Implementation of PLC Based VFD Fed Three Phase ...

I MT I C’ 15 CONFERENCEPROCEEDI NGS 4THI NTERNATI ONALMUL TI  TOPI CCONFERENCE 11 13FEB. ,2015

I SBN:978969880329

FOREWORD This book includes proceedings of 4th International Multi-TopIc Conference (IMTIC’15) comprising a selection of regular, accepted and short papers. IMTIC'15 is an International forum for researchers to exchange the information regarding novel aspects of technology, application and service development within the multidisciplinary framework of Information & Communication Technologies (ICT). It provides a forum for technical exchange amongst researchers from academia and industries in various emerging fields of Computing, Wireless Communication, Signal Processing, Speech and Pattern Recognition, Computer Systems and Networks, Wireless Sensor Networks, Neural Networks and Control Systems, Power Electronics, Energy Management spanning across seven conference tracks. This year, IMTIC'15 received more than 400 submissions, which after author withdrawals and deleting only abstracts led to a total of 294 papers that were reviewed. A small subset of papers was classified as quick rejects based on their quality and likelihood of rejection. All the papers were checked through TURNITIN for plagiarism before the review process. The selected papers were thoroughly reviewed. Each paper has been reviewed by at least one International expert and one National expert. The review process included several stages including TPC member reviews, an on-line discussion among the reviewers, TPC lead summary recommendations, and additional reviews (as needed). Based on the TPC recommendations, we were only able to accept a total of 66 papers for the publication and presentation at IMTIC'15; this represents an acceptance rate of approximately 22.4%. Among 66 the best 25 papers were published as special issue at Mehran University Research Journal of Engineering and Technology. Special thanks to all the authors for promptly revising their papers to meet the requirements of reviewers. Thanks to all the valued reviewers for contributing with their expert comments to improve the papers. Hope you will have the benefit by reading this book.

EDITORS: Dr. Bhawani Shankar Chowdhry, MUET, Jamshoro, Pakistan Dr. Faisal Karim Shaikh, MUET, Jamshoro, Pakistan

TABLE OF CONTENTS Regular Papers 1. 2.

3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Application Of Fractional Order Control Technique To An ElectroHydraulic Position Servomechanism Optical Character Recognition Technique For Recognizing And Splitting Text Of Restaurant Receipts: A MATLAB Based Graphical User Interface Project Performance Evaluation Of Wireless EEG Sensor Network A Context-Aware Switching Mechanism between Heterogeneous Wireless Networks Performance Evaluation Of Secure Force Symmetric Key Algorithm Performance Analysis Of 3-D Orthogonal Frequency Division Multiplexing Transceiver For Satellite Communication Generating Alerts Using Context Aware Security And Brahms Model For Customer Service Improvement Onboard GSM Triangulation Tracking Device Cyber-Secure Energy Measurement System For Smart Grid Quality And Context-Aware Data Collection Architecture From CrowdSourced Data Simulation Of PMSG Based Wind Energy System With Multi Level NPC Converter Extension to Partner Based Grid Workflow Scheduling Heuristic A Fast Encryption/Decryption Scheme for Biometric Images Using Multiple Chaotic Maps

1 11 17 26 33 39 47 55 64 73 84 92 104

Accepted Papers 14. Analyzing The Influence Of Prosodic Features On Emotional States Recognition In Regional Languages Of Pakistan 15. Development Of Gesture Based Wireless Appliances Control System For Hospital Ward 16. Analyses Of Intercultural Challenges In Offshore Software Development Outsourcing Relationships Based On Continent And Company Size 17. SDR Design Of Reconfigurable Transceiver For Rapid Implementation On DSP Platform 18. Implementation Of PLC Based VFD Fed Three Phase Induction Motor Via OPC Server 19. A Survey Based Study On Communication And Coordination Challenges In Offshore Software Development Outsourcing Relationships from Vendors’ Perspective 20. Development Of Horizontal Single Axis Solar Tracking System For Efficient Energy Monitoring And Control With Timing Algorithm 21. Evaluation Of Software Birthmarks Using Fuzzy Analytic Hierarchy Process 22. Slicing PERN With Software Defined Networking: An Approach Towards Network Experimentation 23. Interconnection Of Asynchronous Power Grids Using VSC-HVDC Systems 24. Different Types Of Millimeter Wave Antenna Designs For Imaging Applications 25. Risks In Software Multi-Sourcing Relationship From Vendors’ Perspective: A Systematic Literature Review 26. Knowledge Management In Very Small & Temporary Organizations 27. Area-Performance Analysis Of Hamming Code On 90-Nm CMOS Technology

116 122 129 137 145 153 164 171 176 184 192 200 207 211

28. Optimized Traffic Analysis Attack Settings For OpenSSH Based Large Data Sets 29. Performance Comparison Of Classifiers For Activity Recognition Using big dataset 30. Towards Implementation Of Autonomous Fire Detection Using Salient Image Features 31. Implementation Of Industrial Wireless Closed-Loop Control Network Using Zigbee 32. Analysis Of Mobility Driven Spectrum Sensing Cognitive Radio Under Suburban Environments 33. Development of Secure VOIP Handset Based on Open-Source SIP Client 34. Machine Learning Based Multicast Multipath Routing In Cognitive Radio Networks 35. Efficient Hardware Design & Control Of A Single-Link Inverted Pendulum Using Conventional, Fuzzy and Hybrid Techniques 36. Frequency Dependent Human Head Phantom Design for SAR Analysis 37. Quadcopter UAV Control by Skeletal Tracking - A Humanized Approach 38. A Simple, Fast and Near Optimal Approximation Algorithm for Optimization of Unweighted Minimum Vertex Cover 39. Communication and Coordination Challenges Mitigation Model (CCCMM) for Offshore Software Development Outsourcing Vendors 40. IoT Framework Based Dynamic Traffic Control System 41. Device-To-Device Communication in LTE-Advanced and its Prospects for 5G Cellular Communication

218 225 234 238 248 254 261 270 282 291 296 304 311 316

Short Papers 42. Secure Recognition-Based Graphical Authentication Mechanisms 43. High Qos Adaptive Cross Layer (AC) Protocol For Adhoc Networks 44. I/O Standard Based Low- Energy Pseudo Noise Generator For Optical Communication 45. Energy Management Through Social Networking In Smart Grid - A Review 46. Systematic Literature Review Protocol For The Identification Of Practices In Software Outsourcing Contract 47. Load Forecasting And Dynamic Pricing Based Energy Management In Smart Grid- A Review 48. Pervasive Learning Environment With Emerging Technologies And Learning Transformation 49. Harnessing Of Electric Power Through Multiple Renewable Sources Using Cascaded MLI For Power Optimization 50. Flexible AC Transmission System Controllers: A Review 51. Use Of Information Technology In Hospitals At Pakistan: Case Study Of Shaukat Khanum Memorial Cancer Hospital And Research Center (SKMCH&RC), Lahore 52. Performance Analysis Of Cross Layer Protocols For Flying Ad-Hoc Networks 53. A Simulink Based Adaptive UFLS Scheme 54. Scrum Adaptation To Improve Software Quality In Software Engineering Semester Projects 55. Software integration practices for GSD vendors - A systematic literature review protocol

326 333 343 351 359 368 376 384 393 403 408 416 423 429

I MT I C’ 15

REGULARPAPERS

Regular Papers-Proceedings of IMTIC’15

Application of Fractional Order Control Technique to an Electro-Hydraulic Position Servomechanism Salman Ijaz1, Mohammad Ahmad Choudhry1, Ahsan Ali1, Umair Javaid2 1 Department of Electrical Engineering, University of Engineering &Technology Taxila, Pakistan [email protected], {dr.ahmad, ahsan.ali}@uettaxila.edu.pk 2 Department of Electrical Engineering, University of Engineering &Technology Lahore, Pakistan [email protected]

Abstract The paper presents a method to design optimally tuned fractional order controller for position tracking control of electro-hydraulic servo system (EHSS). A proposed controller has simple structured and is based upon fractional order model of system. To optimally tuned controller parameters particle swarm optimization technique is applied. Simulation result on linear and non linear system show significant improvement in performance of fractional order PID controller compared to classical PID. Keywords: Electro-hydraulic servo system, particle swarm optimization, fractional order PID, fractional order controller

Introduction EHSS is being used worldwide in several fields of engineering due to the fact that it has small volume, swift response, high arduousness and huge output power [1]. Due to nonlinear dynamics of system, there are extremely unpredictable behaviors in this system like the non linear flow-pressure relationship, dead band caused by internal outflow, hysteresis, stiffness of fluid, friction force [2]. Hence, for EHSS, a precise mathematical model is very tough to develop hypothetically that include all such nonlinearities [3-4]. Position tracking control of EHSS has wide range of application in industries and studied extensively in the past few decades. The key element is to achieve precise positioning, widen bandwidth and robust stability. Several techniques have been applied for position tracking control of EHSS. A variable structure control is designed for EHSS in [5, 6] showed a worthwhile results. But the critical issue with this class of controller is the excitation of un-modeled. Adaptive techniques are deployed to exploit the rapid variation of system parameter [7]. A non linear adaptive control law is formulated to compensate the uncertain parameters. Adaptive fuzzy control law is also presented for position tracking control of EHSS in [8]. Fuzzy self tuning mechanism is introduced for the adaption of sliding mode control parameters. This configuration has effectively reduced the chattering phenomena. Classical PID controllers are still active in many industries due to ease in their design and implementation. No doubt, such controllers have limited performance and bandwidth. But the proper selection of controller parameters can enhance performance. In this regards, [9] proposed an optimal PI controller for EHSS. In order to achieve robustness and precise performance, the parameters of controller are tuned using particle swarm optimization (PSO). The designed controller is simple structured and provide robustness towards stability. In [10] PID controller is designed to control the angular position of rotator hydraulic actuator and its parameters are tuned using genetic algorithm (GA). Simulation as well as experimental results shows the effectiveness of proposed controller. The aforementioned approach found its application in control of movable surface of space shuttles. A similar approach is followed by [11] to control angular position of EHSS, showing satisfactory tracking. Another parameter optimization algorithm is presented in [12] to control the position of EHA. The positioning mechanism is installed in hydraulic stand in plate rolling. To control the roll-gap, a time delay controller is designed and parameters are optimized using evolutionary strategy (ES). Robust 𝐻∞ controller is demanding since the last two decades and has plenty of applications in industry. [13] proposed a robust 𝐻∞ controller to control the force exerted by EHA.. Simulation and experimental results determine the stability in term of performance trade off. In the recent year, a substantial amount of studies and applications are focused on fractional order system (FOS) especially on fractional order PID (FOPID) controller. A simple structured controller found very effective as compared to other classical controllers. FOPID provides robustness on external disturbance and external load variation [14]. FOPID controller is applied to torsional system's backlash vibration suppression control in [15] and found robustness against backlash non linearity. Performance of FOPID controller is evaluated through bioreactor control system [16]. Results show flexibility of FOPID controller in term of improving gain and phase margin compared to classical PID. A fractional order PID controller is also applied to multi input multi output (MIMO) twin rotor aero dynamical system (TRAS) for pitch and yaw angle control. Two separate FOPID

1

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 controllers are designed for pitch and yaw. Simulation results verified the performance of FOPID controller as compared to classical PID controller [17]. In this paper FOPID controller is designed for position tracking control of EHSS due to its superior performance. A fractional order model of system is developed first and FOPID controller is designed. The performance index is minimized using PSO technique. To check the robustness of FOPID controller, a disturbance signal function of state of system is applied. Simulation results are verified on non linear model of EHSS. The rest of paper is organized as follows: a brief description of experimental setup and its mathematical model is presented in Section 2. The linear time invariant (LTI) and fractional order (FO) model of system is developed in section 3. Section 4 enlightens the design and analysis of PID and FOPID controller. PSO is developed in section 5 to tune controller parameters. Simulation results at different input are analyzed in section 6. Finally comparison and discussion in section 7 shows the effectiveness of proposed scheme. EHSS Description A laboratory setup of EHSS is shown in fig 1. It consist of • Hydraulic pump with synthetic fluid. It can built pressure of 850 PSI. • 4WRAEB6E12 proportional valve configuration. • Double acting rod of length 10 inches. • Position Transducer serves as sensor. • Dc power supply of 12 Volts for biasing of electronic circuit in proportional valve.

Fig.1. A laboratory Based setup of EHSS Problem Formulation and Mathematical Model of EHSS A 4- way proportional valve configuration is shown in fig 2. Hydraulic actuator is a double acting piston and is connected to proportional valve thorough cylinder port. The flow rate 𝑄1 and 𝑄2 in and out of cylinder controls the piston movement. Based upon that configuration, a mathematical model of EHSS is developed [18].

2

ISBN: 978-969-8680-32-9


Fig .2. A 4 Way Proportional Valve Configuration A mathematical equation representing fluid flow through orifice can be described as 𝑄 = 𝐶𝑑 𝐴0 �2�𝜌 (𝑃1 − 𝑃2 ) Where 𝐶𝑐 𝐶𝑟 𝐶𝑑 = = Discharge Coefficient 2

(1)

�1−𝑐𝑟2 �𝐴0� � 𝐴1

𝑄 = Flow through orifice 𝑃1 = Pressure at orifice 1 𝑃2 = Pressure at orifice 2

The actuator is a double-acting hydraulic cylinder and is connected to the pressure supply through servo valve. Flow of fluid to the cylinder depends upon the spool displacement which is proportional to input current. The pressure 𝑃𝐿 delivered to the load is described as 2𝛽 𝑃̇𝐿 = 𝑒 (𝑄𝐿 − 𝐴 𝑥̇𝑣 + 𝐶𝑡𝑡 𝑃𝐿 )

(2)

𝑉𝑡

Where 𝐴 = Spool Area 𝐶𝑡𝑡 =Total leakage coefficient 𝛽𝑒 = Effective bulk modulus 𝑥𝑣 = Spool displacement 𝑉𝑡 = Total actuator volume

Due to load pressure 𝑃𝐿 , the piston displacement (𝑥𝑝 ) in cylinder can be modeled as

𝑃𝐿 𝐴 = 𝑚𝑥𝑝̈ + 𝑏𝑥𝑝̇ + 𝐾𝑣 𝑥𝑝

(3)

The relationship between spool displacement 𝑥𝑣 and input current 𝑖 is determined by first order approximation as 𝑥𝑣̇ = 𝜏𝑣 𝑥𝑣 + 𝐾𝑣 𝑖

(4)

From (1-4) the system can be written in state space form 𝑥̇ 1 = 𝑥2 1 𝑥̇ 2 = (−𝑘𝑥1 − 𝑏𝑥2 + 𝐴𝑥3 ) 𝑚 𝑥3̇ = 𝛼𝑥2 − 𝛽𝑥3 + 𝛾�𝑃𝑠 − 𝑠𝑠𝑠 (𝑥4 )𝑥3 𝑥4 (5) 1 𝐾 𝑥4̇ = − 𝑥4 + 𝑢 𝜏 𝜏 Where States and inputs 𝑥4 = Valve position 𝑥1 =Actuator Piston position 𝑥2 =Actuator Piston Velocity 𝑢 = Input Current to servo

valve

3

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 𝑥3 =Load Pressure

System Parameters

Flow Rate Constant 4𝛽𝑒 𝐴 𝛼= = 3.6 × 10−11 𝑉𝑡 4 𝐶𝑡𝑡 𝛽𝑒 𝛽= = 7.6 × 108 𝑉𝑡 4𝐶𝑑 𝛽𝑒 𝑤 𝛾= = 2.917 × 109 𝑉𝑡 √𝜌

𝑚 = Mass of actuator 𝑘 =Spring Constant

𝑏 =Damping Constant

The state space model in (5) is the complete representation of system dynamics. The model is tested at various inputs and validated through experimental setup. The experimental validation results at sinusoidal input is shown in fig 3 10

Reference Input Simulation Data Experimental Output

8 6 4

Position

2 0 -2 -4 -6 -8 -10

0

500

1000

1500 Samples (N)

2000

2500

3000

Fig .3. Model Validation Curve

Model Identification In this section, EHSS is represented in term of LTI model and FO model of system. To excite the internal dynamics of system, a multi sine wave input signal is applied. Input-output data samples are used to identify the system model. LTI Model of EHSS The dynamics of EHSS is highly nonlinear so linearization of system at some equilibrium point is necessary before controller design. In this subsection, LTI model is identified by fitting 2nd order polynomial to system input-output data. A continuous 2nd order transfer function representation of EHSS is 𝐺(𝑠) =

0.0119

𝑠 2 +563.33𝑠+1.809

(7)

Fractional Order Model of EHSS A FOS can be defined by a LTI fractional-order differential equation of the form [19]. 𝛽 𝛽 𝛽 𝑎𝑛 𝐷𝑡 𝑛 𝑦(𝑡) + 𝑎𝑛−1 𝐷𝑡 𝑛−1 𝑦(𝑡) + ⋯ … + 𝑎0 𝐷𝑡 0 = 𝛼𝑚 𝛼𝑚−1 𝑑𝑚 𝐷𝑡 𝑢(𝑡) + 𝑑𝑚−1 𝐷𝑡 𝑢(𝑡) + ⋯ + 𝛼 𝑑0 𝐷𝑡 0 𝑢(𝑡) (8) Or by a continuous transfer function of the form 𝐺(𝑠) =

𝑑𝑚 𝑠 𝛼𝑚 +𝑑𝑚−1 𝑠𝛼𝑚−1 +⋯..+𝑑0 𝑠𝛼0

(9)

𝑎𝑛 𝑠 𝛽𝑛 +𝑎𝑛−1 𝑠 𝛽𝑛−1 +⋯..+𝑎0 𝑠 𝛽0

Where 𝛼𝑘 , 𝛽𝑘 , ( 𝐾 = 0,1,2,3 … …) are real numbers, 𝛽𝑘 > ⋯ . . > 𝛽1 > 𝛽0 , 𝛼𝑘 > ⋯ … > 𝛼1 > 𝛼0 and 𝑎𝑘 , 𝑑𝑘 are random constants. The transfer function representation of FOS can be obtained by taking well known Laplace transform of fractional order differential equation [20]. Fractional order transfer function (FOTF) model leads to the concept

4

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 of fractional poles and zeros in the complex s-plane. For the commensurate FOS (9) with the characteristic equation in terms of the complex variable 𝛼 is said to be bounded-input bounded-output (BIBO) stable, if the following condition is satisfied 𝜋 |arg(𝜆𝑖 )| > 𝛼 2 Where as 𝜆𝑖 are the roots of characteristic polynomial in terms of 𝜆. The fractional order elements can be rationalized by several iterative techniques. However in our case Oustaloup method of approximation is used [21]. It gives a very good fitting to fractional order element (𝑠 𝜆 ) within chosen frequency band FO model of EHSS is obtained using FOMCON toolbox in MATLAB [22]. Oustaloup method of approximation is chosen. The model validation curve shows the BFT factor of 92.49 %. A FOTF of EHSS is obtained as 1 𝐺(𝑠) = 1.5 (10) 𝑠 +𝑠+𝑠 0.5 +1 The model in (10) is utilized as benchmark in design of fractional order controller (FOC).

Control Strategy Position tracking control of EHSS is thought to be a challenging engineering problem due to its nonlinear dynamics. Kept in view the above said nonlinearity and to improve the performance of PID controller, a selection of FOPID controller is promising. The proposed controller is expected to improve stability margin, operating frequency range, wide bandwidth and maintain robust stability. To evaluate the controller performance, several performance criteria are available that includes integrated absolute error (IAE), mean square error (MSE), integrated square error (ISE), and integrated time weight square error (ITSE). The transient analysis of controller is based upon specifications such as (i) settling time (𝑇𝑠 ), (ii) rise time (𝑇𝑟 ), (iii) overshoot (𝜎𝑒 ) and (iv) steady state error (𝐸𝑆𝑆𝑆 ). The tools used for frequency domain analysis are gain margin (𝐺𝑀 ), phase margin (𝑃𝑀 ) and operating bandwidth. The performance criteria 𝐾(𝑠) is defined by assigning weight to each factor as 𝐾(𝑠) = 𝑤1 𝜎𝑒 + 𝑤2 𝐸𝑠𝑠𝑠 + 𝑤3 𝑇𝑠 + 𝑤4 𝑇𝑟 + 𝑤5 ∫|𝑒| 𝑑𝑡 +

𝑤6

(11)

𝐺𝑀

A block diagram representation of overall control strategy is shown in fig 5. A disturbance signal at the input of system is applied to check the robustness of proposed controller. A brief description of FOPID controller is discussed in coming subsection 1

Source data Identified model

0.8

0.6

Position (m)

0.4

0.2

0

-0.2

-0.4

-0.6

-0.8

0

10

20

30

40

50 Time (s)

60

70

80

90

100

Fig .4. Fractional Order Model Identification Curve

5

ISBN: 978-969-8680-32-9


Fig.5. A Controller Design Strategy Fractional Order PID Controller The FOC were introduced by Oustaloup (1991) and developed CRONE controller [16]. The transfer function representation of FOPID is: 𝐺𝑐 (𝑠) = 𝐾𝑃 + 𝐾𝐼 𝑠 −λ + 𝐾𝐷 𝑠 µ , 𝜇, λ > 0

(13)

Since from (13), FOPID controller has five tuning parameters i.e. proportional gain (𝐾𝑝 ), integral gain (𝐾𝑖 ), derivative gain (𝐾𝑑 ), integral fractional power (λ) and derivative fractional power (μ). FOPID controller offers many benefits such as elimination of SSE, robustness and provision of phase margin and gain cross over frequency. A block diagram representation of FOC is shown in fig. 6

Fig .6. Fractional Order Controller In the application of FOPID controller, mainly two type of problem arises. The first one is the model realization or approximation and the other is tuning of controller parameters. For model approximation Oustaloup filter technique is used. Since the problem of optimal tuning of controller parameter is more critical then the first one. For this purpose, PSO technique is employed and is discussed in detail in coming section.

Particle Swarm Optimization Technique One of the major issues discussed in this paper is to develop a simple design strategy for tuning controller parameters that guarantee’s (i) the closed loop system is internal stable (ii) the robust performance criteria in (11) is satisfied. This section contains online algorithm to tune the parameters of PID and FOPID controller using PSO. PSO is population based stochastic optimization technique inspired by social behavior of bird flocking. It is the one of most powerful tool to solve optimization problems. It has gained a lot of attraction in recent years as it requires less computer code for its realization. It does not require strong assumption in order to solve the problem efficiently. It is computationally inexpensive, require less memory and speed for its working. Finally, the solution is less dependent on initial state of particle, which could be a great advantage in optimization field

6

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 [23]. An augmented lagrangian PSO algorithm is developed [24] that have ability to handle optimization problem subject to equality and non-equality constraints. PSO is adopted to tune the parameters of classical PID and FOPID controller. The algorithm is initialized by system parameters, particle position and velocity. The size of swarm and maximum bird step is assumed to be 50. Local optimum and global optimal values are calculated at each particle’s step. The algorithm provides online tuning of controller parameter using system data. The position and velocity of particles are updated at each step until fitness function in (11) satisfied [25]. A flow chart representation of PSO algorithm is shown in figure 7. The optimal value of FOPID parameters is shown in Table 1.

Fig .7. Flow Chart Representation of PSO

Simulation Results The section contains simulation result of FOPID controller applied to EHSS. The initial state of actuator is set to be zero. The controller is tested in closed loop configuration at various tracking signals. Simulation results at step and sinusoidal inputs applied to linear and nonlinear system are included in this section. The results are compared with classical PID controller. To check the robustness of FOPID controller, a disturbance is applied at the control input channel. In fig 7, a step and sinusoidal reference signals are applied to LTI and FO model of EHSS. It is clear from the simulation results that FOPID controller is capable to overcome transient condition and maintain steady state earlier as compared to PID controller. Fig. 8 shows tracking performance of controller applied to nonlinear model of EHSS. A step disturbance signal is applied at the input channel of EHSS. It is clear from the results that FOPID controller has good disturbance rejection capability as compared to PID. Table.1. FOPID Controller parameters using PSO Parameters 𝐾𝐾 𝐾𝐾 𝐾𝐾 𝜆

Values 4.0795 99.954 44.178 0.5309

𝜇

0.5740

7

ISBN: 978-969-8680-32-9


1.4

1.5

PID REF FOPID

1.2

FOPID REF PID

1

1 0.5

Position (m)

0.8

0.6

0

0.4 -0.5

0.2

0

-1

0

1

2

3

4

5

6

7

8

1.4

9

10

1

2

3

4

5 Time (s)

6

7

8

1.5

FOPID Reference PID

1.2

0

9

10

PID FOPID Reference

1

1

Position

Position

0.5 0.8

0.6

0

-0.5 0.4

-1

0.2

0

0

1

2

3

4

5

6

7

8

9

10

-1.5

0

1

Time (s)

2

3

4

5 Time (s)

6

7

8

9

10

Fig.9. Tracking Performance of Nonlinear System in the Presence Of Disturbance

Comparison and Discussion Due to non linear dynamics of EHSS, it undergoes the variation of parameters. The open loop response of EHSS has large settling time and undergoes large steady state error. Such type of system response is not appreciable as industrial application requires precise tracking in minimum time. Fractional order PID controller has capability to account parameter variation. Furthermore PSO technique provides optimal value of controller parameters. The performance of both controllers is compared according to performance index. From the comparison table 2, it is concluded that performance of FOPID controller dominates over classical PID in term of rise time and settling time. Moreover, FOPID controller is able to maintain the frequency domain characteristic (i.e. phase and gain margin) to optimal value. FOPID controller provides robust performance in the presence of disturbance.

Conclusion Application of FOPID controller in position tracking control of EHSS is presented in this paper. Furthermore the PID controller is applied on the same system for comparison. In this paper, EHSS is represented in term of FOS. The optimal value controller parameters are obtained using particle PSO. Simulation results show the performance of FOPID controller dominates over the conventional PID controller. FOPID controller has an ability to maintain transient and frequency domain characteristic of system under various operating conditions.

8

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Table.2. Comparisons of PID and FOPID

Performance Parameters

Settling time (sec) Rise Time (sec) Over Shoot (%) Steady State Error (%) Gain Margin Phase Margin

PID using PSO

FOPID

Linear Model

Nonlinear model with disturbance

Linear Model

1210

0.9

3.7

0.2

0.5

680

0.45

1.6

0.09

0.21

0

0

20

0

5

25

0.1

0

0

0

∞

40.2

33.1

∞

∞

EHSM

105

57.21

66.1

60

Nonlinear model with disturbance

53

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

Alleyne, A., Liu, R.: Systematic Control of a Class of Nonlinear Systems with Application to Electro-hydraulic Cylinder Pressure Control. IEEE. Trans. Cont. sys. Tech. 8(4), 623–634 (2000) Zhao, P., Wang, S., Li, X., Zhang, B .: A Novel Method of Parameter Identification of Nonlinear Electrohydraulic Servo Systems. In: International Conference on Fluid Power and Mechatronics, FPM, pp.547-551 (2011) Garagic, D., Srinivasan, K.: Application of Nonlinear Adaptive Control Techniques to Electrohydraulic Velocity Servomechanism. Cont. Sys Tech. 12(2), 303- 314(2004) Ferreira, J., Sun, P., Gracio, J.: Close Loop Control of a Hydraulic Press for Spring back Analysis. Jour. Mater. Pro. Tech, 377–381(2006) Chern, T.L., Wu, Y.C.: An Optimal Variable Structure Control with Integral Compensation for Electrohydraulic Position Servo Control Systems. IEEE. Trans. Indus. Elect. 39, 460-463(1992) Fung, R.F., Wang, Y.C., Yang, R.T., Huang, H.H.: A Variable Structure Control With Proportional and Integral Compensation for Electrohydraulic Position Servo Control System. Mechatronics. 7(1), 67-81(1997) Guan, C., Uangxia, P.: Nonlinear Adaptive Robust Control of Single-Rod Electro-Hydraulic Actuator with Unknown Nonlinear Parameters. IEEE. Trans. Cont. Sys. Tech. 16(3), 434-444 (2008). Cerman, O., Husek, P.: Adaptive Fuzzy Sliding Mode Control for Electrohydraulic Servo Mechanism. Exp. Sys. App. 39, 10269-10277 (2012) Olranthichachat, P., Anidvilai, K.: Design of Optimal Robust PI Controller for Electro-Hydraulic Servo System. Engg. Lett. (2007) Elbayomy, K. M., Zongxia, J., Huaqing, Z.: PID Controller Optimization by GA and its Performances on the Electrohydraulic Servo Control System. Chinese. Jour. Aeronaut. 21, 378-384(2008) Ayman, A.A.: PID Parameter optimization Using Genetic Algorithm Technique for Electrohydaulic servo Control System. Int. Cont. Auto. 2, 69-76(2011) Kim, M.Y., Lee, C.O.: An Experimental Study on the Optimization of Controller Gains for an Electro-hydraulic Servo System using Evolution Strategies. Cont. Engg. Prac. 14, 137-147 (2006) Laval, L., MSirdi, N. K., Cadiou, J. C.: 𝐻∞ Force Control of a Hydraulic Servo-Actuator with Environmental Uncertainties. In: IEEE International Conference on Robotics and Automation, Proceedings, pp. 1566-1571 (1996) Jiangbo, Z., Junzheng, W., Shoukun, W.: Fractional Order Control to the Electro-hydraulic System in Insulator Fatigue Test Device. Mechatronic. (2013) Chenbin, M., Yoichi, H.: The Application of Fractional Order Control to Backlash Vibration Suppression. In: Proceedings of conference on American control, pp. 2901–2906 (2004) Karad, S., Chatterji, S., Suryawanshi, P.: Performance Analysis of Fractional Order PID controller with the Conventional PID Controller for Bioreactor Control. Int. Jour. Sci. Eng. Research. 3 (2012) Sunil Kumar Mishra and Shubhi Purwar.: To Design Optimally Tuned FOPID Controller for Twin Rotor MIMO System. In IEEE conference (2014) Merritt, H.: Hydraulic Control System. John and Waley (1967) Wang, C., Jin, Y., Chen, Y.: Auto-Tuning of FOPI and FOPI Controllers with ISO-damping Property. In: 48th Decision and Control IEEE, pp 7309-7314, Shanghai, China(2009) Monje, C. A., Chen Y. Q., Vinagre, M., Dingyü, X, Vicente, F.: Fractional-order Systems and Controls - Fundamentals and Applications. Springer-Verlag London Ltd (2010)

9

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 21. Oustaloup, A., Levron, F., Mathieu, B., Nanot, F.M.: Frequency-band Complex Noninteger Differentiator: Characterization and Synthesis. IEEE. Trans. Circuits. Syst. I. Fundam. Th. App. 47(1), 25–39 (2000)

22. Tepljakov, A.: Petlenkov, E.: Belikov, J.: FOMCON: a MATLAB Toolbox for Fractional-order System Identification and Control. Int. Jour. Microelect. Computer. Sci. 2(2) (2011)

23. Vinagre, B.M., Monje. A.: Fractional Order Control-A Critical Review. In: 7th Control Portuguese Conference on Automatic Control, pp 11-13, Portugal (2006).

24. Eberhart, R.C., Tsakalis, K.: A New Optimizer using Particle Swarm Theory. In: 6th Proceeding of International Symposium on Micromachine and Human Science, pp .39-43 (2000)

25. Sedlaczek, K., Eberhard, P.: Using Augmented Lagrangian Particle Swarm Optimization for Constraint Problem in Engineering. Struct. Multi-Disp. Opt. 32(4), 277-286 (2006)

26. Kennedy J., Eberhart R.: Particle swarm optimization. In Proceeding IEEE International Conference on Neural Network, pp.1942 – 1948 (1995)

10

ISBN: 978-969-8680-32-9


Optical Character Recognition Technique for Recognizing and Splitting Text of Restaurant Receipts Tayaba Abbasi1, Isra Abbasi2a and Aamir Saeed Malik2b 1

Quaid-e-Awam University College of Engineering, Science & Technology Larkana, Pakistan [email protected] 2 University Technology PETRONAS, Bandar Seri Iskandar, Tronoh 32610, Perak Malaysia a [email protected], [email protected]

Abstract Dinning out with group of friends brings its own challenges. Selecting a place that suits everyone’s choice and making early enough reservation to book a table are just few of them. However, the most daunting task comes at the end when it’s time to pay. The bill contains the list of all the items ordered and respective prices. In this kind of scenario, one among the group has to do the Math and divide the bill to ensure that everyone pays what they are required to pay. However, almost every time, someone either ends up underpaying or overpaying due to miscalculation. This work is aimed to resolve this issue by developing a user friendly interface that can make splitting the receipt an easy task. The initial interface is developed on MATLAB and can be extended to a Mobile App or Excel Interface depending on the restaurant’s suitability. The scanned image of receipt is loaded in the system and the user can calculate the due amount to pay with four easy steps. A simplified Optical Character Recognition (OCR) engine is proposed to recognize the text in the receipt with accuracy of 94 %. Keywords: Optical Character Recognition; MATLAB; Graphical User Interface; Restaurant Receipt

Introduction Planning dine out with group of friends comes with its own challenges. Selecting a place that fits everyone’s choice and booking it early enough to reserve a large table are just few of them. However, the most difficult tasks among all comes at the end, the time when the money has to be divided and paid. The bill contains all the items listed, with its total price. At times like this, usually one among the group has to pull out calculator and start doing the Maths to make sure everyone pays what they are supposed to pay. But mostly all the time, it happens that everyone’s contribution doesn’t add up exactly to the total amount. That means, either someone is underpaying or overpaying. The goal of this work is to resolve this issue and make splitting the receipt an easy task. This work is motivated from a project titled “L’Addition: Splitting the Check, Made Easy” from Digital Image Processing Class at Stanford that developed an Iphone application to snap a receipt and extract the text from the receipt using tesseract Optical Character Recognition (OCR) engine and then perform text processing to define individual items in receipts [1]. A snap shot of the designed application is given in Figure 1. However, the proposed work in this paper is aimed to develop a MATLAB based graphical user interface (GUI) integrated with a simplified OCR engine to extract text from receipt and then perform text processing to define individual items in receipt.

Fig. 1. Related Work: L’Addition: Split the Check

11

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Proposed Algorithm Pipeline Proposed algorithm is based on four major components as shown in Figure 2. 1. User Input (Scanned Copy of Receipt) 2. Optical Character Recognition (OCR) Engine on Receipt Items 3. Text-Processing to parse Receipt items (Quantity, Item Name and Price) 4. MATLAB based Graphical User Interface (GUI)

3

1 2

4

Fig. 2. Proposed Algorithm Pipeline User Input The user input stage is quite straightforward; a user scans the copy of receipt and load the scanned image to system for the next stage. In this work, a mobile scanner is used to scan the receipt, hence the resolution of the scanned image is not really good. Hence some preprocessing steps are required to enhance the image and prepare it for the next stage. Optical Character Recognition Optical Character Recognition is the mechanical or electronic conversion of scanned or photographed images of typewritten into computer readable text [2]. With intensive amount of research done in the field, there are several algorithms designed to recognize characters as accurate as possible [3-4]. OCR is the key in the development of whole system. The algorithm used to extract and recognize the text affects the performance of the whole system. If OCR doesn’t perform accurately, the following steps in pipeline will take this inaccurate data as input, resulting the end output to be inaccurate as well. The OCR algorithm used in proposed system consists of three phases as given in Figure 2. Pre-processing Scanned Input Image. Pre-processing is the preliminary step which transforms the data into a format that will be easily and effectively processed. Thus preprocessing is an essential stage prior to feature extraction since it controls the suitability of the results for the successive stages [5]. Pre-processing steps are based heavily upon the type of input image. If the input image has some kind of known noise, so a specific noise filter is applied to denoise the image [6]. The pre-processing steps applied in this algorithm are relatively simple and listed below. • Convert to Gray Scale Image using rgb2gray • Convert to Black & White Image with selective threshold value using im2bw • Apply Morphological Operation (Dilate) using imdilate Feature Extraction. In order to extract the characters, connected component analysis is performed on the whole image in order to find the connectivity of each character and isolate each character from the rest. The command used to find connectivity is bwareaopen and bwlabel. Once the connected components (or, individual characters) are isolated then the image of each character is cropped by finding respective rows and columns of each

12

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 connected component. The Psuedo code for cropping each word is given as following. Examples of cropped image is shown in Figure 3. The cropped image of letter is then passed to next stage for classification. [R C] = bwlabel(imgn); for n=1:C [r,c] = find(R==n); % Extract letter n1=imgn(min(r):max(r),min(c):max(c)); % Resize letter (same size of template) img_r=imresize(n1,[42 24]); end

Fig. 3. Example of Cropped Image Classification of Characters. This is the last stage in the OCR engine, where the extracted image of character is to be rightly classified as the original character in image. Among several techniques of classification, a very simple yet an efficient method is correlation classification method that is used in the proposed OCR engine. In order to implement correlation method, a template is created that contains images of all alphabets (A-Z), numbers (0-9) and a decimal point (.) in ARIAL font style. Each image in template is assigned its respective character, for instance an image of alphabet A is assigned character A. The template contains total of 37 images. Classification of characters is performed by computing correlation coefficient between extracted image and all the images stored in the template. Hence, for each extracted image, there are 37 values of correlation coefficient computed. Among these 37 values, the image in the template that gives the highest correlation coefficient is selected and the respective character associated with that image is chosen as a recognized character. The algorithm used to compute correlation co-efficient is given in Equation 1, where A is the extracted/cropped image (matrix) from the receipt and B is the image (matrix) from template. 𝑟=

∑𝑚 ∑𝑛(𝐴𝑚𝑚 −𝐴̅)(𝐵𝑚𝑛 −𝐵�)

�1�

�(∑𝑚 ∑𝑛(𝐴𝑚𝑚 −𝐴̅)2 )(∑𝑚 ∑𝑛(𝐵𝑚𝑚 −𝐵�)2 )

At the end, the outcome is string of words found in the image. The result found after classification for image is shown in Figure 4. The bold and underlined letters, highlight the text that is misclassified. MERRYBROWN AEONSTATVON18 PERAKMALAYSAA 3FRAEDCHACKENPLATERM3145 2DANNERPLATERM16 2FRESHSALADRM10 5BOSLEDVEGETABLERM51 2ORANGEJUACERM10 2MANGOSHAKERM10 TOTALRM12845

(a)

(b)

Fig. 4. (a) Scanned Image of Receipt (b) Recognized Text Text Processing Given the string of words, the system now has to parse and make meaning of the text on the receipt. The receipt format used in this work has found to be uniform for many restaurants. The format is as shown below. [Quantity] INT.CONST

[Item Name] INT.STRING

[Price] INT_CONST.INT_CONST

The algorithm designed for text processing detects if the first character of each line is a number or string of words. If it is a number, then it is taken as quantity of items ordered. The algorithm reads the number (i.e the

13

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 quantity) to determine number of items ordered, and divide the total price that comes at the end of line with quantity to get the individual price. The string of words between quantity and price is taken as it is and printed as item name. In the other case, if the first character in a line is a string of words then it is taken as either the header or footer of receipt and printed as it is. The output text after text processing is shown in Figure 5. Quantity

Item Name MERRYBROWN PERAKMALAYSAA 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 TOTALRM12845

Price

FRAEDCHACKENPLATERM FRAEDCHACKENPLATERM FRAEDCHACKENPLATERM DANNERPLATERM DANNERPLATERM FRESHSALADRM FRESHSALADRM BOSLEDVEGETABLERM BOSLEDVEGETABLERM BOSLEDVEGETABLERM BOSLEDVEGETABLERM BOSLEDVEGETABLERM ORANGEJUACERM ORANGEJUACERM MANGOSHAKERM MANGOSHAKERM

10.300000 10.300000 10.300000 8.000000 8.000000 5.000000 5.000000 10.200000 10.200000 10.200000 10.200000 10.200000 5.000000 5.000000 5.000000 5.000000

Fig. 5. Parsed Output Text after Processing However, it must be noted that there may be other restaurants that follow different format. Hence algorithm for processing text depends heavily on the receipt format. If the receipt format is not as assumed in this work, then a customized algorithm for that format has to be written to parse the text. User Interface Once the internal representation of receipt items has been constructed, all that is left is to provide the user with an interface to split the items amongst different customers in the group. This phase, has no algorithm sophistication; though it has some coding sophistication as the interface is designed on Graphical User Interface in MATLAB. The interface let the user select items from a list, computes the subtotal and stores an object containing the user’s name and the total price that they are required to pay. Each user repeats the process for themselves and in the end, everyone has the exact amount they are required to pay without having to perform any manual calculation. The interface is shown in Figure 6 and 7.

Fig. 6. MATLAB GUI of developed System

Usage The application flow of designed system is as follows: a scanned image of receipt (from a mobile scanner) is loaded in the system, upon clicking “SPLIT”, within seconds, a list of all items on receipt are presented under section “Select Items”. A user can then select one or more items on the list and proceed to check out by entering

14

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 their name in the space given and clicking “PAY”, which will then show the items selected with the total amount due in the section “List of Paid Items”. The name and corresponding items are recorded to keep track of the totals that everyone must pay. Once a user is done, all of the previously selected items are removed from the “Select Items” section. New user can continue the same process by clicking “NEW USER”. The section titled “List of Paid Items” will give list of all people who have already calculated their required total contributions. This process is repeated for every customer in the group until everyone has their total amount.

Fig. 7. Application Flow of developed System

Results & Limitations As inferred from interface, the system works quite successfully. The success of the system is validated from several sample receipts as given in Appendix. The accuracy on character level is calculated to be 94 % by using Equation 2. The accuracy could have increased, if the receipt is scanned by a good resolution scanner as the receipt used in this system is scanned from a mobile device. The proposed algorithm is in initial stage and can be extended to a Mobile App or Excel Interface depending on the restaurant’s suitability.

Accuracy =

𝑁𝑁𝑁𝑁𝑁𝑁 𝑜𝑜 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑊𝑊𝑊𝑊𝑊𝑊𝑊 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑇𝑇𝑇𝑇𝑇 𝑁𝑁𝑁𝑁𝑁𝑁 𝑜𝑜 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

�

�2�

In comparison with Iphone app developed at Stanford that uses built in open source tesseract engine, the proposed algorithm develops its own OCR engine by using image processing algorithm sophistication that exposes applicability of image processing toolbox in such an application. However, the drawback in comparison is its inaccuracy in terms of detecting decimal point “.” and alphabet “I” in the receipt. As can be inferred from Figure 4, the decimal point “.” is always recognized as “4” and alphabet “I” is recognized wrongly as letter “A” or “V” or “S”. Considering the application of proposed system, the wrong recognition of letter “I” is not as critical as of decimal point “.”. Because, numbers are one of the main items of the receipt that we want to extract accurately to do further division. In this work, to undo the misclassification in decimal point, post-processing steps are implemented that analyzes the numbers recognized from receipt for misclassification and places a decimal point instead of misclassified digit ‘4’. The text processing stage to parse the recognized text as shown in Figure 5 is performed only after the post-processing stage. Hence the final division obtained after text processing stage as given in Figure 5 is accurate and doesn’t use misclassified number. Further, misclassification of decimal point can also be resolved by improving pre-processing phase that will deliver better image of decimal point to be recognized correctly. Hence an amelioration in algorithm is required by improving pre-processing phase that can classify the decimal point correctly.

15

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Future Work The main part in the project that requires further attention is pre-processing phase in OCR part. As shown in previous section, the accuracy is good enough but not so high as some of characters are misclassified. The higher the accuracy of OCR gets, the better would be the performance of the whole system. Hence it is recommended to improve pre-processing steps to filter the image in the best way. It is also recommended to try different classification methods e.g neural net to increase accuracy.

References 1. 2. 3. 4. 5. 6.

A. Heng, “L’Addition: Splitting the check, Made Easy”, Project Report, Digital Image Processing, Spring 2012-2013, Stanford. https://stacks.stanford.edu/file/druid:yt916dh6570/Heng_LAddition_Restaurant_Check_Splitting.pdf U. Patel, “An Introduction to the Process of Optical Character Recognition,” International Journal of Science and Research (IJSR), Vol.2. (5), May 2013. S. Mori, C. Y. Suen, and K. Yanamoto, “Historical review of OCR research and development,” Proceedings of the IEEE, Vol.80, pp.1029-1058, July 1992. R. Plamondon and S. N. Srihari, “On-line and off-line handwriting recognition: comprehensive survey,” IEEE Transactions on Pattern Analysis and Machine Vision, Vol.22, No.1, January 2002. Y. Alginahi, “Preprocessing Techniques in Character Recognition,” Character Recognition, Minori Mori (Ed.) Q. Chen, “Evaluation of OCR Algorithm for Images with different Spatial Resolutions and Noise” M. Sc Thesis, Elect. Eng., University of Ottawa, Canada, 2003.

16

ISBN: 978-969-8680-32-9


Performance Evaluation Of Wireless EEG Sensor Network Muhammad Khizar Abbas, Bilal Muhammad Khan Dept. of Electronics & Power Engg. National University of Sciences and Technology (NUST) Islamabad, Pakistan [email protected], [email protected]

Abstract In the field of medical Electroencephalography (EEG) is widely used for neural diseases diagnosis. Conventional EEG device uses wired interface which significantly restrained patient mobility. In this paper we present wireless sensors network based on IEEE 802.15.4 standard for wireless EEG devices. WSN enables patient mobility, remote monitoring capability, centralized resource sharing i.e. monitoring & analysis. This results in cost and human resource reduction in diagnosis. The system can also be used in natural disaster emergency situation like hurricane, floods etc in which a dedicated person cannot be provided to each patient so an automated wireless EEG system can take care of device controls. We will analyze cross layer wireless sensor node performance at Mac and Network layer respectively. By applying three different routing protocols namely flooding, Probabilistic broadcast and converge cast the result shows optimum performance analysis of these protocols thus enabling the selection of the most efficient routing protocol for various applications. Keywords – EEG, BCI, Flooding, Probabilistic broadcast, Converge cast, Wireless Sensor Network, IEEE 802.15.4.

Introduction In modern world use of electronic devices has increased exponentially in every field. It has also induced many changes in area of medical diagnosis, imaging and treatment. Therefore hospitals uses several types of electronic medical devices for various applications, such as Portable Vital Signs monitors to measure blood pressure, body temperature, Electrocardiograph (ECG), monitors is used for analysis of heart activity, medication and delivery of fluids into patients body is controlled by Electronically designed Infusion Pumps and likewise many other devices [1]. As there is increasing number of electronic medical devices their implementation need hierarchical based approach, proper infrastructure, and efficient communication protocols. Because of timely needs world has shifted from infrastructure based networks to infrastructure less networks for the communication. As implementation of infrastructure based networks has higher infrastructure cost, more complexity, and need greater level of expertise. It tends to use infrastructure less networks due to low cost, easy implementable and application diversity at the user end. The involvement of communication technology opened up research areas targeting integration of communication technologies in order to develop efficient and smart devices. Several standards were developed like WPAN, Zigbee etc. Out of several different wireless communication standards IEEE 802.15.4 Wireless Sensor Network (WSN) standard is being used for communication purpose. WSN has several applications including military sensing, data broadcasting [2], agriculture [3], environment monitoring [4], industrial automation [5] and patient monitoring [6] to name a few. Wireless sensor network is generally made up of inexpensive power constrained sensors which has a small size and organized in different topologies so that the data can properly be transmitted to the sink. Typically WSN are data centric networks. The main advantage of WSN using in EEG applications includes reliability i.e. to provide complete data from all nodes, scalability so as the number of nodes increases so the quality of service of the network is not degraded, Fault tolerance as in majority of medical scenario data transmission is crucial and by making wireless protocol fault tolerance. It is made sure that the data is being routed from multipath towards the destination. With all the advantages comes the challenges of WSN that needs to be overcome so that it can be used in EEG application scenarios; these includes energy constrain, limited computation capabilities, smaller storage capacity and narrow network bandwidth. In this paper we are proposing a wireless sensor network for EEG detection. Firstly we will discuss background and advantages. Conventional EEG Detector uses wired interface with multiple electrodes placed in order to record the signals. This conventional methodology has several constraints that can be resolved by using the 802.15.4 wireless sensor nodes. We have presented a comprehensive analysis of Mac and Routing Layer by applying three different protocols namely flooding, probability broad cast and converge cast. The analysis is done on the basis of latency, energy consumption and reliability. As well as over the individual node performance and its effects on the whole wireless sensor network.

17

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 The rest of the paper is; in section II Brief background and literature review is presented, followed by Advantage of using Wireless EEG in section III; Section IV contains Methodology and Architecture; Section V explains IEEE 802.15.4 standard; Results are presented in section VI and finally Conclusion in Section VII.

Related Work Electroencephalography (EEG) records electrical activity around the scalp. It measures voltage fluctuation which is generated by ionic current flows within the neurons of the brain [7]. The early detection of Electroencephalography (EEG) was started in 1842 by Carlo Matteucci. In which electrical activity was detected over the frog muscles by using galvanoscopic leg [8]. In 1924 a German physiologist and psychiatrist Hans Berger recorded the first human Electroencephalography (EEG) [9]. Berger recorded the EEG during a neurosurgical operation on a 17 year old boy and paved the way for research in EEG by introducing alpha and beta waves [10]. In the evolution of Electroencephalography (EEG) instrument development various themes were introduced. The need of medical devices incorporated different engineering fields such as electrical, electronics and communication engineering for development of devices to be used for medical purpose. And computerized or digital EEG evolved further as compared to conventional EEG. Because digital EEG provided more functions for analysis such as event detection, mathematical reconstruction possible etc. Furthermore cost of machines was reduced with enhanced feature in digital EEG [11]. This paved a way forward for further research and development in corresponding field. As every application needs a customized way for the collection of the data therefore for collection of EEG signals conventional EEG machines uses wired caps containing electrodes in it for the recoding of EEG signals [12]. Research is being done to improve EEG recording machine to maximize its uses in many other fields; one such example is wireless cap which is also being used as brain computer interface. These devices have many different types of set of electrodes to acquire EEG signals from brain. Therefore cap is being designed in such a way that electrical contact with the head can be reduced in order to facilitate the patient [13]. Usually the wired EEG devices limits the patient mobility resulting patient to bed confined, in long term it causes headache, by using wireless EEG this issue can be eliminated easily, minimizing psychology impact such as and saving money used for psychological treatments. Wireless EEG provides mobility therefore it is not mandatory for a person to be admitted to hospital room for the recording of EEG, it can also be done by having device at home, which will save cost per patient of hospital admission especially in developing countries where the admission fees for a common person is out of reach. At remote locations like village or distant populated areas where there is no facility of a modernized equipped hospitals wireless devices can be used for recording of EEG which can be beneficial for under privileged people residing at that location. Centralized monitoring unit can be setup in order to maintain the EEG records of both hospitalized and remote location patients which will reduce monitoring resources. Moreover wireless sensor nodes provide wire free inter sensor communication which helps the patient’s convenience from hazel of wires over the head. Currently Wireless Electroencephalograms (EEG) devices is in use for transmission of data from brain sensors to a computer via wireless link and paves the way for development of many future medical applications [14]. It also provides brain computer interface enabling a disabled patient to control a robotic arm and other electronic gadgets [15].

Methodology and Architecture In wireless sensor network topology play an important role for network organization, connectivity, performance, reliability, energy efficiency, latency, robustness etc. Presently wsn topology includes star, mesh, tree and clustered hierarchical architecture. Requirement of our application is reliability as most important factor along with acceptable latency and energy consumption. As study suggests that in star topology reliability is product of reliable communication link between each node and the sink with no redundant path available. In cluster hierarchical reliability is the path from sink to top cluster head and goes downwards it is usually used for large network. While mesh provides reliability and robustness as each sensor node has multiple neighboring node for relaying data [16]. Therefore in this paper we proposed wireless EEG sensor network using mesh topology. In Mesh networks data is transmitted only to the nearest neighboring nodes. Provided that Nodes are identical in mesh networks and are also referred to as peer-to-peer connections. Generally there are multiple routing paths between nodes, therefore it is robust to failure due to link breakage of individual nodes. An advantage of this scheme is that all nodes with same computing and transmission parameters may be identical but specific nodes can be made as ‘Cluster head’ which can perform additional functions. In case of failure of any cluster head, another node can be assigned the responsibility of cluster head. The wireless EEG detector node operates at IEEE 802.15.4 standard. In prescribed topology there are total 13 sensor nodes which can either be a Reduced Function Device RFD that has a reduced level of functionality and only hops data to the next node Or a Full Function Device ‘FFD’ which not only transmit but also can make limited decision based on the protocol implemented. And one is Main

18

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Coordinator node. The functionality of Main Coordinator node is to connect to all nodes available in the network; these specific nodes are known as Full function device FFD. FFD exhibits full levels of functionalities which can be used for sending and receiving data, as well as routing of data from other nodes. The nodes are connected in peer to peer scheme. Peer-to-peer networks forms a random connection pattern. Each node can directly access the immediate connected node.

Fig.1. Mesh Topology for EEG Sensor Network

IEEE 802.15.4 Standard IEEE 802.15.4 is the standard used in wireless sensor network. It was defined in 2003 by IEEE 802.15 working group. This standard defines protocol and device interconnectivity using radio communication for Personal Area Network (PAN). ISA100.11a, ZigBee[17], Wireless HART, and MiWi specifications are also based on the same standard and same standard is extended further by adding the upper layers that are not defined in IEEE 802.15.4. This standard gives wireless personal area network (WPAN) focusing on low-cost, lower power consumption, low-speed ubiquitous communication between devices as compared with others such as WI-FI. 802.15.4 basic framework has data rate of 250 kbit/s over the air. Bu data rate can be changed according to application requirement, if lowered nodes will be more power efficient. It has16 channels in 2450 MHz band, 10 channels in 915 MHz band and 1 channel in 868 MHz band. Depending on the application requirement this standard supports star and peer-to-peer topologies. Carrier sense multiple access-CSMA and collision avoidance mechanism is used and. contention based media access is utilized via superframe structure. Main components may include Reduced function device (RFD) and Full function device (FFD). PAN coordinator (FFD) allocates the time slots to devices that have time critical data. Prominent features includes guaranteed time slots(GTS) for real-time suitability, collision avoidance by means of CSMA/CA, power management by energy detection and integrated support for secure communication[18].

Simulation and Results The simulation is done using OMNET++. This platform provides flexibility and comprehensive analysis of cross layer. The major requirement of our application is reliability; hence no one node data can be lost; however on the other hand delay can be tolerated, hence the EEG application is delay tolerated. For our proposed topology we have used three different routing parameters for performance evaluation of sensor nodes. The routing parameters are as follows to get concrete results; Cross Layer Analysis Flooding Flooding is a common and frequently used technique for path detection and information transmission via different nodes. It is cost effective routing strategy which is independent of expensive network topology maintenance that yields towards difficult and complex routing algorithms. Reactive routing approach is used in which each node that receives data or a control packet broadcasts the packet to all of its neighboring nodes. During transmission, a packet opts for all possible routes and paths till the delivery is completed and the packet reaches its destination provided that network is not disconnected [19].

19

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Table.1. Simulation Node Parameters

Physical Layer Parameter Maximum Transmission Power Sensitivity Application Layer Parameters Application Layer Type Header Length Burst Size Mobility Parameters Mobility Type Speed Update interval Battery Parameters Capacity Voltage Mac Layer Parameters Bit rate Detection time Turnaround time Mac Acknowledgement Mac Ack wait duration Mac Max CSMA backoff Contention Window Minimum Backoff exponential Maximum Backoff exponential

1.1 mW -100 dBm Traffic Generator 512 bit 1 Constant Mobility 2 mps 0.1 s

Speed

9999mAh 3.3V 10000bps 0.000128 s / 8 symbols 0.000195 s True 0.1 s 5 2 3 8

Fig.2. Flooding Protocol Routing Network Layer Behavior for Flooding Flooding is used as a network route with the header length of 24 bit. Each node generates its own traffic randomly as the application layer type is traffic generator. As the traffic is generated it is flooded to all neighboring nodes after a node receives the data packet it is further forwarded to other all other neighboring nodes. As a result of flooding the behavior of each node is presented in figure 3. The maximum number of packet received and forwarded is 8 and minimum packet received and forwarded is 3. Number of packets reception at individual node is forwarded equally without any packet drop. There is no loss of data at any individual node which yields that in flooding packets are broadcasted to every possible neighborhood therefore there are multiple routes for data packets to reach its destination which makes the network robust to any node failure and prevents data loss by means of any possible route. As no data loss is observed it is clear to say that by using flooding with mesh topology the reliability of the network increases which results in better through put of the network. Mean hop value is 2-3 hops as each node broad cast the packet which increases number of hops at every single node. It means the network consumes slightly more energy as it is using multiple hops that can

20

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 increase latency and decrease energy efficiency of each node which affects the overall latency and energy consumption of the network. Thus for network layer we can say that by using flooding our network is reliable, energy efficient and robust.

Fig.3. Number of Packets / Mean Hops Vs Individual Node Performance Mac Layer Behavior for Flooding In figure 4 Mac Layer behavior of every node is shown. The maximum back off is at node 1 which is 66 therefore frame drop at this node is also maximum i.e. 10 frame drops. While node 11 has minimum number of back off’s which is 24 hence also has lowest frame drop i.e. 2 frame drop. As every node has almost high Back off it is causing the frame drop at every node with back off duration is less than 1 s. The frame drop at any node does not exceeds from 10 drops. Therefore say the mac layer behavior of flooding will cause some increase in latency and power consumption. By analyzing behavior of both Mac and Network layers; it is clear from the results that by using flooding our network is reliable and robust but with little compromise over energy efficiency and latency.

Fig.4. Number of Back Off / Frame Drop Vs Individual Node Performance

Probabilistic Broadcast In Probabilistic broadcast message is sent only to possible direct neighbors of a node. In case a node receives a message from multiple neighbors, this will be treated as collision and the message will be considered as corrupt by the specified node. Thus in probabilistic broadcasting scheme every individual node on the basis of certain probability decides whether to forward its message or not. Hence chance of message collision lowers which reduces the general communication costs [20]. Network Layer Behavior for Probabilistic Broadcast Probabilistic Broad cast is used as a network route with the header length of 24 bit. By using probabilistic broadcast each node decides whether to forward a packet to its neighbors in order to avoid collision at receiving node. Therefore number of packet received by each node is much less as compared to flooding as shown in

21

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 figure 5. This attribute saves a lot of energy as most of the node energy will not be consumed in receiving multiple packets. Therefore data at each node forwarded is higher as compared to reception which makes the network energy efficient and decreases latency of the overall network significantly. It also yields to mean hop count value ‘1’ which proves that traffic congestion is minimal and collision of data is also eliminated due to probabilistic forwarding of the data packets. Whereas 3 nodes respectively node ‘1’,’2’ and ‘10’ has only forwarded their own sensor data. Hence saving energy and decreasing latency of the sensor network which increases reliability.

Fig.5. Number of Packets / Mean Hops Vs Individual Node Performance Mac Layer Behavior for Probabilistic Broadcast In figure 6 at Mac Layer each node number of back offs does not increases more than 34.In case of flooding it was much higher ranged from 20-70. Therefore back off duration is also very less at each node. This results in significant decrease number of frame drops at Mac layer and does not exceeds from 5 frame drop. The highest number of back off is observed at node 11 i.e. 34 which has resulted in most frame drop at this node to 5 dropped frames. Node 1 & 10 has minimum back off value 2 & 1 respectively therefore these nodes have no frame drop. It can be seen from the results that overall performance of Mac and Network layer using probabilistic broad cast is much more efficient than flooding in terms of energy, reliability, latency but it does not cover robustness as in flooding.


Converge cast Converge cast yields to a communication pattern that converges data multiple node data onto a single node. Flow of data is in such a manner that it flows from a set of nodes or multiple nodes to a single node within the network. Converge cast is just invert in behavior as compared to broadcast functionality, where data is generated from a single node and sent to all multiple neighboring nodes within the network. In Figure 7 a example

22

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 illustrates the functionality of the broadcast and converge cast routing for a small network. Broadcast example is at left side in which ‘A’ node is the source node that generates data packets and direct receiving nodes is B. As B hears the data packet from source node ‘A’ it broadcast the data packets to all of its neighboring nodes that is C and D. This yields the working mechanism of Broadcasting. Second is a converge cast example at right side which explains node A is the sink or the destination node. B, C, and D nodes wants to send a data packet to node A. As Node B lies between node C and D therefore it serves as a relay node for nodes C and D and data packets coming from nodes C and D will be converged at relaying node B and forwarded to destination node A [21].

Fig.7. Broadcast and Converge Cast Protocol Routing Network Layer Behavior for Converge Cast In converge cast routing packets are being generated from node with interval of 1200 seconds. In this multiple packets is being sent from source node to sink which increases duplication at a receiving node. This results in increase latency and energy consumption at each node. Due to duplication of packets at each node sensor network faces congestion and packet drop that yields to non reliability of the network. Received number of packet at each nodes are equally forwarded but results in high duplicate packet reception. The protocol shows less mean hop count along with 7 nodes which has no hop value, this means that each node has traffic congestion which limits the packet forwarding, which increases packet drop. Therefore due to these reasons at network layer converge cast fails to provide reliability, energy efficiency, robustness and has very high latency factor.

e Fig.8. Number of Packets / Mean Hops Vs Individual Node Performance Mac Layer Behavior for Converge Cast At Mac Layer each node has around more than 25 number of back offs. The maximum number of back off is at node 2 i.e. 56 with highest number of frame drop i.e. 8 drop. While the minimum back off is at node 5 which is 26 along with frame minimum frame of 2 as shown in figure 9. It has less back off than flooding but more back off than probabilistic broadcast. And almost every node has frame drops which affects energy level of the network and latency. Therefore analyzing the overall effect of both Mac and Network Layer for converge cast it is appropriate to say that using this protocol will be the most inappropriate approach for our desired application; since it has highest latency factor, consumes a lot of energy, provides non reliability to the network.

23

ISBN: 978-969-8680-32-9



Conclusion In this paper we have proposed wireless EEG system using cross layer approach. The most important requirement of our application is reliability and robustness of the sensor network; however latency and energy efficiency can be compromised. Through extensive simulation analysis of these protocols it is clear from the results that flooding will be more efficient than probabilistic broadcast or converge cast protocols. The probabilistic broadcast can easily provide energy efficient and minimum latency; however robustness and reliability of the network is less than the flooding protocol. Moreover Converge Cast increases latency of the network due to duplication of packet reception at each node. It can also cause route failures and network will not be reliable and energy efficient. Thus making it the most un appropriate protocol to be used in EEG sensor network. Based upon the comprehensive analysis it is proposed that flooding will be most efficient to use in EEG application due to the reason that it shows higher packet delivery ratio, have average hop count, no duplication of packets resulting in enhanced reliability of the network, low energy consumption and acceptable latency as compared to other routing protocols.

References 1. 2. 3.

4.

5.

6. 7. 8. 9. 10. 11. 12. 13.

Wireless Network Sensor and Server Architecture For Legacy Medical Device by Paul Frehill, Desmond Chambers, Cosmin Rotariu. Tien-Wen Sung, Ting-Ting Wu, Chu-Sing Yang, Yueh-Min Huang, “reliable data broadcast for zigbee wireless sensor networks”, international journal on smart sensing and intelligent systems, vol. 3, no. 3, september 2010 Pedro Mestre, , Carlos Serôdio, Raul Morais, Jorge Azevedo, Pedro Melo-Pinto, “Vegetation Growth Detection Using Wireless Sensor Networks”, Proceedings of the World Congress on Engineering 2010 Vol I WCE 2010, June 30 - July 2, 2010, London, U.K. Ang, Won-Suk, William M.Healy, “Assessment of Performance Metrics for Use of WSNs in Buildings”, 2009 International Symposium on AUTOMATION and ROBOTICS in CONSTRUCTION (ISARC 2 009) p.570-575, 2009. F. Chen, T. Talanis, R. German, and F. Dressler, “Realtime Enabled IEEE 802.15.4 Sensor Networks in Industrial Automation,” in IEEE Symposium on Industrial Embedded Systems (SIES 2009). Lausanne, Switzerland: IEEE, July 2009, pp. 136–139. N. GolmieD. CypherO. Rebala “Performance analysis of low rate wireless technologies for medical applications”, Computer Communications Volume 28 , Issue 10 June 2005 Pages:1266-1275 ISSN:0140-3664. Niedermeyer E. and da Silva F.L. (2004). Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Lippincot Williams & Wilkins. “History and Evolution of Computerized Electroencephalography “ Thomas F Collura, Section Of Neurological Computing, Department Of Neurology, The Cleveland Clinic Foundation, Cleveland, Ohio, USA. Hans Berger (1873–1941), Richard Caton (1842–1926), and electroencephalography. Hans Berger (1873-1941)--the history of electroencephalography. The advantages of digital over analog recording techniques, 1VA Medical Center, West Los Angeles, Department of Neurology, UCLA, CA 90073, USA. http://www.brainproducts.com/productdetails.php?id=9 Wearable Brain Cap with Contactless Electroencephalogram Measurement for Brain- Computer Interface Applications, Proceedings of the 4th International IEEE EMBS Conference on Neural Engineering Antalya, Turkey, April 29 - May 2, 2009.

24

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 14.

15. 16. 17. 18.

19. 20. 21.

Integrating Wireless EEG into Medical Sensor Networks by Behcet Sarikaya, M. Abdul Alim, and Siamak Rezaei, Department of Computer Science, University of Northern British Columbia, 3333 University Way, Prince George, BC, Canada V2M 4Z9 Analysis of a Commercial EEG Device for the Control of a Robot Arm G.N. Ranky, S. Adamovich New Jersey Institute of Technology Biomedical Engineering Department, Fenster Hall Newark, NJ 07103. A Performance Comparision Of Different Topologies For WSN by Akhilesh Shrestha and Liudong Xing, University of Massachusetts Dartmouth, 285 Old Westport Road, North Dartmouth, MA 02747. Gascón, David (February 5, 2009 ). "Security in 802.15.4 and ZigBee networks". IEEE 802.15.4-2006 Standard for Information technology- Telecommunications and information exchange between systems- Local and metropolitan area networks- Specific requirements Part 15.4: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (WPANs). IEEE Computer Society. Revision of IEEE Std 802.15.4-2003). ISBN 0-7381-4996-9. 8 September 2006. Book : Wireless sensor network by John, Page# 203 Probabilistic Broadcast Mark Hayden and Ken Birman, Computer Science Department, Cornell University. Research Paper Radial Coordination for converge cast in wsn by Qingfeng Huang and Ying Zhang

25

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 A Context-Aware Switching Mechanism between Heterogeneous Wireless Networks Muhammad Diyan Department of Computing and Technology Iqra University, Islamabad, Pakistan

Abstract The integration of heterogeneous wireless networks invokes many design and implementation issues regarding the dynamic topology and mobility management. The mobility management in heterogeneous wireless networks is becoming a hot area of research during the last couple of years. One of the important key elements of a switching system is the triggering of a switching process. Traditional triggering schemes are mainly based on the signal strength coming with each packet from the current network. But, signal strength based triggering schemes are mainly suffered from the ping-pong effects. Therefore, to provide a mobile user with context-aware switching capabilities, we proposed a scheme that initiates a switching process based on the proposed optimal signal strength indication. Moreover, the target network selection is based on the ELECTRE decision model. The contexts used in the proposed switching mechanism include the bandwidth, delay, jitter, packet loss, and monetary cost. The ELECTRE decision model normalized these contexts in a matrix and then compared it with the reference context it selects a network providing with highest contexts. The proposed approach is tested in various mobility scenarios with different number of mobile user having variable speeds. The performance evaluation shows that the proposed scheme selects a network and stay in the network for longer time. Similarly, the proposed switching process reduces the ping-pong effects and number of switching.

Introduction One of the visions of future generation of networks is to provide a Mobile User (MU) with the service of always connected anywhere anytime. Now-a-days, the Network Access Operators (NAO) is providing a variety of services to enhance the quality of its network and functionality. The Quality of Service (QoS) of the network depends on a number of contexts and criteria. These criteria and contexts are further divided into two parts. The first contain those criteria that are important for a user-satisfaction and the second part consists of parameters that directly depend on the network resources. In the last decade, several schemes have been presented to enhance the quality of the network context based schemes. The interconnection of network and user centric resources gives birth to three new concepts: 1) Quality of Context (QoC), 2) Quality of Experience (QoE), and Perceived Quality of Service (PQoS). Both of QoS [1] and QoE are closely related with each other in the context of improving the performance of mobility and handover management. The QoS manage the interconnection between the user, application, and network. Similarly, QoE provides the user with the best quality and hence improves the user satisfaction. In heterogeneous wireless networks (HWN), enabling context-aware services and provides user-centric approach is still a challenging job. Traditional, schemes are mostly based on those parameter and contexts which create imbalance between the network resources and the user satisfaction level. For example, most the switching mechanisms from one network to another in HWN environment are initiated when the link connectivity (RSSI) of current network is dropped below a particular level as shown in Figure 1. One of the problems related with using RSSI level for switching between networks is that it does not provide the user-satisfaction. Because, sometimes user do not want to initiates switching from one network to another. Similarly, The Signal to Interference and Noise Ratio (SINR) and bandwidth based switching mechanisms are mostly suffered from noise and bandwidth dynamics problems, respectively. Therefore, identifying those parameters that satisfy both user and network requirements is a challenging job. There are two types of parameters presents in case of network-centric approach [2]. The first types of the parameters are directly proportional to the QoS of the network. In general, if we increase the value of this type of parameter, an increase in QoS of the network is achieved. Similarly, there are other types of parameters that are indirectly depending on the QoS of the network. The second set of parameters includes delay, jitter, network load, etc. Furthermore, using these two different types of parameters for the selection of a network while switching between HWN creates the imbalance problem between these parameters.

26

ISBN: 978-969-8680-32-9


Fig.1. RSS based switching between networks In order to provide a balanced relation between these parameters, different decision-making schemes have been proposed. These schemes efficiently solve the problem of imbalance between these direct and indirect parameters. The decision of switching from one network to another is a context-aware problem. Therefore, in the last decade, researchers diverted their focus to integrate the decision-making methods with context aware switching for mobility management in HWN. This integration can only be possible by classifying the parameters into different groups [3]. This classification is presented in Table 1. Table 1. Switching of Network based Parameter Classification

Classification

Parameters

User centric

Cost, security, PQoS, QoE, QoC, etc.

Network centric

Delay, jitter, bandwidth, data rate, RSSI, Bit Error Rate (BER), and SNIR, etc.

Terminal based

Battery consumption level, network interfaces, speed, and location information, etc.

NAO based

Profit, user level, load on network, and security level, etc

In order to improve, the performance of the existing context-aware switching decision, we proposed a scheme that initiates the switching from one network to another on the basis of an optimal RSS based threshold hold scheme. Moreover, the target network selection while switching from one network to another is performed using the ELECTRE decision model. The rest of the paper is organized as follows. In section 2, a detail literature review of the existing switching schemes is presented. Section 3 presents the proposed model in detail. Section 4 illustrates the results and discussion section and finally, the paper is concluded in section 5.

Related Work Recently, researchers proposed several different schemes to enhance the functionality of the context-aware switching mechanisms. These schemes are target different contexts which directly affect a switching process from one network to another [4] [5]. Most of these decision models ranks the available networks and then select one of them providing with the best QoS. These decision models select a network on the basis of QoS, which directly depending on a network centric approach. But, providing a user centric based switching methods is discussed by very few researchers [6]. Recently, researcher has been proposed various schemes depending on multi-attribute decision-making mechanisms. One of a similar decision-making method has been presented in [7]. The proposed scheme employs the approach of Simple Additive Weighting (SAW) for decision making. This approach first assigns weights to each criterion and then constructs a normalized matrix of it. The proposed scheme multiply each criterion with the respective user defined weights as shown in following equation 1. (1) 𝑄𝑄𝑄𝑆𝑆𝑆 = 𝑀𝑀𝑀𝑖∈𝑁 ∑𝑛𝑗=1 𝑤𝑗 𝑎𝑖𝑖

27

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Where N is the number of parameters, a ij is the normalized context parameter j of a network i and w j is the weights assigned by the user. The working of Weighted Product Method (WPM) is similar to that of SAW; only the weighted criterion is multiplied instead of adding it. Similarly, a scheme based on ranking of the available networks on the basis Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) has been presented in [8]. The TOPSIS decision model selects an ideal situation and then compares each criterion of the network with the ideal situation. The distance of each criterion is then mapped in normalized decision matrix and then using following Equation 2, the network with highest rank is selected for switching. 𝑅𝑖∗ =

𝐻𝑖− + 𝐻𝑖 +𝐻𝑖−

(2)

Where 𝐻𝑖− and 𝐻𝑖+ represents the distance from the negative and positive ideal situations, respectively, and 𝑅𝑖∗ represent the degree of each network. A scheme based on another decision model called Grey Relational Analysis (GRA) has been presented in [9]. The GRA decision model is based on the grey relational coefficient (GRC). The network which has GRC closer to the ideal GRC is selected for the switching. The working of GRA is presented in following Equation 3. 𝐺𝐺𝐺𝑖 =

1

𝑂𝑖

𝑂

𝑗 ∑𝑗=1

∆𝑚𝑚𝑚 + ∆𝑚𝑚𝑚

(3)

∆𝑖 +∆𝑚𝑚𝑚

Where O i is the total number of objectives and ∆𝑚𝑚𝑚 and ∆𝑚𝑚𝑚 represent the difference between two minimum and maximum values of an objective in a sequence. A scheme has been proposed based on Vlsekriterijumska optimizacija I KOmpromisno Resenje (VIKOR) multi decision modeling in [10]. The proposed scheme utilizes the qualities of different contexts and then decides the best alternative among them. The user first computes the degree of regret and maximum favorable utility. The proposed scheme selects the best alternative by considering one which provide with maximum utility and low regret. The decision models are used richer set of context for the switching decision from one network to another. Handling many numbers of input criteria is challenging and time-consuming job. Thus, to deal with such problems, a scheme has been presented based on the fuzzy logic systems [11]. A modified RSS is used to triggered handover. The modified RSS predication is obtained using the differential predication algorithm. Before, switching from one network to another a filtration is performed on the user to separate them on the basis of velocity. The proposed scheme efficiently reduces the cell-dropping rate and unnecessary handoffs. The IEEE published a new standard called Media Independent Handover (MIH) in 2008. This new standard is used to perform and control the switching between IEEE based technologies and cellular. The proposed standard is extensively studied by different researchers [12] [13] [14]. The MIH standard proposed a new logical layer between the lower layers and upper layers. The communication between these layers is handled through different events that work under the MIH function. The MIH function is located in the heart of MIH logical layer, and this function handles all of the communication between the Access Point (AP) or Base Station (BS) and an MU. The MIH standard performs handover triggering on the basis of the RSS. If the connectivity between the MU and an AP or BS drops below than a threshold of RSS, the MU will trigger the handover. Similarly, the selection of the network is also performing through the RSS level. Recently, researchers proposed different schemes that are using other parameters such as SNIR and bandwidth, etc. for initiating the handover [15]. The integration of handover decision mechanism with MIH standard has been proposed in [16]. The MIH standard is made enhanced by the concatenating a new entity called the Vertical Handover Management Engine for fast handover decision method. The proposed scheme performs handover decision on the basis of the MU previous experience in the same environment. The handover efficiency is significantly increased, and the handover delay is decreased.

The proposed Scheme The proposed scheme is further divided into two phases, in order to clearly understand the working of the proposed scheme. Switching Initiation Phase In general, a switching phase is initiated if any one of the following three criteria matches. 1) If a node wants to switch off a particular interface while communicating with the access network through an interface, it initiates switching to a different access network using another interface. 2) When a switching attempt fails to a particular network, a switching is triggered. 3) If the RSS from the current network drops below a predefined threshold, then the MU initiates switching to an access network with strong RSS. We use a threshold mechanism to

28

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 provide efficient switching initiation. The threshold mechanism reduces the number of false switching indications, number of failing attempts to a network with overloaded APs/BSs. Since the coverage distance around WLAN AP is small than the coverage area of the UMTS BS, the number of switching attempts inside an AP coverage area is greater than that of a BS. On the boundary of coverage area, we put a threshold of RSS level (𝜃𝑡ℎ ) for initiating the network selection. Let d 0 represent the coverage area of an AP and BS, then the time when the RSS drops below 𝜃𝑡ℎ is given by 𝑇𝑡ℎ =

𝑑

(4)

2×𝑣

where v is, the velocity of the MN and d is the distance of coverage area of an AP or BS. Let ∆t be the time difference between the initiation of network selection phase and the receiving network selection parameters from the available networks. The following inequality must be true for seamless switching condition. 𝑇𝑡ℎ ≥ ∆𝑡

(5)

From (10) and (11), we get 𝑑0 ≥ 2 × (𝑣 × ∆𝑡)

(6)

𝜃𝑡ℎ [𝑑𝑑] = 𝐾1 − 𝐾2 log( (1 − 𝛿)𝑑0 )

(7)

According to the signal propagation model in [17], the threshold should be set based on the distance (1 − δ)d from an AP or BS where δ is a fluctuation to represent the variation in network data rate dynamics. The value of δ is taken between 0 and 1. Thus, the RSS for handover initiation can be written by where K 1 represents the antenna gain and the signal wavelength, and K 2 represents the path loss factor. When the RSS drops below from θth [dB], the MN initiates handover. In the proposed scheme, we efficiently optimize the working of RSS parameter. The false handover initiation is significantly minimized using the proposed threshold mechanism.

Selection of target network The applications of ELECTRE are widely adopted in different fields. The ELECTRE performs pair-wise comparison of the alternative by considering each context separately. There are different contexts used by researchers to decide a network for switching. We used five different contexts which include delay (D), bandwidth (BW), jitter (J), utilization (UT), and Monterey cost (MC). The first and important step of ELECTRE is to combine these contexts in a vector for decision-making purpose. Let N represent a network then the vector of contexts become as follows: 𝑁𝑖 = [𝐷𝑖

𝐵𝐵𝑖

𝐽𝑖 𝑈𝑈𝑖

𝑀𝑀𝑖 ]

(8)

Similarly, for N different networks, the context is arranged in a matrix M NW as follows:

𝑀𝑁𝑁

𝐷1 𝐵𝐵1 𝐷2 𝐵𝐵2 = � ⋯ ⋯ 𝐷𝑚 𝐵𝐵𝑚

𝐽1 𝑈𝑈1 𝑀𝑀1 𝐽2 𝑈𝑈2 𝑀𝑀2 � ⋯ ⋯ ⋯ 𝐽𝑚 𝑈𝑈𝑚 𝑀𝑀𝑚

(9)

Moreover, we provide every user with the functionality to design its own desired network. This concept enables context-awareness in the proposed scheme. The desired vector has the contexts with maximum possible values of each context. The main advantage of designing desired vector is to provide the user to compare the contexts of available networks with the desired vector. Similarly, to remove the irrelevant information obtained from available networks, each context is normalized using following relation. max𝑗=1⋯𝑚 �𝐶𝑗 �−𝐶𝑖 𝐶𝑖̅ = (10) max𝑗=1⋯𝑚�𝐶𝑗 �−min𝑗=1⋯𝑚 �𝐶𝑗 �

Where 𝐶𝑖̅ represent the normalized context. The users are also providing with functionality to assign weights to each context obtained from each network. The weights assignment enables a user to adjust the value of each context for during switching. Thus, a

29

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 user gets successful in arranging the contexts in a way that can help in selecting an optimal network according to user requirements. The weighted normalized matrix WN NW matrix is shown as follows. � �� 𝑤𝐽 𝐽1̅ 𝑤𝑈𝑈 �� 𝑈𝑈1 𝑤𝑀𝑀 �� 𝑀𝑀1 ⎤ ⎡ 𝑤𝐷 𝐷1 𝑤𝐵𝐵 𝐵𝐵1 �� ̅ �� 𝑤 𝐷 𝑤 𝐵𝐵 𝑤 𝐽 𝑤 𝑈𝑈 𝑤 𝑀𝑀 𝐵𝐵 2 𝐽 2 𝑈𝑈 2 𝑀𝑀 2⎥ 𝑊𝑁𝑁𝑁 = ⎢ 𝐷 2 (11) ⋯ ⋯ ⋯ ⋯ ⎢ ⋯ ⎥ �𝑚 𝑤𝐵𝐵 𝐵𝐵 ��𝑚 𝑤𝐽 𝐽𝑚 ̅ 𝑤𝑈𝑈 𝑈𝑈 ��𝑚 𝑤𝑀𝑀 𝑀𝑀 ��𝑚 ⎦ ⎣𝑤𝐷 𝐷 In order to compare the networks with each other to get the optimal network, the ELECTRE uses the concept of concordance and discordance. Both concordance and discordance measure the satisfaction and dissatisfaction level of available networks. Thus, the network that provides the best values of the proposed contexts is selected for the switching. To achieve the concordance level of network, we define a threshold value for each context. The threshold value (𝜃𝐶 ) is computed using following equation. 𝜃𝐶 =

𝑚 ∑𝑚 𝑖=1 ∑𝑗=1 𝐶𝐶𝑖𝑖

(12)

𝑚∗(𝑚−1)

Where m and CD represent the total number of contexts and elements in concordance matrix, respectively Similarly, we define another threshold (𝜃𝐷 ) on the discordance level as shown in following relation. 𝜃𝐷 =

𝑚 ∑𝑚 𝑖=1 ∑𝑗=1 𝐷𝐷𝑖𝑖

(13)

𝑚∗(𝑚−1)

Where DC represents the elements presents in the discordance matrix From above relations, we can conclude that the context that are below than the 𝜃𝐶 is passed to the 𝜃𝐷 . Thus, using this way, the elements which have high values in the concordance is used for the selection of network for switching. The network which have more concordance context than the another network is the best choice of switching. In general, the architecture of proposed approach using ELECTRE is shown in following Figure 1. ELECTRE CONCORDANCE (C) Networks Contexts Weights

Available Networks

High Contexts C-Level Low Contexts

Pair-wise comparison

C- High Contexts

Contexts D- Low Contexts Weights

Best Network Selection

ELECTRE DISCONCORDANCE (D) High Contexts Networks Contexts Weights

D-Level Low Contexts

Fig.2. Working of proposed scheme using ELECTRE method

Results and Discussion The proposed scheme is tested in a heterogeneous environment consisting of three different networks i.e. WIFI, WIMAX, and Cellular using C++ programming lanuage. The coverage area of cellular and WIMAX BSs are same (500m) while that of WIFI is low equal to 100m. The number of applications assigns to each to each MU randomly. We supposed three different types of applications ranging from low bandwidth to high bandwidth i.e. elastic, audio cal, and multimedia applications. The weight assignment is dependent on each user, for example, a user can assign a high weight to one type of applications while low to another. The weight values are assigned between 0 and 1. The highest value is 1 while the lowest value is 0. It also depends on the applications in use while switching process. Similarly, the user can assign weights to each context. For example, a user is interested in high bandwidth with a reasonable cost. Then the user assigns a high weight to the bandwidth and low weight to the cost. Similarly, the user can change its weight assignment after performing switching from one network to another. Thus, the proposed scheme becomes dynamic. An example scenario is shown in Figure 3. The big black, blue and small blue circles represent the cellular BS, WIMAX BS, and WIFI AP, respectively.

30

ISBN: 978-969-8680-32-9


Mobile User

WIFI Access Point

WIMIX/Cellular Base Point

Fig.3. An Example Simulation Scenario The MU is randomly distributed in different locations and is shown in the above scenario. The MU performs switching at different locations from one network to another. We test the performance of the proposed scheme with existing scheme [8] in context of a number of switchings. The proposed scheme performed very less number of switching because of the proposed switching triggering approach. The comparison is shown in following Figure 4. Proposed Switcing mecha nism

Existing Scheme

350

Number of switchings

300 250 200 150 100 50 0 10

20

30

40

50

60

70

80

90

100

Simulation Time (sec)

Fig.4. Analysis of a number of switchings Similarly, the MU stays for a longer time in the network because of the selection of proposed network selection scheme. The average staying time of MU in case of proposed scheme is compared with existing scheme [12]. We found that the proposed scheme stays for a longer time on the network and hence reduce the frequent switching between networks as shown in following Figure 5.

Conclusion In this paper, we proposed an efficient switching and network selection scheme. The proposed scheme initiates the switching mechanism on the basis of an optimized RSS value. The threshold set for initiating switching between networks greatly optimized the working of existing switching mechanisms. The proposed switching mechanism reduces the number frequent switching as well as the ping-pong effect. Similarly, the proposed network selection scheme significantly maximizes the time require for staying in a particular network. The existing schemes are suffered from frequent switchings and less stay time in the network. The proposed scheme is compared with the existing schemes, and it outperforms the schemes in contexts of number of switchings and stay time in the network.

31

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Existing Scheme

Proposed Scheme

18

Average Stay time (sec)

16 14 12 10 8 6 4 2 0 1

2

3

4

5

6

7

8

9

10

Number of Switchings

Fig.5. Average stay time

References 1. 2.

3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

13.

14. 15. 16.

17.

P. Tinkhede and P. Ingole, "Survey of handover decision for next generation," in International Conference on Information Communication and Embedded Systems, Chennai, India, 2014. M. Khan, G. Park, W. Cho, G. Seong and K. Han, "A Handover Management Scheme Based on User-Preferences and Network-Centric Approach," KSII Transactions on Internet and Information Systems, vol. 9, no. 1, pp. 344 357, 2015. M. Zekri, B. Jouaber and D. Zeghlache, "Context aware vertical handover decision making in heterogeneous wireless networks," in IEEE 35th Conference on Local Computer Networks (LCN), Denver, CO, 2010. H. Liao, L. Tie and Z. Du, "A Vertical Handover Decision Algorithm Based on Fuzzy Control Theory," in First International Multi-Symposiums on Computer and Computational Sciences, Hanzhou, 2006. J. Márquez-Barja, T. Calafate and J.-C. Cano, "An overview of vertical handover techniques: Algorithms, protocols and tools," Computer Communications, vol. 34, no. 8, pp. 985 - 997, June 2011. N. A. a. Y. G.-D. Q-T. N-Vuong, "Novel Approach for Load Balancing in Heterogeneous Wireless Packet Networks," in IEEE Network Operations and Management Symposium Workshops, Salvador da Bahia, 2008. P. Chan, R. Sheriff, Y. Hu and P. Conforto, "Mobility management incorporating fuzzy logic for heterogeneous a IP environment," IEEE Communications Magazine, vol. 39, no. 12, pp. 42 - 51, Dec 2001. H.-Y. Huang, C.-Y. Wang and R.-H. Hwang, "Context-Awareness Handoff Planning in Heterogeneous Wireless Networks," in Lecture Notes in Computer Science, Xi’an, China, 2010. M. Khan and K. Han, "An Optimized Network Selection and Handover Triggering Scheme for Heterogeneous Self-Organized Wireless Networks," Mathematical Problems in Engineering, vol. 2014, p. 11, June 2014. M. Alkhawlani, "MCDM based Joint Admission Control for Heterogeneous Wireless Networks," in Seventh International Conference on Computer Engineering & Systems (ICCES), Cairo, 2012. L. Xia, L.-g. Jiang and C. He, "A Novel Fuzzy Logic Vertical Handoff Algorithm with Aid of Differential Prediction and Pre-Decision Method," in IEEE International Conference on Communications, Glasgow, 2007. G. R. Tamijetchelvy, "An optimized fast vertical handover strategy for heterogeneous wireless access networks based on IEEE 802.21 media independent handover standard," in Fourth International Conference on Advanced Computing (ICoAC), Chennai, India, Dec 2012. M. Navitha, R. Tamijetchelvy and G. Sivaradje, "Robust vertical handover scheme using IEEE 802.21 Media Independent Handover," in International Conference on Communications and Signal Processing (ICCSP), Melmaruvathur, India, 2014. M. Khan and K. Han, "A Zone-Based Self-Organized Handover Scheme for Heterogeneous Mobile and Ad Hoc Networks," International Journal of Distributed Sensor Networks, vol. 2014, p. 8, June 2014. "A Review of Handover Techniques in Wireless Ad hoc Networks Based on IEEE 802.21 Media Independent Handover Standard," IETE Technical Review, vol. 31, no. 5, pp. 353 - 361, Sep 2014. M. Zekri, B. Jouaber and D. Zeghlache, "An enhanced media independent handover framework for vertical handover decision making based on networks' reputation," in IEEE 37th Conference on Local Computer Networks Workshops, Clearwater, FL, 2012. N. A. a. Y. G.-D. Q-T. N-Vuong, "A user-centric and context-aware solution to interface management and access network selection in heterogeneous wireless environments," Computer Networks, vol. 52, no. 18, p. 3358–3372, Dec 2008.

32

ISBN: 978-969-8680-32-9


Performance Evaluation of Secure Force Symmetric Key Algorithm Shujaat Khan, Mansoor Ebrahim*, Kafeel Ahmed Khan Iqra University, Defence View Campus Karachi, Pakistan

* Sunway University, Selangor, Malaysia [email protected],[email protected], [email protected]

Abstract Since the devices in wireless sensor network (WSN) are usually small in size, resource constraint is one of the critical issues that need to be taken into consideration. Therefore, WSNs must be designed in such a way that a compromise should be established by balancing between the security level and the processor overhead. The aim of this research work is to evaluate and analyze the performance of proposed Secure Force (SF) Algorithm [1] in terms of security and computational complexity. As presented in our previous work, the SF algorithm has a simple architecture consisting of basic mathematical operations and requires less computational resource as compared to other standard encryption algorithms. In this paper the key pluses and open issues of our proposed SF algorithm based on certain performance evaluation parameters such as Avalanche effect, Execution time, Entropy change, and image histogram are highlighted. Moreover, based on the evaluation results, the paper also suggest the possible solutions for the open issues that will be considered in the future work. Keywords: SF, WSN, Security Algorithms, Secure Force, Avalanche Effect, Image Encryption

Introduction Conventional cryptographic algorithm is not suitable for WSN because of its distinctive characteristics [3]. The key issue in designing the cryptographic algorithms for WSN is to deal with the trade-off among security, memory, power, and performance. To achieve the high security requirements, numerous efforts have been made on assessing cryptographic algorithms and proposing energy efficient ciphers [4, 5] for WSN [6]. In our previous research work we proposed a low-complexity symmetric key algorithm for WSN, denoted as Secure Force (SF) and compare it with several existing symmetric key algorithms based on architecture, flexibility, and security level [1]. The major pros and cons of the earlier proposed low-complexity symmetric key Secure Force (SF) algorithm based on certain parameters are highlighted in this paper. The introduction of Secure Force algorithm is discussed in Section 2. In section 3 the performance evaluation criteria is described. All the simulation results based on evaluation criteria are presented and discussed in Section 4. Finally, the paper is concluded in Section 5.

Secure Force Algorithm The design of SF algorithm provides low-complexity architecture for implementation in WSN. To improve the energy efficiency, the encryption process consists of only five encryption rounds. It has been suggested in [19] that a lower number of encryption rounds will result in less power consumption. In order to enhance the security, each encryption round encompasses six simple mathematical operations operating on only 4 bit data (designed to be compatible with 8-bit computing devices for WSNs). This is to create enough confusion and diffusion of data to encounter different types of attacks. The key expansion process, which involves complex mathematical operations (multiplication, permutation, transposition and rotation) to generate keys for the encryption process, is implemented at the decoder. This shifted the computational burden to the decoder and indirectly, this will help to increase the lifespan of the sensor nodes. However, the generated keys must be transmitted securely to the encoder for the encryption process. In this case, the LEAP (Localized Encryption and Authentication Protocol) [18] is adopted. It is an energy efficient, robust and secure key management protocol that is designed for the WSN. Overall, the process of SF algorithm consists of 4 major blocks. The detail description of each block of the proposed algorithm can be found in [1].

33

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 • Key Expansion Block: Key expansion is the primary process that is used to generate different keys for encryption and decryption. Different operations are performed in order to create confusion and diffusion. This is to reduce the possibility of weak key as well as to increase the key strength. The round keys (Kr) are derived from the input cipher key by means of the key schedule. The process consists of two components: key expansion and round key selection. The key expansion performs logical operations (XOR, XNOR), left shifting (LS), matrix multiplication using fix matrix (FM), permutation using P-table and transposition using T-table. • Key Management Protocol: The key can be securely sent to the encoder with the aid of LEAP [18].It is a simple and energy efficient protocol designed for large scale WSN, which allows secure key establishment through the use of four types of keys. They are known as the individual key, group key, cluster key, and pair wise shared key. • Encryption Block: The encryption process is initiated once the keys generated by the key expansion block are securely received by the encoder through the secure communication channel created by using the LEAP protocol. In the encryption process, simple operations, which include AND, OR, XOR, XNOR, left shift (LS), substitution (S boxes) and swapping operations, are performed to create confusion and diffusion. • Decryption Block: The decryption process is just the reserve of the encryption process described above.

Performance Evaluation Criteria The evaluation of SF algorithm was carried out on certain well know parameters used by various authors [11], [12], [13], [14] and [15] in order to assess the performance of different conventional algorithms. Avalanche Test A very well-known parameter used to analyze the security (randomness) of an encryption algorithm. The avalanche test measures the effect of change in the number of bits of encrypted text (cipher) due to one bit change in either key or plain text. The avalanche test is considered best if half of the bits of the cipher text are changed as per the strict avalanche criterion SAC [2]. Execution Time (Encoding/Decoding) The execution time is one of the essential parameter that needs to be considered along with security in the development of an encryption algorithm. The execution time of an encryption algorithm is defined as the total time required for the encoding/decoding of a particular data. Image Histogram Image histogram is a recently used parameter; it shows the randomness in the encrypted image data distribution. In this parameter the histogram of encrypted and unencrypted images are compared to know overall change in the data image intensities due to encryption. Image Entropy Digital Images are combination of discrete valued pixels, combined together to form a visual perception of image. Image entropy measure is the simplest parameter used to analyze the randomness in the encrypted image. In this parameter the difference between the original and encrypted image’s entropy is measured. The greater the entropy change, the better will be the encryption. Entropy of an image can be calculated by the given relation (1) E = ∑N i=1 X i (log2(X i ))

�3�

Where ‘E’ is entropy of image, ‘X’ is the probability of the intensity level in image and ‘N’ is the total number of intensity levels.

34

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Experimental Results Avalanche Test The results in the table1 show that Secure Force algorithm can cause a significant number of bits change with the single bit change in the key or plain text. On Average SF 64-bit can change 58.2% of cipher bits due to the change of one bit in text or key bits where as the avalanche results of DES is 65.63%[23] and for AES it is 44.92%[12]. Table 1. Avalanche Test Results Sno.

Key

Input

Avalanche Test

Output

FFFFFFFFFFFFFFFF

000A4A6DE8DB6667

925BEDEAD4E631EB

FFFFF7FFFFFFFFFF

000A4A6DE8DB6667

B83D3E9D07911E50

FFFFFFFFFFFFFFFF

000A4A6DE8DB6667

925BEDEAD4E631EB

FFFFFF7FFFFFFFFF

000A4A6DE8DB6667

4D2CF304C7D5E1EB

000A4A6DE8DB6667

FFFFFFFFFFFFFFFF

388D97729772388D

000A4A6DE8DB6667

FFFFFFFFFFFFDFFF

D5501BBEE5D8F550

000A4A6DE8DB6667

FFFFFFFFFFFFFFFF

388D97729772388D

000A4A6DE8DB6667

FFFFFFFFEFFFFFFF

E772F99D6DD8288D

1

0.6250

2

0.5156

3

0.5938

4

0.5938

Mean percentage avalanche value

0.5820

Simulation Time SF is a light weight algorithm, it uses very low amount of computer resources. The results in table 2 show overall computation time for encryption. The comparison of SF with standard algorithms is possible on common hardware like FPGA or DSP. Table 2. Simulation Time Analysis and Experimental Setup Mean Simulation Time

27.5 mSec

Processor

Intel Core2Duo T6500 @ 2.10GHz

RAM

4.00 GB

Operating System

Windows 8 64-bit

MATLAB Version

R2011a 64-bit

Image Entropy Entropy is the measure of information contents of the data, more random the data more difficult it is to be recognized after encryption. The entropy change for six popular images, namely Cameraman [22], Rice [22], Lena [21], Football [22], ORL Faces [20], and Onion [22] is presented in table 3. On average 10.39% of entropy change is observed. Image Histogram (Intensity Variation) Histogram is a very useful way to analyze the effect of encryption over the image. The ideal resultant histogram after encryption should be flat (even). SF shows pretty decent results, however, for few tests SF cause the minor changes in the histogram; this is because of the original intensity distribution of image. The results for selected images are shown in Figure 1 and 2.

35

ISBN: 978-969-8680-32-9


Fig. 1.( a) : Cameraman original and encrypted image

Fig. 1. (b):Rice original and encrypted image

Fig. 1.(d): Football original and encrypted image

Fig. 1.(c): Lena original and encrypted image

Fig. 1.(f):Onion original and encrypted image Fig. 1.(e): ORL Faces original and encrypted image

Fig. 1. Various Original and encrypted images

Table 3. : Image Entropy Test

Entropy

Entropy

Percent

(original)

(Encrypted)

Change

256X256

7.0097

7.8705

12.28

Rice.tif

256X256

7.0115

7.9448

13.31

3

Lena.jpg

220X220

7.4618

7.9643

6.73

4

Football.jpg

256X320

6.6861

7.8210

16.97

5

ORLFace.jpg

490X467

7.5332

7.9723

5.83

6

Onion.png

135X198

7.3299

7.9300

8.19

7.1720

7.9172

10.39

SNO.

Image

Dimension

1

Cameraman.tif

2

Mean Entropy values

Open Issues and Possible Solutions The results in this paper show the best performance results after the extensive testing; SF algorithm has some built-in flaws in its key generation. In our future work we are planning to propose the improved version, few observed flaws in the proposed SF algorithm are as follows. Weak Keys: During Avalanche test it is observed that many results produce no significant change in the cipher when change in key or plain text occurred.

36

ISBN: 978-969-8680-32-9


Fig. 2. (a). Histogram of "Cameraman"

Fig. 2. (b). Histogram of “rice”

Fig. 2.(c). Histogram of “Lena”

Fig. 2.(d). Histogram of “Football”

Fig. 2.(e). Histogram of “ORL Faces”

Fig. 2.(f). Histogram of “Onion”

Fig. 2. Image histogram results Visual results: Visual results for all six images show that although it is almost impossible to identify the actual content of the image, but we a very good idea of the image can be perceived as the outlines or the borders of the objects are not well randomized (changed). Such weakness in the SF is due to the fact that the algorithm is tested for only 64-bit architecture and also because of the above mentioned weak key issue. Computation Time: Although SF shows good results and its performance is comparable to other algorithms [7], [8], [9], [10] and [15] in terms of computation cost, but it is still not as claimed. The originally proposed SF is not as promising as was expected to be, but the flexibility of its architecture can be exploited to improve the performance easily. This problem can be easily rectified by applying few optimizing programming techniques as well.

37

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Conclusion Modern advancements in the field of communication and computer networks increase the challenges for network security, scalability and reliability [16], [17]. Power constrained networks like wireless sensor network (WSN) demands an algorithm which can provide a reliable security at an affordable computational cost. Recently proposed Secure Force algorithm is one of the candidates for WSN. In this paper we implemented the SF(64-bits) on MATLAB® platform and perform the standard tests for image and text data, SF(64-bits) perform reasonably well in terms of computational resources (time) and randomness. In the result of security analysis some weaknesses are identified in its default architecture configuration that can be rectified by taking advantage of its flexibility.

References 1.

Ebrahim, Mansoor, and Chai Wai Chong. "Secure Force: A low-complexity cryptographic algorithm for Wireless Sensor Network (WSN)." Control System, Computing and Engineering (ICCSCE), 2013 IEEE International Conference on. IEEE, 2013.

2.

Webster, A. F.; Tavares, Stafford E. (1985). "On the design of S-boxes". Advances in Cryptology - Crypto '85. Lecture Notes in Computer Science 218. New York, NY,: Springer-Verlag New York, Inc. pp. 523–534. ISBN 0-387-16463-4. J. Luo, P. Papadimitratos, and J.P. Hubaux, “Gossicrypt: WSN data confidentiality against parasitic adversaries”, SECON '08. 5th Annual IEEE Communications Society, June 2008. Perrig, R. Szewczyk, J. Tygar, V. Wen, and D. Culler, “Spins: security protocols for sensor networks”. ACM/Kluwer Wireless Networks, 8(5):521–534, 2002. Karlof, N. Sastry, and D. Wagner. TinySec: A link layer security architecture for wireless sensor networks. In Proc, of the 2nd ACM SenSys, 2004. M. Ebrahim, S. Khan and U.B. Khalid,” Symmetric algorithm survey: a comparative analysis”. International Journal of Computer Applications 61(20), January 2013, pp. 12-19. USA National Institute of Standards and Technology,"SkipJack and KEA algorithm specifications (Version 2.0)," May 1998. D. Hong, J. Sung, S. Hong, J. Lim and S. Lee “HIGHT: A new block cipher suitable for low-resource device”, Cryptographic Hardware and Embedded Systems Vol. 4249, 2006, pp. 46-59. R.L. pavan, M.J.B. Robshaw, R.Sidney, and Y.L. Yin. “The RC6 Block Cipher”. Ver 1.1, August 1998. S.I. Huang, and S. Shieh, “SEA: Secure encrypted data aggregation in mobile WSNs”, International Conference on Computational Intelligence and Security, IEEE, 2007. Kumar, A., & Tiwari, M. N. (2012). effective implementation and avalanche effect of AES. International Journal of Security, Privacy and Trust Management (IJSPTM), 1(3/4), 31-35. Shivkumar, S., & Umamaheswari, G. (2011, July). Performance Comparison of Advanced Encryption Standard (AES) and AES key dependent S-box-Simulation using MATLAB. In Process Automation, Control and Computing (PACC), 2011 International Conference on (pp. 1-6). IEEE. Zeghid, M., Machhout, M., Khriji, L., Baganne, A., & Tourki, R. (2007). A modified AES based algorithm for image encryption. International Journal of Computer Science and Engineering, 1(1), 70-75. Elminaam, D. S. A., Abdual-Kader, H. M., & Hadhoud, M. M. (2010). Evaluating The Performance of Symmetric Encryption Algorithms. IJ Network Security, 10(3), 216-222. Yoon, J. W., & Kim, H. (2010). An image encryption scheme with a pseudorandom permutation based on chaotic maps. Communications in Nonlinear Science and Numerical Simulation, 15(12), 3998-4006. Ebrahim, Mansoor, Shujaat Khan, and UmerBin Khalid. "Security Risk Analysis in Peer 2 Peer System; An Approach towards Surmounting Security Challenges." arXiv preprint arXiv:1404.5123 (2014). Ebrahim, Mansoor, Shujaat Khan, and Syed Sheraz Ul Hasan Mohani. "Peer-to-Peer Network Simulators: an Analytical Review." arXiv preprint arXiv:1405.0400 (2014). S. Zhu, S.Setia,, S. Jajodia, “LEAP: Effcient security mechanisms for large scale distributed sensor networks”. In Proc. of the 10th ACM Conf. on Computer and Communications Security. Washington DC USA, (2003), pp. 62-72. R. Chandramouli, S. Bapatla, and K. P. Subbalakshmi, “Battery power-aware encryption”, ACM Transactions on Information and System Security, Vol. 9, No. 2, May 2006, , pp. 162–180. “The Database of Faces” formally known as “The ORL Database” Computer Laboratory, University of Cambridge, UK Roberts, Lawrence G (1961). "Picture Coding Using Pseudo-Random Noise". MIT MATLAB, The MathWorks, Inc Mandal, A. K., Parakash, C., & Tiwari, A. (2012, March). Performance evaluation of cryptographic algorithms: DES and AES. In Electrical, Electronics and Computer Science (SCEECS), 2012 IEEE Students' Conference on (pp. 1-5). IEEE.

3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

38

ISBN: 978-969-8680-32-9


Performance Analysis of 3-D Orthogonal Frequency Division Multiplexing Transceiver for Satellite Communication Muhammad Farrukh Shah1, Tariq J. S. Khanzada2, Syed Naveed Jaffri2 1

COMSATS Institute of Information & Communication Technology Lahore, Pakistan IICT, Mehran University of Engineering and Technology, Jamshoro, Pakistan

2

[email protected], [email protected], [email protected]

Abstract Satellite communication can still be dominant over terrestrial communication system to provide the coverage over the areas where terrestrial networks can’t be deployed or are not feasible. Recent developments in modulation techniques have ameliorated the performance of wireless communication systems. This paper is intended to address the implementation of 3-D OFDM transceiver over satellite communication. In this work, 3OFDM transceiver is implemented using 2-D Inverse Fourier transform (IFFT) which is used to allocate signal to sub-carriers and to perform modulation of the signal. 3D OFDM transceiver is evaluated using QAM modulation having levels 4, 8 32 and 64. 3D OFDM is also compared with conventional OFDM transceiver in terms of bit error rate (BER). Results vivid that 3-OFDM has better error performance when compare with conventional OFDM transceiver. It is because minimum Euclidian distance in 3-OFDM signal mapper is much farther than the 2-D signal mapper. 3D OFDM is appropriate choice as a modulation scheme to be used in satellite communication Keywords—3 D OFDM, OFDM, 2 D IFFT, MATLAB

Introduction Satellite communications have embarked revolution in the field of technology. Starting in the year 1965, Satellite communications have penetrated into many services which include communication, broadcasting and navigations. Recent revolution in the field of Satellite communication is the replacement of analog technology with the digital and the software control over variety of system components [1]. Satellite communication is being used for broadband services to provide high data rates to end user. Satellite mobile communication is also getting attention now days. Figure 1 shows satellite space segment.

Fig.1. Satellite space segment

39

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Recently, satellite communications have gained much attention for high data rate applications. For high data rate services inter symbol interference (ISI) is matter of concern due to the frequency selective fading (FSF). The performance of Orthogonal Frequency Division Multiplexing (OFDM) has been exploited in wireless communication services. OFDM has been adopted as a physical layer modulation technique in IEEE 801.11/WiFi, IEEE 802.16/WIMAX and Digital Video Broadcast-Terrestrial(DVB-T)[2]. For satellite communication OFDM has been adopted by BVB-SH [3] which is a broadcast standard for delivering multimedia services over hybrid satellite/ terrestrial networks to a variety of fixed terminal with compact antennas and very limited directivity [4].OFDM modulation has been explored in JTEO‟s research for finding optimal air interface for MILSATCOM network [4]. The use of guard intervals makes OFDM one of the best modulation techniques for physical layer transmission. 3-D OFDM is emerging OFDM modulation technique which makes use of 3D signal mappers to map the signal [5]. 3-D OFDM uses 2 D inverse Fast Fourier transform (IFFT) to modulate 3 D signals. 3D OFDM system has better error performance when compared with the tradition OFDM system. 3D OFDM transceiver has been realized in MATLAB in this work using 2D IFFT to modulated OFDM symbols. This paper is organized as follows. Section II describes conventional OFDM followed by 3 D OFDM with mathematical model. Section III highlights simulation setup and results description. Finally section IV presents conclusion.

3D OFDM for Satellite Communication The rapid pervasion of mobile telephony and equally growth of voice, data and video communication over the internet, justify the great expectation for next generation communication. Wireless Broadband Multimedia Communication Systems (WBMCS) are now under considerations [6]. The next generation WBMCS systems may be able to integrate variety of functions and able to support numerous applications. WBMCS require providing their users to avail high data rates. To support such high data rates choice of suitable modulation technique is required [7]. OFDM seems to be appropriate transition proficiency for WBMCS systems. OFDM is being used as a powerful modulation technique in wireless communication to tune interface standards such as Institute of Electrical and Electronics Engineers (IEEE) 802.11/Wireless Fidelity (WIFI), IEEE 802.16/ World Wide Interoperability for Microwave Access (WiMax) and Digital Video Broadcast-Terrestrial (DVB-T). OFDM is also applied in satellite and military applications [7]. OFDM has been exploited in Joint Terminal Engineering Office (JTEO„s) research effort for identifying the optimal air-interface MILSATCOM networks via Wideband Global Satcom (WGS) system [8]. OFDM has been investigated for satellite communication since last decade. Satellite systems use high data rate transmission, consequently there always is a possibility of Inter Symbol Interference (ISI) to occur. OFDM is a potential candidate which can eliminate ISI effectively by use of orthogonal subcarriers [9]. Fig 02 shows OFDM based satellite communication

Fig. 2. OFDM based Satellite system [2]

In the sequel we are reproducing our OFDM analytical model presented in reference [10] and [11] with minor changes where necessary. Fig. 3 and Fig. 4 shows OFDM transmission system and OFDM carriers in time domain respectively.

40

ISBN: 978-969-8680-32-9


Fig. 3. OFDM system

Fig. 4. OFDM Sub carriers [3] OFDM is a kind of multi-carrier transmission system scheme which splits single stream of data into identification bit of low data pace subcarriers. In OFDM high data charge per unit r is divided into sub-blocks of small rate for transmission. Then the low rate blocks are mapped onto symbols of lower transmission rate r s = r/k. The symbols have a continuance of T s and they are separated by guard interval of continuance T g shown in Fig.5.

Fig. 5. Bit mapping and symbol duration with guard interval The subcarriers in OFDM are orthogonal and it is achieved by choosing BT s = N over T s . The orthogonal subcarriers in time and frequency domain are shown in Fig.06.

Fig. 6. Orthogonality of subcarriers (a) Time domain and (b) Frequency domain view [Courtesy of R.Prasad , Chapter 1 page 23] The subcarriers are modulated using QAM or PSK modulation scheme. In OFDM ISI is effectively eliminated by using T g longer than the max channel delay spread τ c as shown in Fig.7. ISI is removed due to the orthogonal subcarriers [11]. It is given as (1). To = Ts + Tg

Tg ≥ τc

(1)

41

ISBN: 978-969-8680-32-9


Fig. 7. Elimination of ISI by keeping Tg ≥ τc [11] By using additional signal processing at the receiver, ISI can be removed in time varying channels. For timevarying channel pilot signals are transmitted for updating channel characteristics [11]. The OFDM frame is shown in Fig. 8.

Fig. 8. OFDM symbol frame [10] The transmission of N PSK/QAM symbols with distinct frequencies is shown in Fig.09

Fig. 9. Distinct frequencies of carriers [11] The subcarrier frequencies are given as; 𝑓n =

𝜇+𝑛 µ = integers, n = 1,2,3 … . N 𝑇𝑠

𝑓𝑛+1 − 𝑓𝑛 =

𝐵𝑇 1 𝑁 = ≥ 𝑇𝑠 = 𝑁 𝐵𝑇 𝑇𝑠

The PSK or QAM symbols that are assigned to the nth subcarriers can be expressed as {B 1 (n), B 2 (n) ………………B M (n)} where B i (n) = a i (n) ejФn

i = 1,2,3….. M n

(3)

th

and M n is the configuration of n subcarrier. The number k is given as 𝑛

𝑘=�

The mth OFDM symbols is then expressed as 𝑀

𝑆𝑚 (𝑡) = �

𝑛=1

𝑛+1

𝑁

𝐿𝐿𝐿2 (𝑀𝑛 ) = 𝐿𝐿𝐿2 �

𝑀𝑛

𝑛=1

𝑁

𝑆𝑚 (𝑡) = �

𝑛=1

𝑎𝑚 cos[2𝜋𝜋𝜋(𝑡 − 𝑚𝑇𝑜 ) +

(𝑛)

Ф𝑚

The M-ary coding of k-bits blocks for the OFDM symbol can be seen as being shown in Fig.10. The nth digit have M n different likelihood {B 1 (n) , B 2 (n) ………………B M (n)}

42

ISBN: 978-969-8680-32-9


Fig. 10. Bit streams converted to parallel groups [10] Overall OFDM symbols can be formulated to be M = M 1 , M 2 …………………… M N = ∑𝑁 𝑛−1 𝑀𝑛

which have k – bit segments calculated by 2k equivalent to 𝑁

𝑀= �

𝑀𝑛 = 2𝑘

𝑛=1

In 3 D OFDM system, a symbol can be represented as a column vector S k = (x k , y k , z k )T , 0 ≤ k ≤ M-1, where M is the size of a signal constellation and T is the transpose operation. x, y and z are the coordinates of a symbols. The base band signal of the 3D OFDM system can be given as [5]. S = [S o T, …………. S N-1 T] x o , ………….. x N-1 = y o , …………... y N-1 z o , ………….. z N-1 N is the number of sub carriers. The 2 D IFFT is used to modulate OFDM symbols and the modulated symbol can be represented as [5] 𝑠(𝑛2 , 𝑛1 ) =

𝑁−1 𝑁−1 𝑗2𝜋𝑘1 𝑛1 𝑗2𝜋𝑘2 𝑛2 1 � � 𝑆(𝑘1 , 𝑘2 ) . exp � � . exp � � 𝑁1 𝑁2 𝑁1 𝑁 2 𝑘=0 𝑘=0

for 0 ≤ n 1 ≤ N 1 – 1 and 0 ≤ n 2 ≤ N 2 – 1 where k 1 and k 2 are the indices for columns and rows. The columns represent sub carriers of the OFDM system and rows shows dimension of the signal space for the 2 D matrix respectively. Fig 11 depicts 3D OFDM system and Fig. 12, Fig 13 show constellation in 3D of modulated symbols respectively.

Fig. 11. 3DOFDM system

43

ISBN: 978-969-8680-32-9


Fig. 12. 3D signal constellation with M = 4

Fig.13. 3D signal constellation with M = 8

Simulation Setup Simulations have been done using MATLAB. The 3D OFDM modem has been tested for BPSK, 8, 16 and 32 QAM. 192 subcarriers with 8 pilots have been selected for the modem. 2 D IFFT is used to modulate OFDM symbols and 256 points IFFT have been taken. The length of cyclic prefix is kept variable. AWGN channel is wielded in a modem. Table 1. 3D OFDM configuration parameters Parameters

Parameters set

No of Sub carriers

192

2D FFT

256 points

Pilots

8 pilots

Modulation techniques

BPSK,QAM

Fig. 12 shows the BER performance of the 3D OFDM modem employing BPSK compared with theoretical BPSK .It is depicted that performance of 3D OFDM closely follows theoretical BPSK. It is because mapping of two BPSK symbols in 3 D would result significance change in Euclidean distance. Hence error performance would remain same.

44

ISBN: 978-969-8680-32-9


Fig. 14. BER Performance of 3D OFDM with BPSK modulation Fig. 13 shows BER performance of 3D OFDM using 4,8,32 and 64 QAM. The satellite channel is assumed to power limited. Hence by increasing level of modulation higher bandwidth is achieved at the cost of BER. Hence 32 QAM or 64 QAM are suited when use in 3D OFDM for satellite communication.

Fig. 15. BER performance of 3 D OFDM with M-QAM modulations Fig. 16 depicts performance comparison of 3D OFDM with OFDM. It is shown that 3D OFDM performance is better than OFDM because of 3 D OFDM symbol mapping. This enhanced error performance can be used to mitigate the problem of FSF which results ISI. Thus 3 D OFDM is suitable modulation technique which can be implemented in satellite communication for providing broadband services and other high data rate services.

Fig.16. BER performance comparison of 3D OFDM with OFDM

Conclusion This paper presents implementation of 3D OFDM modem for satellite communication. The Matlab window

45

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 provides easy and interactive way of designing communication modem which facilitates the improved 3D OFDM modem. The performance of 3D OFDM is evaluated using 4,8,32 and 64 QAM assuming power limited satellite channel and it is shown that 32 QAM gives better performance. 3D OFDM comparison is presented with OFDM using AWGN channel. It is shown that 3D OFDM exhibits good BER performance than the OFDM. 3D OFDM has the major advantage of good error performance hence it is well suited for satellite broadband communication services.

References 1.

N. Chuberre, G.E. Corazza, M.G. Francon, C. Nussli, C. Selier, A.Vanelli-Coralli, P. Vincent, ”Satellite Digital Multimedia Broadcasting for 3G and beyond 3G systems”, IST Mobile Summit 2004, Lyon France, 28-30 June 2004. 2. P. Salmi, M. Neri, and G.E. Corazza, ”Design and Performance of Predistortion Techniques in Ka-band Satellite Networks”, 22thInternational Communication Satellite Systems Conference (ICSSC), AIAA,May 2004 3. A. Ganz, Y. Gong, and B. Li, “Performance Study of Low-Earth Orbit Satellite Systems,” IEEE Trans. Commun., vol. 42, 1994, pp. 1866–71. 4. Dissanayake, S.D.; Armstrong, J., Comparison of ACO-OFDM, DCO-OFDM and ADO-OFDM in IM/DD Systems, Lightwave Technology, Journal of , vol.31, no.7, pp.1063,1072, April1, 2013 5. S. G. Kang, “An OFDM with 3-D signal mapper and 2-D IDFT modulator,” IEEE Communications Letters, vol. 12, no.12, pp.871–873, 2008 6. Borkar, N.J.; Bormane, D.S., BER performance of OFDM system with adaptive modulation, Complex Systems (ICCS), 2012 International Conference on , vol., no., pp.1,8, 5-6 Nov. 2012 7. Nieto, J.W., An investigation of coded OFDM and CEOFDM waveforms utilizing different modulation schemes on HF channels, Communication Systems, Networks and Digital Signal Processing, 2008. CNSDSP 2008. 6th International Symposium on , vol., no., pp.358,362, 25-25 July 2008 8. Abu-alhiga, R.; Haas, H., Subcarrier-index modulation OFDM, Personal, Indoor and Mobile Radio Communications, 2009 IEEE 20th International Symposium on , vol., no., pp.177,181, 13-16 Sept. 2009 9. Kelley, Brian, Software defined radio for broadband OFDM protocols, Systems, Man and Cybernetics, 2009. SMC 2009. IEEE International Conference on , vol., no., pp.2309,2314, 11-14 Oct. 2009 10. Khanzada, T.J.S.; Ali, A.R.; Omar, A.S., "An Analytical Model for SLTDM to Reduce the PAPR and ICI in OFDM Systems for Fast Varying Channels," IEEE Multitopic Conference, INMIC '06. pp.56,61, 23-24, Islamabad, Pakistan, Dec. 2006. 11. Khanzada, T.J.S.; Memon , S; Jaffari , N.A., “An Efficient Channel Model for OFDM and Time Domain Single Carrier Transmission Using Impulse Responses” , Mehran University Research Journal of Engineering & Technology, Vol 30, No.1, pp 95-100, Jan 2012

46

ISBN: 978-969-8680-32-9


Generating Alerts using Context-aware Security and Brahms Model for Customer Service Improvement Abid Ghaffar1,2, Mohamed Ridza Wahiddin1, Asadullah Shaikh3,4, and Akhlaq Ahmad1,5 1

Department of Computer Science, Kulliyyah of Information and Communication Technology, International Islamic University Malaysia, P.O Box 10, 50728 Kuala Lumpur, Malaysia. [email protected], [email protected] 2 Department of Computer Science, Foundation Year Program, Umm Al-Qura University, Makkah, Saudi Arabia. 3 Faculty of Computer Science and Information Technology, Institute of Business and Technology Karachi, Pakistan. [email protected] 4 Department of Computer Science and Information Systems, Najran University, Najran, Saudi Arabia. [email protected] 5College of Engineering and Islamic Architecture, Umm Al-Qura University, Makkah, Saudi Arabia. [email protected]

Abstract Every organisation set some objectives to achieve success which is closely related with the quality assurance. Human behavior brought many challenges in variety of situations in an organisation which is sometimes untraceable phenomena. The real challenge is to control the collective human behavior at customer service department in an organisation which may pose a serious threat to the top management in terms of its defined objectives. Human behavioral activities can be monitored using Brahms Model Technique along with the Warning Alert Generation System which would maximize the efficiency and improve the system performance at customer service level. In this paper, we propose context aware security measure which would complement the system performance in terms of human error detection and rectification. Generating timely alerts using context aware security measure along with Brahms Model would improve the system performance. Keywords: Brahms Model, Human Behavior, Warning Alerts, Context Aware Security, Cognitive Science, Organisational performance.

Introduction Human behavior is a key issue while dealing with organisational collective performance and improvement [3], [24]. Every institution has some goals and objectives, but some organisations do not come up with progress and success. Tools and equipment might be replaced but employee replacement is not so much easy, it takes long time to set up an organisation and coordination among its employees [3]. It has been revealed that the most important asset and strength of an organisation is its employees, and good organisation takes care of it [3]. Customer service improvement in any organisation is a primary concern and related with the quality assurance. Unfortunately, problems in customer service department are not addressed properly and organisational performance continued to suffer [1], [2], [3], [12]. The need of an hour is to control the human behavior factor and monitor the job activities of employees in the customer service department. Once we are able to capture the human behavior in an organisation, we shall be able to rectify the human error timely and effectively. Brahms Modeling and Sumulation tool provides us an opportunity to capture the human behavioural activities in an organisation [5], [6], [17]. The captured human behaviour using Brahms Model could be used as an input into the Context Aware Secuirty Model which would filter the information as an output. Context Aware Security Model is depended on a rule-based system which serves as an engine. Alert could be easily generated to the concerned person in the customer service department, once information is verified through Context Aware Security measure. Human behavioural activities in the customer service department could be improved and monitored by the application of Context Aware Security measure using Brahms Model and Warning Alert Generation System. The remainder of the paper focuses on Context Aware Security measure along with Warning Alert Generation System and Brahms Model which would rectify human behavioural errors in the customer service department in an organisation. Section II focuses on the development of the strategy proposed for generating alerts. Section III follows up with the related material which was used for the solution of the problem. Sections IV, V and VI carry details about Brahms Model, Context Aware Security Model and Warning Alert Generation System accordingly. Section VII explains the contribution part, while section VIII draws conclusions and future work.

47

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Research Methodology Our research methodology started from bibliographic research approach and then qualitative and design research is used for the proposed solution. Online survey carrying twenty two questions has been conducted from different professionals working at different levels in different international organisations [2]. The questions were asked about their daily job activities and it has been concluded that human behaviour errors exists in different organisations while performing different job activities and as a result, organisational objectives are compromised [2]. There is no timely alert system exists which could rescue the system and minimize the human errors. Therefore, we propose Context Aware Security measure along with Brahms Model and Warning Alert Generation System for the solution of existing problem. It would rescue the system from the failure state to the success state. Brahms Modelling and Simulation tool is considered as the suitable tool for capturing human behavioural activities at the work practice system [2]. We chose Brahms tool for monitoring job activities of an employee in the customer service department [2]. It would help us to monitor and detect the human error in the system. Ethnography study was performed to capture the data during job activities of different employees in an organisation [18], [19]. Two departments of International Islamic University, Malaysia were considered as case studies, one is International Student Division (ISD) and the other is Graduate School of Management (GSM). Data is captured in the form of personal obervations, pictures and video footages during working schedule of the department [18], [19]. Data sets would be used for the modeling and simulation of job activities of different employees in the system. Quantitative research methodology would ensure the following factors while moving towards the solution of a problem: • Finding different tools for monitoring human behavioural activities in the work practice system • Choosing the suitable tool among different available tools for capturing human activities • Analysing formalism for the selected tool Design research is also known as improvement research which would help us to solve different problems at different levels [15], [16]. It is divided into the following steps: • Knowledge flows • Process steps (Awareness of problem, Suggestion, Development, Evaluation and Conclusion) • Output Context Aware Security Model receives an input from the Brahms Model and information is verified from the rule-based system as shown in Fig. 1. Appropriate alert may be generated for the rectification of human error in the system which may cause survival of the system.

Related Work Abid Ghaffar et al. Feb. 4, 2015 compared different multi-agent tools with the Brahms Modelling and Simulation tool and proved that Brahms is the suitable tool for capturing human behavioural activities in the work practice system [2]. Abid Ghaffar et al. Jan. 2015 presented an idea about the framework to generate warning alerts using Brahms Modeling and Simulation Tool. The paper focuses on monitoring human activities and generating alerts in case of human error in the work practice system [1].

48

ISBN: 978-969-8680-32-9


Fig. 1. Framework for generating alerts using Context Aware Secuirty Measure and Brahms Model.

Abid Ghaffar et al. May 2013 proposed a framework which used Brahms Model and Simulation Technique for monitoring human activities in an organisation. Alerts are generated at the time of human error in the work practice system [23]. Martin Chapman 2013 presented an idea to improve the decision making process of available automated software tools to protect the networks or other secuirty systems using the concept of context aware security measure [21]. Mohamad Fauzan Noordin 2013, presented his ideas about the relationship between Information Communication Technology (ICT) and Islam. He focused on heartware among the peopleware which is the most important element in the components of information technology. Heartware determines the direction of the peopleware [24]. Andy Pasztor and Drew Hinshaw 2013, revealed that human error was the main cause of Nigeria Air plane crash and pilot was made responsible for the major cause of accident [13]. Stephen T. Robbins et al. 2012 discussed about different aspects of organisational behaviour which involves interpersonal skills, management and behavioural science. These factors play a vital role to achieve organisational objectives [3]. Geong Sen Poh et al. 2012, proposed a security framework called Human Behaviour Security Framework. It would simulate the human activities acrosss an organisation which involves security related issues like information leakage to protect the information flow [11]. Jim Blythe and L. Jean Camp 2012, presented an idea about an effective communication and warning alerts using mental model approach of users. This would improve the user behaviour in the network security environment [10]. Saad Almutairi et al. 2012 discussed about the application of context aware security systems based on the previous history of users. Time and location keeps on changing which needs different approach like context aware system for providing security to the systems [20]. Lorrie Faith Cranor 2011, proposed a framework called human in the loop framework (HILF). She presented an idea about the decision making process of users is based on their background education, psychology, emotions, knowledge and motivational factors [8]. Jim Blythe USC et al. 2011 discussed about risk factors involved with users for possible cyber threats during improper messaging and communication. Cyber threats are common and users do not understand alerts and messages properly. Users may be protected from possible cyber threats, if proper communication is peformed with users through alerts [9].

49

ISBN: 978-969-8680-32-9


Cristian Bravo-Lillo et al. 2011, discussed about security warning dialogues which people do not understand properly and consequently, lead the user to wrong decisions. If these warning dialogues are designed properly then people might be secured from possible loss or threats [4]. L. Jean Camp 2009, presented an idea about Privacy and Security issues which is based on certain policies. Lack of risk communication is involved in Security and Policy matters specially in case of medical, environmental and life’s style [7]. Maarten Sierhuis et al. 2007, discussed about modelling and simulating work practice system at an individual level where people, objects, timings, activities, geography, knowledge and communication is involved. Brahms language environment provides an opportunity which can simulate work practice system which is developed at NASA Ames Research Center [6]. Scott Shappell and Doug Wiegmann 2004, discussed about the causes of civil and military accidents. They examined 16000 US civil and military accidents and found that human errors had major role in these accidents [14]. Maarten Sierhuis et al. 2002, presented an idea about modeling and simulation of work practice system using Brahms. People perform different tasks using machine and objects at different timings and locations. They communicate with each other and use knowledge to achieve set targets. Victoria lunar mission work system was considered for the application of Brahms [5]. Maarten Sierhuis 2001, explained about collaborative work practices in an organisation. He proposed a framework which is based on human centered approach where people collaborate each other using different tools in an organisation to peform the assigned tasks [17]. Chris Johnson 1999, suggested that organisations are failed mostly due to human failure. He proposed that there is connectivity between human failure and organisational failure [12]. Kevin C. Gross et al. 1998 presented an idea about generating mail messages based on rule-based system. Actions are define based on certain conditions and appropriate rule is fired [22]. Brigitte Jordan 1996, discussed about ethnography study to understand about work practice system in an organisation. She also focused on data collection through team of ethnographers in order to analyse complete information about work flow[19]. Brigitte Jordan 1994, suggested different methods to perform Ethnography Study in an organisation to collect data. The methods involved Interviews, Observations, Questionnaire and Video Recordings. Data could be collected from different perspectives, for example, emic data and etic data [18]. Existing research is focused on the monitoring of human behaviour activities in the work place environment and generating alerts without verification and counter check. It does not give us a solution to detect the human errors and generating alerts using the concept of Context Aware Security measure. Context Aware Security would be coupled with the Brahms Model and Warning Alert Generation System in order to detect and rectify human error in the work practice system.

Brahms Model (BM) Human error detection has always been a real challenge in any organisation and as a result, organisation fails to deliver [3]. Job activities in the customer service department could be monitored and captured directly using Brahms Modeling and Simulation technique [5], [6], [17]. Brahms provides an opportunity to define the humanmachine interaction system at certain time and location. It gives us details like, how human is interacting with his colleagues, dealing with machines, communicating using different tools like computers or papers to perform different activities at certain time and location. On the basis of different aspects and dimensions in an anogranisation, Brahms Model (BM) is further sub-divided into further seven models known as Agent Model, Object Model, Timing Model, Activity Model, Communication Model, Knowledge Model and Geography Model [5], [6]. All these different models complement each other and capture activities of an individual at certain interval of time and location in an institution. Ethnography study would be used to capture and monitor the data used by Brahms Model [18], [19]. Data collection in the form of images, videos, real-time observation of different activites would be captured in order to be used as input in the Brahms Model. Complete picture of a customer service department can be viewed using Agent-Viewer tool in the Brahms Modeling and Simulation process [6].

Context Aware Security Measure Context Aware Security has played a signifact role in the field of Artificial Intelligence [20], [21]. Every employee in the customer service department has significant role and responsibilities to achieve desired targets.

50

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 All the activities of an employee could be monitored in the work practice system using Brahms Modeling and Simulation Technique [5], [6]. The context of each employee job-profile could easily be recorded for future references and decision making process. Any unusual activity against any employee job-profile would be checked before taking some decisions. Context Aware Security measure would help us to filter the information about each employee in case of miscalculation or wrong observation. This is equivalent to counter check the situation before we take some decision against an individual who’s credibility can not be questioned easily. In other words, Context Aware Security measure may protect the interests of employees and the employer both at the same time. Human behaviour error detection with an application of Context Aware Security feature may fine tune our observation and monitoring the activities of an employee in the right direction. It would ensure that alerts are generated to the concerned person in the customer service department based on his user profile and background history. Consequently, alerts would be generated precisely using Brahms Model and Context Aware Security in order to detect and rectify the human error in the system as shown in Fig. 2. We can design the mathematical model for the context aware security by considering different parameters in which the related task is performed by an employee in an organisation [20], [21]. The customer service department has different number of employees and tasks in compliance with the objectives of an organisation. We represent “ m ” as number of employees and “ n ” as number of different tasks assigned to different employees. In general, the symbol "τ p " represents a task assigned to an employee; where i

= i 1,= 2,3, , n; p 1, 2,3, , m . The symbol “ τ 4 ” respresents 3rd task for a 4th employee. A set of primitive contexts would be required which could define an appropriate alert for an employee. The symbols C, C 1 , C 2 , C 3 , C 4 , C 5 , C 6 , and C 7 represents different contexts based on certain job environment. 3

C1 = “Priority”, represents three values of range, “High”, “Medium”, and “Low”. C2 = “Time Allocation”, which represents time interval. We consider ∆t = t2 − t1 , where t1 and t2

represent

the start and finish time.

C3 =

“Job Acknowledgement”, represents two values either, “Yes” or “No”. This context represents either an

employee knows about the job allocation to him or information is still unknown. In case of “Yes”, the employee would be communicated with ∆t

= t2 − t1 , which shows time duration, and in case of “No”, a gentle reminder

would be sent to an employee. In case of any further delay, the manager would analyze the situation and may assign the task to an alternative employee.

C4 =

“Employee Leave at time t2 and t1 ”, which represents two values “Yes” or “No”. In case of “Yes”, the time could be adjusted within ∆t . If an employee on an emergency leave or there is an unavoidable situation, then the same task has to be reported to an official in order to be assigned to some other employee.

C5 =

“Employee Presence at workplace”, which carries two values, either “Yes” or “No”. It could be

confirmed by the employee attendance database either an employee was present in the office or remained absent. If “No”, then manager would analyze the context accordingly and may adjust the time within ∆t .

C6 =

“Instrument Working Conidtion”, it carries two values, either “True” or “False”. This primitive context

will explain the physical condition of instruments like computer, fax machine and other objects required to perform the task. If there is any problem, the alternative resource would be used.

C7 =

“Communication”, which represents the medium of communication for example, Email, Fax, Telephone,

Verbal Communication and Paper Communication. The above set of primitive context would be defined collectively as given below:

C = C1 , C2 ,C3 , C4 , C5 , C6 ,C7 For example;if we have a set of primitive context as,

C = High, 48Hrs, No This means that an employee needs a gentle reminder to acknowledge the acceptance of task with receipt.

51

ISBN: 978-969-8680-32-9


= If C

High, 48Hrs, Yes, +12 Hrs

This means an employee was on leave and there is a need to adjust 12 hours for the time allocated to the task.

C = High, 48Hrs, Yes,0, Yes,Working The relevant context would be alarming, if the task has not been accomplished on time and an alert may be generated based on the premitive context. The rule based system which may be applied in the following form [22]; If condition 1 , Condition 2 , …..Condition n Then Action 1 , Action 2 ….Action n

Fig 2. The exclusive communication among three models.

Warning Alert Generation System Reputable organisations spend lot of budget into the training and improvement programes of their valuable employees and try to retain their skilled force. They understand the real strength and value of their well trained staff. Every employee in the customer service department has specific role and job responsibilities, but it has been observed in some cases, employees do not comply with their job responsibilities very well and consequently, organisational performance continue to suffer and objectives are compromised [12], [13], [14]. Timely alerts to the employees of an organisation could play a vital role in terms of rectification of human error in the system [9], [23]. Designing and preparing messages or alerts understandable by users or employees would carry its own importance and help them to take corrective measures while performing their job activities [4], [7], [8], [10], [11]. Wrong message communication would carry only wrong decision making [4], [9]. Sometimes employees do not understand their role and make mistakes [3]. Key mistakes always bring failure into the system and institutional objectives are compromised [3]. Warning Alert Generation System using Brahms Model and Context Aware Security would generate the alerts in case of human error in the system [1], [23]. We can focus on the problematic areas and may liberate the system from the failure state to the success state by using Warning Alert Generation System along with Brahm Model and Context Aware Security. Warning Aert Generation System comprised of different alerts based on different human errors in the customer service department and becomes activated once human error occurred in the system [1], [23]. It is coupled with Brahms Modeling and Simulation tool and Context Aware Security Measure, an alert is generated once human error is triggered in the system which may emancipate the system from the cause of serious failure.

52

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Our Contribution The main objective for the proposed framework is to ensure that alerts are generated properly and filtered in case of human error in the work practice system. Every employee in an organisation is equally important in terms of his responsibilities and duties [3]. In this paper, we propose context aware security measure which would keep a counter check before alerts are generated to the concerned person in the work practice system. Context Aware Security would keep an employee user-profile based on his daily activities in the system and verified before alerts are generated to the concerned people. Rule based system serves as an engine for the Context Aware Security and plays a significant role in decision making. Alerts would be suspended automatically, if there is any contradiction in the information. On the other hand, alerts would be generated after confirmation through the Context Aware Security using Brahms Model. Fig. 1. Shows the framework comprised of different components including Brahms Model, Context Aware Security Measure and Warning Alert Generation System. Brahms Model captures the human behavioural activities in the work practice system and connected with the Context Aware Security Model which is based on Rule Based System. Warning Alert Generation Ssystem takes the input from the Context Aware Security Model and generates the alerts using cellular network to the work practice system in the form of feedback loop.

Conclusion and Future Work Monitoring and controlling human behavioural activities at the work place is indeed a challenging task and organisational strength and objectives depends upon these factors. Detection and rectification of human error in the system is a great question mark. In this paper, we propose Context Aware Security measure coupled with Brahms Model and Warning Alert Generation System to deal with the human error detection and rectification in the customer service department. Context Aware Security depends on rule-based system which rectifies the decision making process in terms of sending alerts to empoyees in an institution. Context Aware Security receives input from Brahms Model, filters the information using rule-based system and sending alerts to the employees in the customer service department with the help of warning alert generation system. Once employees receives timely alerts based on human error in the system, strength and performance of an organisation may be improved. There are two factors which may need to be improved in terms of future work, one is ethnography study and other is to update rule-based system over the period of time. Context Aware Security measure receives an input from Brahms Model and sends ouput to the Warning Alert Generation System at the time of human error in the system. Therefore, output of our Context Aware Security system greatly depends upon the Brahms Model which is based on ethnography study. Once, ethnography study and rule-based system are improved, our proposed model would be more effective to deal with the human error detection and rectification in the system.

Acknowledgements The current research is partially funded by the Malaysian Ministry of Education ERGS 11-010-0010 and partly supported by Umm Al-Qura University, Makkah, Kingdom of Saudi Arabia. We are thankful to Dr. Mohamad Fauzan Noordin and Dr. Asadullah Shah for their useful comments.

References 1. Abid, G.,Mohamed, R.W., Mohamad Fauzan, B.N.,Asadullah, S. :A Framework to Improve Customer Service Using Brahms Model, Int. J. of Engineering Innovation and Research-IJEIR, (Jan. 31, 2015) 2. Abid, G.,Mohamed, R.W., Mohamad Fauzan, B.N., Asadullah, S. : Evaluation of Tools and Techniques for the Generation of Warning Alerts: A Survey Paper, Accepted paper in International Journal on Information Technology (IREIT, Feb. 4, 2015) 3. Robbins, S. P., Judge, T. A.:Organizational Behavior 15th Edition. prentice Hall. (2012) 4. Cristian, B., Lorrie, F.C., Julie, S.D., Saranga, K.: Bridging the Gap in Computer Security Warnings - A Mental Model Approach, IEEE (March-April 2011) 5. Maarten, S., William, J. C. : Modeling and Simulating Work Practice: A Method for Work Systems Design, IEEE (2002)

53

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 6. Maarten, S., William, J.C., Ron, V.H.: Brahms: A multiagent modelling environment for simulating work processes and practices, Int. J. of Simulation and Process Modelling, Vol. X, No. Y (2007) 7. L., JEAN, C.: Mental Models of Privacy and Security, IEEE Tech & Society Magazine (FALL 2009) 8. Lorrie, F.C., Carnegie Mellon University: A Framework for Reasoning About The Human in the Loop (2008) 9. Jim, B., Jean, C., Vaibhav G. : Targeted Risk Communication for computer security, IUI’11, February 13–16, 2011, California, USA (2011). 10. Jim, B., L. Jean, C. : Implementing Mental Models, IEEE Computer Society (2012). 11. Geong, S.P., Nik, N.A., Muhammad, R.Z., Mohamed, R.W.: Reasoning of Collaborative Human Behaviour in SecurityCriticial Work Practices: A Framework, Atlantis Press (2012). 12. Chris, J. : Visualizing the Relationship between Human Error and Organizational Failure, Department of Computing Science, University of Glasgow, Glasgow (1999). 13. Andy, P., Drew, H. : Human Error Seen in Nigeria Air Crash, The Wall Street Journal (February 11, 2013). 14. Scott, S., Doug, W. : HFACS analysis of military and civilian aviation accidents: A North American comparison, Proceedings of the Annual Meeting of the International Society of Air Safety Investigators (2004). 15. Sandeep, P.: Design research in the technology of information systems: Truth or dare., http://iris.nyit.edu/~kkhoo/Spring2008/Topics/DS/000DesignSc_TechISResearch-2002.pdf (2002) 16. Alan, R.H.r, Salvatore, T.M, Jinsoo, P., Sudha, R.: Design science in information systems research., http://em.wtu.edu.cn/mis/jxkz/sjkx.pdf (2004) 17. Maarten, S.: Modeling and simulating work practice : BRAHMS: a multiagent modeling and simulation language for work system analysis and design, Source Type: Book, Pages: 318, ISBN No. ISBN 90-6464-849-2, Faculty of Social and Behavioural Sciences, The institutional repository of the University of Amsterdam (2001). 18. Brigitte, J.: Ethnographic Workplace Studies and Computer Supported Cooperative Work (June 1994) 19. Brigitte J. : Transforming Ethnography-Reinventing Research (1996) 20. Almutairi, S., Aldabbas, H., Abu-Samaha, A.: Review on the security related issues in context aware system. International Journal of Wireless & Mobile Networks (IJWMN), 4(3), 195-204 (2012) 21. Chapman, M.: Intelligent, Automated Decision-Making in Cyber Defence (2013) 22. Gross, Kevin, C.D., Charles, J., L., Eugene, H.: Event-driven rule-based messaging system, Patent No. 5283856, Publication number US5802253 A (Sep 1, 1998) 23. Ghaffar, A., Wahiddin, M. R., Shaikh, A.: Computer Assisted Alerts Using Mental Model Approach for Customer Service Improvement. Journal of Software Engineering and Applications, 6, 21 (May 2013) 24. Noordin, M.F. : ICT and Islam, IIUM Press (2013)

54

ISBN: 978-969-8680-32-9


Onboard GSM Triangulation Tracking Device Imran Jattala1, Junaid Farooqi1, Shakeel Durani1, Mustafa Minhas1, Nauman Jattala2, Junaid Jattala2, Imran Ghafoor3, Muhammad Tahir Ch3 1

Isra University, Islamabad, Pakistan Riphah International University, Islamabad, Pakistan 3 National University of Sciences & Technology (NUST), Islamabad, Pakistan {imran.jattala,junaid.farooqi,shakeel.durranii,mustafaminhas}@gmail.com, {numan_rasool99,engr.junaidsarwar}@yahoo.com, [email protected], [email protected] 2

Abstract GSM triangulation is a method to compute mobile station (MS) location based on its relative distance from three known base transceiver stations (BTSs). GSM triangulation provides an alternate positioning solution to GPS receiver. The paper presents a novel idea of onboard GSM triangulation (OGT) tracking device. OGT computes the location of MS within the device, independent of a GPS receiver. Onboard triangulation requires the live GSM network data in-addition to BTSs location database to be stored within the device. Sierra Wireless Q2687 based GSM only positioning-solution has small form factor, is a portable device and has efficient power consumption. TinyAVR microcontroller has been introduced in OGT tracking device as a power-supervisor for an improved power management. OGT tracking device can provide a backup time of up to 1 month. All communication by OGT tracking device to and from the tracking server is encrypted. Special emphasis has been laid on the information assurance & cybersecurity for the tracking server. The paper presents the Systems Development Life Cycle (SDLC) of the onboard GSM triangulation tracking device. The applications for OGT tracking device can be positioning for Device to Device (D2D) communications and telemetry solutions. Keywords: Onboard GSM Triangulation; Sierra Wireless Q2687; Timing Advance; Matlab Mapping Toolbox.

Introduction GSM triangulation is the process of computing the position of a GSM Mobile Station (MS) with the help of three known positions, i.e. the base transceiver stations BTSs to which MS is connected [1]. GSM triangulation is an alternative to GPS for geographic position calculation. The GSM triangulation techniques are [2]: • Timing Advance (TA) • Angle of Arrival (AoA) • Round-Trip-Time-of-Flight (RToF) • Time Difference of Arrival (TDoA) • Enhanced Observed Time Difference (EOTD) Timing Advance based technique, can be used for triangulation computation, without adding extra hardware to the existing GSM Network [3]. Onboard GSM triangulation (OGT) is an innovative approach as it computes the location of MS within the device (onboard), independently of a GPS receiver. OGT algorithm requires live GSM network data and on-chip BTSs latitude/longitude coordinates database. GSM network data provides home CELL-ID of MS and six neighbor CELL-IDs. GSM network also provides TAs for the GSM home CELL and six neighbors CELL [4]. The value of TA varies from 0 ~ 65 and provides the approximate distance from the BTS. The unit value of TA 3.69 µs corresponds to 550 meters [5]. The BTSs location database is maintained and updated by GSM service providers and can be obtained from GSM service providers under a license/NDA. The Onboard GSM Triangulation (OGT) Tracking Device has been developed to fulfill the following purposes: • To have an alternative positioning solution to GPS • To have a small form factor tracking device • To have a power-efficient tracking device • To develop a GSM-only positioning solution

55

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Related Work A survey of GSM localization techniques and algorithms including CellID, AOA, TDOA, OTDOA, UTDOA, RSS, GPS, and A-GPS is presented by A. Roxin, J. Gaber, M. Wack, and A. Nait-Sidi-Moh [6]. E-OTD technique takes 5 seconds to locate a mobile with the accuracy is about 30 – 50 meters but requires external hardware for positioning computation. Each of the presented methods has its own merits and demerits. The algorithms presented enable very accurate positioning, but the draw-backs or challenges of these algorithms are in the practical implementation of such systems, because these algorithms require installation of dedicated hardware for GSM positioning. Triangulation of mobile device using a Timing Advance (TA) method has been presented by S. A. Hussain et al. [7]. The received signal is measured along this grid and recorded in the signature database. When a mobile device moves in the same area, the signal received from it is compared with the entry in the database, and thus its location is determined. They have shown that TA based approach provides better results in terms of accuracy as compared to signature database approach, which requires that the covered area be continuously monitored and the signature database be continuously updated. RSSI-based GSM localization system for rural and urban areas has been presented by M. Ibrahim, and M. Youssef [8]. They have proposed a system using a hybrid technique that combines probabilistic and deterministic estimations to achieve both high accuracy and low computational overhead. The survey of the related work showed many implementations of the GSM Triangulation or Positioning techniques but all of these require addition of external hardware to a GSM Network. Thus it was observed that there is a requirement of an Onboard Triangulation or Device-only positioning solution that can serve as an alternate to GPS.

On-Board GSM Triangulation (OGT) Tracking Device OGT Algorithm has been developed utilizing the Timing Advance based technique, since this technique doesn’t require any hardware modification in the existing GSM network [6]. An on-chip database of BTSs coordinates is maintained to enable the onboard GSM triangulation. A minimum of three BTSs connected to MS and with known distances from MS, are triangulated for its location as shown in Fig. 1.

Fig.1. Mobile Phone Positioning through GSM Triangulation

The Systems Development Life Cycle (SDLC) of the project was divided into four major components Onboard GSM Triangulation (OGT) Algorithm, Hardware design of the Device, Power Management, and Information Assurance & Cyber-Security. OGT tracking device computes its location by OGT triangulation algorithm and transmits the computed location via SMS/TCP to a tracking server which plots the required info on a GIS map, as shown in Fig. 2.

56

ISBN: 978-969-8680-32-9


Fig. 2. GPS-Independent Tracking System Architecture

Onboard GSM Triangulation Algorithm The algorithm requires at-least three connected BTSs positions and if the modem is connected to less than three, new position cannot be triangulated. This inherent limitation in the algorithm can produce performance degradation in rural areas were GSM BTSs density is less than urban areas. The triangulation computation is triggered by three connected CELL-IDs and their respective TAs. A circle intersection algorithm is used to get intersection points between two circles [7]. The process is repeated to get three pairs of intersection points. The inner intersection points are selected based on proximity to opposite circle center. Inner intersection points form a triangle, and the device position is within this triangle. The centroid of the triangle is derived as in (1), where x, y is the coordinates of centroid of the triangle. x 1 , x 2 , x 3 and y 1 , y 2 , y 3 are vertices of the triangle. 1 1 (x, y) = (𝑥1 + 𝑥2 + 𝑥3 ), (𝑦1 + 𝑦2 + 𝑦3 ) (1) 3

3

MS provides home CELL-ID and six neighbors CELL-IDs along with their respective TAs. The triangulation algorithm is built on this information.

OGT Algorithm • • • • • • • • • •

Get GSM Network Neighbors Information Extract CELL-IDs and Timing Advance Info for Home CELL and Six Neighbor CELLs Search CELL-ID in the BTS Database for Latitude & Longitude 1 Unit of TA Corresponds to 550 Meters. TA has a range of 1 to 65 (upto 35 KMs) Pick Minimum 3 CELL-IDs [Lat, Long] with Corresponding TAs Find the Circle Intersection Points for All three Circular Intersections Pick the Inner Intersection Points from three intersection points pair From one circle’s Center, pick nearest intersection point of other two circles Compute the Triangle Centroid Resultant Point is the Triangulated Location

Flow Chart of OGT Algorithm The OGT algorithm was developed and modeled in Matlab. The OGT algorithm was than coded in ‘C’ for Sierra Wireless Q2687 module. The corresponding flow chart of OGT algorithm is shown in Fig. 3.

57

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 START-UP

Read GSM Network Neighbors CELLs Information

Valid CELL Information

No

Yes

Extract CELL-IDs & Timing Advance For Home Cell & Six Neighbor Cells

Min 3 CELL IDs & TAs

No

Yes

Extract CELL-IDs Lat/Long from On-Chip Database

Compute CELL-IDs Circle Radius from TAs

Compute CELL-Circles Intersection Points Compute Inner Intersection Points Average the Inner Intersection Points Triangulated Point END

Fig. 3. Flowchart of Triangulation Algorithm

OGT Device Development Sierra Wireless Q2687 module with a built-in ARM9 application processor was selected for OGT tracking device development. Moreover, OGT tracking device required the appropriate selection of its peripherals. The battery selected for the device was a Li-Poly, with a capacity of 1600mAh [11]. A durable slide-in SIM socket was selected from Molex [12]. The socket is designed for the automotive industry which ensures that it meets the strict requirements of the industry. The embedded GSM antenna selected was from Toglas that provided the needed GSM coverage even when the device was placed under the vehicle; this also reduced the footprint of the final device [13]. A USB charging IC was selected from Texas Instruments [14]. Prototype Development The development kit for Sierra Wireless Q2687 was initially used for functionality testing and firmware development [15]. The internal power management module of Q2687 was tested and this enabled the device to have a working range from 18 hours right up to 2 days depending on the update rate, with the Q2687 enabled. Sierra Wireless Q2687 development kit as shown in Fig. 4.

Fig.4. Q2687 Development Kit

58

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Onboard GSM triangulation tracking required further improvement in its power management in-order to increase its power backup time. TinyAVR microcontroller was selected as power-supervisor. The development of the firmware was done with the help of evaluation kit from Sierra Wireless Q2687 and interfaced to a starter kit for ATtiny5 and a complete lab device was completed and tested. The prototype OGT tracking device is shown in Fig. 5.

Fig. 5. OGT Prototype Tracking Device

A 3.7V 1600 mAh Li-Poly Battery is used to power-up the OGT Tracking Device. The OGT Board has a provision for either SMA GSM Antenna or a Patch Antenna to be attached with it, as shown in Fig. 6.

Fig. 6. OGT Tracking Device Board

Onboard GSM triangulation tracking required further improvement in its power management in-order to increase its power backup time. TinyAVR microcontroller was selected as power-supervisor. The development of the firmware was done with the help of evaluation kit from Sierra Wireless Q2687 and interfaced to a starter kit for ATtiny5 and a complete lab device was completed and tested. Power Management In order to improve up-time of OGT tracking device and go beyond the 8 days bench mark an ultra-low power microcontroller was introduced into the device. TinyAVR (ATtiny5), microcontroller has multiple modes of operation with nominal power consumption; Idle Mode 25uA and in Power-down Mode less than 0.1uA at 1.8V [16]. The introduction of a microcontroller into the device enabled the device to put the GSM modem in power down mode. The microcontroller acts as the timer circuit for the device and wakes up the GSM modem depending on the update rate. This approach increased the running time of the device up to 30 days with 24 updates per day, with a battery of 4800 mAh. Finally a commercial-off-the-self waterproof casing was selected for the OGT tracking device. The selected casing has strong permanent magnets at its base, in-order to attach the OGT tracking device anywhere on the vehicle body. USB charging solution has been incorporated in the OGT tracking device for ease of use. This device was tested extensively and further optimizations were made where needed. The final layout of the board of OGT tracking device is shown in Fig. 7.

Fig.7. OGT Tracking Device – Board Design

59

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Memory Requirements for OGT Onboard GSM Triangulation requires the storage of Lat/Long database of BTSs in the program memory i.e. RAM. Therefore OGT requires more RAM as compared to conventional tracking or positioning systems. Mobilink, the leading GSM service provider in Pakistan, has installed over 8,500 cell sites (BTSs) in whole of Pakistan [17]. For a single Lat/Long of a BTS, two float numbers are required i.e. 4 bytes for Lat plus 4 bytes for Long (since float is of 4 bytes in ‘C’). Therefore for storage of 8,500 BTSs Lat/Long values 68 KBs of RAM is required. Q2687 has 2MB of RAM, therefore the OGT algorithm easily accommodated Lat/Long database of BTSs of complete Pakistan region. The Q2687 was tested with dummy floating point values of 8,500 Lat/Long database BTSs and the code was found to be stable.

Information Assurance and Cybersecurity The covert device was hardened by turning off all non-required peripherals and the provision of the Over the Air Programming (OTAP) was disabled. The crypto certificates are feed into the device via a separate Certification Authority (CA) under a cryptographic system administrator, while the system is not online. The device sends and receives data to and from control center via TCP socket and SMS. Both TCP and SMS channels are secured in order to ensure cyber-security. The primary communication channel for OGT tracking device is GPRS with SMS as backup communication channel. The OGT tracking device connects to the remote sever on a TCP Secure Socket Layer (SSL). The device computes and transmits it location through a TCP socket-write operation. In case of a socket-write failure, location data is transmitted via encrypted SMS. TCP/IP Socket Communication The system’s main security feature is based on OPENSSL 1.0.1 [18]. This is an open source toolkit maintained by the larger crypto community that provides regular bug fixes & patches, and continuously update against reservations raised by the crypto community. Every unit is authenticated and all data is secured via a host of crypto algorithm available in OPENSSL suite. The key exchange is done via RSA present in the suite. The unit comes with a tamper proof button that first sends an alert to the Control Center and then deletes all information within the memory of the device. The device also deletes mission critical information prior to the depletion of the battery charge. SMS Communication The OGT device uses SMS (Short Message Service) as backup rudimentary form of communication. Device sends the required Lat/Long Information via SMS to the control center in case the primary form of communication is not available. The Advance Encryption Standard (AES) is used to encrypt the data [19]. OGT Tracker uses AES-256 for SMS data encryption. The 32 byte AES key can be exchanged on a secure TCP socket. The encryption algorithm produces non-printable characters, which cannot be transmitted via SMS. In-order to transmit location data over the SMS a 7-bit GSM encoding scheme is used [20].

Triangulation Tracker Trails A testing scenario was designed for OGT tracking device whereby the results generated by the device were plotted on the GIS map using Mapping Toolbox of Matlab [21]. The free available shape files of Pakistan were plotted using Mapping Toolbox [22]. The trials for OGT Tracker were conducted on predefined paths in Islamabad, Pakistan [23]. The GIS data of Pakistan Population Density was plotted in Matlab as shown in Fig. 8. The GSM tele-density map of Pakistan conforms to the Population Density map. The BTSs density of GSM service providers is dictated by the population density of that area.

60

ISBN: 978-969-8680-32-9


Fig.8. Population Density Map of Pakistan

The GPS coordinates for BTSs around the testing routes were collected by physically visiting these BTSs in Islamabad. OGT results were overlaid on this BTSs Map of Islamabad [24]. Thus the mapping function developed in Matlab served as a virtual testing field for OGT tracking device. The triangulated positions computed by OGT tracking device were logged and later-on plotted in Matlab on GIS map of Islamabad. The results showed that the algorithm was efficient and the triangulated points are a good approximation of GPS track path. The actual track plot with a GPS device is shown in red and the results derived from our device is shown in blue as shown in Fig. 9.

Fig. 9. OGT Tracking Device Testing Results on Urban Path

The results showed that the algorithm was efficient and the triangulated points are a good approximation of GPS track path, however there is a performance difference between urban area and rural area triangulated path. The trial of the OGT tracking device was later conducted on a rural path (with lesser BTS density). The resultant track plot is shown in Fig. 10.

61

ISBN: 978-969-8680-32-9


Fig.10. OGT Tracking Device Testing Results on Rural Path on Rural Path

Conclusion The development of the OGT tracking device produced good results and provided the platform to recursively improve the results of the OGT algorithm and Device. The device gave an excellent backup to GPS in GSM coverage areas and high backup time of up-to 30 days. The algorithm presented produced acceptable results in low GSM density areas. The algorithm was improved with provision of error handling and preemptive point tracking to produce satisfactory results in large cities. The second development cycle has certainly produced an improved device that performs better even in low BTS density areas. Future enhancement to the device would be the integration of an LTE/WiFi module to allow for D2D communication, and development of better solutions. The introduction of ad-hoc networking will allow for more interactive scenarios.

References 1. V. G. Ashok, and A. Srinivas. "A novel two-stage self correcting GPS-free localization algorithm for GSM mobiles." IEEE 5th International Conference on Advanced Networks and Telecommunication Systems (ANTS), 2011, 2011. 2. M.N. Borenovic, M.I. Simic, and A.M. Neskovic, Petrovic, M.M., "Enhanced Cell-ID + TA GSM Positioning Technique," Computer as a Tool, 2005. EUROCON 2005.The International Conference on , vol.2, no., pp.1176,1179, 21-24 Nov. 2005. 3. M.A. Spirito, and A.G. Mattioli, "Preliminary experimental results of a GSM mobile phones positioning system based on timing advance," IEEE VTS 50th Vehicular Technology Conference, 1999. VTC 1999 - Fall., vol.4, pp.2072, 1999. 4. R.S. Campos, and L. Lovisolo, "Location methods for legacy GSM handsets using coverage prediction," IEEE 9th Workshop on Signal Processing Advances in Wireless Communications, SPAWC 2008, pp.21-25, 6-9 Jul. 2008. 5. I. Martin-Escalona, F. Barcelo-Arroyo, and A. de la Fuente, "On the Availability of GNSS and Terrestrial Location Techniques: A Field Study," IEEE Vehicular Technology Conference, 2008. VTC Spring 2008, pp.2760-2763, 11-14 May. 2008. 6. A. Roxin, J. Gaber, M. Wack, and A. Nait-Sidi-Moh, "Survey of wireless geolocation techniques," IEEE Globecom Workshops, 2007. pp. 1-9, 2007. 7. S. A. Hussain, M. Emran, M. Salman, U. Shakeel, M. Naeem, S. Ahmed, and M. Azeem, 'Positioning a mobile subscriber in a cellular network system based on signal strength". IAENG International Journal of Computer Science, vol.34, no.2, 2007. 8. M. Ibrahim, M. Youssef, "CellSense: An Accurate Energy-Efficient GSM Positioning System," IEEE Transactions on Vehicular Technology, vol.61, no.1, pp.286-296, Jan. 2012. 9. F. Meneses, and A. Moreira, "Using GSM CellID Positioning for Place Discovering," Pervasive Health Conference and Workshops, 2006, pp.1-8, 29 Nov. - 1 Dec. 2006. 10. F. Nourani, M.A.J. Jamali, "Improved Circles Intersection Algorithm for Localization in Wireless Sensor Networks," 11th ACIS International Conference on Software Engineering Artificial Intelligence Networking and Parallel/Distributed Computing (SNPD), 2010, pp.129-133, 9-11 Jun. 2010.

62

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 11. M. Einhorn, F.V. Conte, C. Kral, and J. Fleig, "Comparison, Selection, and Parameterization of Electrical Battery Models for Automotive Applications," IEEE Transactions on Power Electronics, vol.28, no.3, pp.1429-1437, Mar. 2013. 12. M. Rahnema, "Overview of the GSM system and protocol architecture," IEEE Communications Magazine, vol.31, no.4, pp.92-100, Apr. 1993. 13. Tran Thi Thuc Linh, and Dang Hoai Bac, "Design a decoupled receiver with one dual - polarized antenna to cancel CCI for GSM network," 15th International Conference on Advanced Communication Technology (ICACT), 2013, pp.251255, 27-30 Jan. 2013. 14. Lan Dai, Ke-Qing Ning, and Yan-Feng Jiang, "Design of a multi-mode and high-stability linear charger for Lithium-ion batteries," 10th IEEE International Conference on Solid-State and Integrated Circuit Technology (ICSICT), 2010, pp.345-347, 1-4 Nov. 2010. 15. J. Karim, W.M.A. Bin Wan Amat, and A.H.A. Razak, "Car Ignition System via Mobile Phone," International Conference on Future Computer and Communication, 2009. ICFCC 2009., pp.474-476, 3-5 Apr. 2009. 16. E. Jafer, B. O'Flynn, C. O'Mathuna, and Wensi Wang, "Design of miniaturized wireless sensor mote and actuator for building monitoring and control," 2010 IEEE 17th International Conference on Telecommunications (ICT), pp.887892, 4-7 Apr. 2010. 17. Mobilink, “Mobilink, Cell Sites,” 15-Nov-2013. [Online]. Available: http://www.mobilinkgsm.com/about-us/ourbusiness-financial-results-network-advantage/. [Accessed: 15-Nov-2013]. 18. M. Ordean, and M. Giurgiu, "Implementation of a security layer for the SSL/TLS protocol," 9th International Symposium on Electronics and Telecommunications (ISETC), 2010, pp.209-212, 11-12 Nov. 2010. 19. D. Lisonek, and M. Drahansky, "SMS Encryption for Mobile Communication," International Conference on Security Technology, 2008. SECTECH '08., pp.198-201, 13-15 Dec. 2008. 20. J. Brown, B. Shipman, Vetter, and Ron, "SMS: The Short Message Service," Computer, vol.40, no.12, pp.106-110, Dec. 2007. 21. Mubushar Hussain, Mudassar Hassan Arsalan, Kashif Siddiqi, B. Naseem, and U. Rabab, "Emerging geo-information technologies (GIT) for natural disaster management in Pakistan: an overview," Proceedings of 2nd International Conference on Recent Advances in Space Technologies, 2005. RAST 2005., pp.487-493, 9-11 Jun. 2005. 22. M.F. Mahbub, S. Jawad, and S. Ahmed, "Geographical information systems and digital cartography in environment planning and development of Islamabad," Proceedings of the IEEE Symposium on Emerging Technologies, 2005., pp.370-375, 17-18 Sep. 2005. 23. Xiaomeng Li, Yonghua Sun, Xiaojuan Li, Huili Gong, and Yue Ma, "Flood risk assessment in Pakistan at 1km grid scale," International Conference on Multimedia Technology (ICMT), 2011, pp.321-325, 26-28 Jul. 2011. 24. K.A. Khan, G. Akhter, Z. Ahmad, and M. Rashid, "Development of a projection independent multi-resolution imagery tiles architecture for compiling an image database of Pakistan," 2nd International Conference on Advances in Space Technologies, 2008. ICAST 2008., pp.164-170, 29-30 Nov. 2008.

63

ISBN: 978-969-8680-32-9


Cyber-Secure Energy Measurement System for Smart Grid Imran Jattala1, Junaid Farooqi1, Shakeel Durani1, Mustafa Minhas1, Nauman Jattala2, Junaid Jattala2, Imran Ghafoor3, Muhammad Tahir Ch3 1

Isra University, Islamabad, Pakistan Riphah International University, Islamabad, Pakistan 3 National University of Sciences & Technology (NUST), Islamabad, Pakistan {imran.jattala,junaid.farooqi,shakeel.durranii,mustafaminhas}@gmail.com, {numan_rasool99,engr.junaidsarwar}@yahoo.com, [email protected], [email protected] 2

Abstract Energy theft is a daunting issue in Pakistan. Pakistan has an aged electric energy transmission & distribution system which causes more than 16% line losses to the national grid. Energy theft adds greatly to line losses due to the existing power-shortage in the country. As Pakistan is energy-starved country the paper presents a robust and cost-effective solution for prevention of energy theft and adds a new paradigm to smart grid & smartmetering in third-world countries. The Energy Measurement System (EMS), presented in this paper is based on M2M processor, from Sierra Wireless. EMS consists of two parts, an Energy Measuring Module (EMM) and an Energy Data Server (EDS). EMM has AirPrime Embedded Module as its central processor and is a compact module which can be placed inside a conventional energy meter. The communication from EMM to EDS is via GSM Network. The primary communication channel is SSL secured TCP Socket and secondary communication channel is AES encrypted SMS. The EDS consists of Energy measurement software & web-portal (developed in C#.net & ASP.net) with a SQL server database. Multiple EMM clients, each with a unique serial number, can connect to EDS and update energy consumption data. The EDS can generate daily, weekly or monthly energy consumption reports of a particular EMM/house. Thus EMS provides a low-cost and comprehensive energy theft prevention system. The paper presents complete Systems Development Life Cycle (SDLC) elaborating from an idea to deployment of the system. Keywords: Smart Grid; M2M; AirPrime Q2687; SSL; AES.

Introduction Pakistan like most third world countries has an aged electricity transmission and distribution system. An aged and inefficient system causes heavy line losses of more than 16% to national grid and causes a heavy blow to the national exchequer [1]. The power generated through different sources like hydral, thermal, nuclear, wind etc. is brought to cities and villages through a transmission system and distributed to homes and industries. The lack of a monitoring system in the transmission lines results in inability of power distribution companies to effectively monitor and control the line losses. Moreover the power-failures in the transmission system are dependent on human reporting rather than automatic reporting [2]. The automation of electricity transmission and distribution system is the need of the hour to meet the growing energy demands and manage an aging transmission system. The solution to transmission and distribution system is Smart Grid [3]. Smart Grid is essential to fulfill the national & global goals of economic development, energy security, and climate change mitigation [4]. Automatic energy measurement systems and Smart Grids are the future of humanity to ensure a sustained electric energy supply. The paradigm of electric cars is going to replace oil with electricity as a fuel in net 20 years [5]. The electric power generated through renewable energy resources fed to electric cars will materialize the dream of a “Green” environment. This futuristic “Green-World” will be regulated by Smart Grid [6]. The existing energy transmission and distribution system can be gradually automated to shift towards a Smart Grid. The paper presents a system to automate existing transmission and distribution system. The advent of ‘Stuxnet’ has made Cyber-Security a necessity of every embedded system [7]. The Energy Measurement System has been secured from cyber-attacks to ensure a reliable service. A Cyber-Secure Energy Measurement System has been developed for: • • •

Gradual migration of existing transmission and distribution system towards a Smart Grid, Monitoring and regulation of transmission and distribution losses, Theft detection and prevention in transmission and distribution system.

64

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 The architecture for Cyber-Secure Energy Measurement System is shown in Fig. 1.

Fig. 1. Energy Management System – Architecture.

Establishment of Smart Grid at national and global level is a colossal tak and cannot be achieved by the governments alone. Therefore communities, academia, industry and civil society have to join hands to bring about this revolution. Pakistan has an aged power transmission and distribution system that was costing a loss of over 25% a decade ago. Power Pakistan has two times higher percentage of transmission line losses as compared to rest of the world as shown below in the Fig. 2. 30 25 20 15 10 5 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 Pakistan

World

Fig. 2. Electric Power Transmission and Distribution Losses (%age)

This paper is divided into five parts, first related research work, second is design and development of Energy Measuring Module, third is design and development of Energy Data Server, fourth is Android App for EMS and fifth is test and trial results.

Related Work Design and Development of a GSM based Energy Meter has been elaborated by A. Jain, D. Kumar and J. Kedia [8]. Conventional method of human-based meter reading system has been proposed to be replaced with a GSM based energy meter. The interface of meter to 8051 microcontroller and transmission of energy consumption data via GSM network has been demonstrated. Tamper detection unit has also been deployed to detect power theft. Another implementation of GSM based Energy Metering has been demonstrated by R.G. Sarangle, U.P. Khot, and J. Modi [9]. A single phase electric energy meter has been interfaced and energy consumption data is transmitted to a remote data server software. A control system to control the energy consumption has also been incorporated in this system. Development of automatic meter reading system has been demonstrated by H.G. Rodney Tan [10]. A single phase kWh digital electric energy meter has been interfaced with a SIMCOM GSM modem. The energy data is

65

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 transmitted via GSM network to a remote eBilling system. The billing information of individual clients can be relayed to customers via SMS. The survey of automatic metering showed that although a number of such systems has been developed based on microcontroller and GSM modems. Smart grid requires an intelligent M2M module based system. This paper is based on a market leader intelligent M2M processor that provides flexibility to add more features in the EMS in future.

Energy Measuring Module (EMM) Energy Measuring Module (EMM) is based on M2M processor Q2687 by Sierra Wireless [11]. EMM comprises Q2687 M2M processor, a single phase digital energy meter, battery & power management module and a GPS. The design layout of EMM is shown in Fig. 3.

GPS

LiIon Battery

SIM Interface

Power Management Battery Charging & Protection

GSM Antenna

Fig. 3. Energy Measuring Module (EMM)

M2M Processor – Wavecom Q2687 Q2687 is an industrial leader in M2M market. Q2786 has a built-in ARM7, 84MHz 32-Bit Processor along with a GSM/GPRS Modem in single package. A real-time OS provides an Application Development Layer (ADL) to rapid develop and deploy an M2M or telemetry application. Q2687 M2M processor is shown in Fig. 4.

Fig. 4. M2M Processor – AirPrime Q2687

EMM Interfaces Design Q2687 enables the MAX3375E connected with the UART of GN8402 GPS and reads the RMC frame from the GPS module [12]. This RMC frame has all the information of location (Latitude and Longitude) along with the current date and GMT time. The format of the RMC frame that Q2687 receives is shown in Fig. 5.

66

ISBN: 978-969-8680-32-9


Fig. 5. GPS RMC Sentence

Q2687 first reads the validity of the received data and then retrieves date, time, latitude and the longitude. If the current retrieved location information is different from the previously saved location for more than 8 meters, an ALERT is immediately generated and sent to the server via either GPRS or SMS (if GPRS is unable to establish a connection) to notify that the electric meter is relocated from its originally installed location. Q2687 has 2 UARTs in it. Uart-1 is reserved for the configuration and the firmware downloading while the UART-2 is multiplexed to be communicated with electric meter and the GPS module. For this purpose, 2x MAX4475E are used. The “THREE-STATE” pin of both of the MAX 4475E are controlled by Q2687’s GPIO. UART-2 of Q2687 is connected to the input side of both of the MAX4475E. At the output end of one MAX4475E, GPS module, GN8402, is connected and to the output end of the other MAX3375E, electric meter’s UART is connected [13]. A GSM Patch antenna was used for prototype development that was replaced with an embedded GSM antenna in final product. The initial prototype development for EMM was done a Q2687 development kit, shown in Fig. 6.

Fig. 6. Q2687 Development Kit

Energy Meter – Processor Communication Protocol After reading the GPS location information from GPS module, Q2687 enables the MAX3375E connected with the Electric meter’s serial port and disables the other MAX3375E, i.e. connected to the GPS module [14]. The Q2687 reads the units consumed and save the information along with the time in its internal memory. Q2687 repeats this procedure every 5 minutes. After every 24 hours, the system sends the total units consumed in the past 24 hours to the server. Q2687 has an RTC in it which is used for the above mentioned time calculations [15]. A single phase digital electric energy meter was used for the development of an Energy Measuring Module as shown in the Fig. 7.

Fig. 7. Domestic Electric Energy Meter

67

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Electric Meter Tamper Detection Tamper detection unit detects power theft or physical interference with the energy meter. A hardware interrupt pin of Q2687 is connected to 6 micro-switches to detect the meter tamper. All 6 switches are connected in series and gives 1.8V to the hardware interrupt pin INT0, pin#50. The INT0 is pulled down in hardware. If ever, any of those switch opens, processor immediately sense the voltage transition on INT0 pin from HIGH to LOW and generate an internal hardware interrupt. Similarly a current reversal scenario, or partial earth grounding raises an alert for Power theft [16]. As a result, processor sends a tamper alert to the server. Generation of this Alert is irrespective of any timing and location based information. Meter Configuration Electric meter is supposed to be configured for the first time when installed on site. UART-1 of Q2687 is used for this reason. Configuration includes: • • • •

Server IP address Server port number SMS server phone number The location information of the place of installation

Custom made command set is made for the above mentioned configuration. AT+IP=xx.xx.xx.xx is used to set the IP address of the server. Device respond “OK” as a confirmation of command implementation. AT+PT=xxxx is made to set up the port number of the server. Device responds “OK” to acknowledge the command implementation. AT+PH=+923331234567 is used to set store the reporting server’s phone number. “OK” response from the device confirms the successful execution of command. AT+GETLOC is made to save the date, time and location (Latitude and Longitude) as can be read from the GPS module. This command is supposed to be run only when the unit is installed on the site. This reference location is compared with every new location to confirm that the electric meter is on the same place where that unit was installed. The saved time is used to check the 24 hours of reporting time of the electric meter unit. Whenever the meter send report to the server it saves that time taken from the GPS and overwrites it when the next report is sent, i.e. after every 24 hours. The initial time with location remains saved in the module’s internal memory for record keeping purpose. All of the above mentioned commands are password protected. There are 2 more commands made for this purpose. AT+PASS=xxxxxxxx is made to restrict the unauthorized access of all the above mentioned commands. User has to enter the password by using AT+PASS=xxxxxxxx command before using any of the above mentioned configuration commands. When the configuration is done, user has to enter another command AT+DONE to complete the configuration procedure. This AT+DONE command also disables all of the configuration commands for the system.

Energy Data Server (EDS) Energy Data Server (EDS) is designed to collect energy consumption data sent by remote EMM units over a GSM/GPRS network. EMM provides all the required data to EDS to generate energy consumption reports for a city or whole country. Individual users can get their energy consumption summary reports via a secure android app. The combination of M2M processor, GPS, cellular network, database and robust software makes it possible for energy managers and customers to track their energy consumption 24 hours a day, 7 days a week and 365 days a year. EDS is a comprehensive energy consumption data reception and analysis software, developed in C#.net powered by Microsoft SQL Server database [17]. A webserver developed in ASP.net enables higher management to remotely monitor energy consumptions. A distributed database has been designed to deploy Energy Measurement System across Pakistan [18]. Database servers’ hierarchy moves from cities, regions, provinces all the way up to country level. Thus distributed database architecture allows the regional, and national level monitoring levels for energy managers. All filed EMM units connect to EDS via secure GSM/GPRS channel and transmit energy consumption data at a pre-programmed interval. The received energy consumption data and tamper alerts are logged into a distributed database as shown in the Energy Measurement System architecture in Fig. 8.

68

ISBN: 978-969-8680-32-9


Fig. 8. Distributed Database for EMS

EMS Server Software EMS software is a C#.net based application to monitor energy consumption data received via GSM/GPRS network from remotely installed EMM units. The software is powered by ArcGIS for mapping solution. The software can provide energy consumption details for individual customer, city, region or whole country. Different reports ranging from average consumption to energy theft trends can be generated via this software. EMS Webportal The EDS also hosts a Microsoft IIS web-server to provide remote access to energy managers [18]. The web-portal provides all the functionality of EMS desktop software remotely. The web-portal is powered-by Google Maps for GIS solution. The Webserver architecture is shown in Fig. 9.

Fig. 9. Webserver Architecture for EMS

EMS Android App An android app has also been developed for customers and energy mangers to remotely access individual or complete energy consumption data, depending upon the access level grated to the user [19]. A user can monitor daily energy consumption of his/her house. The data-server can issue a softcopy of client’s electricity bill as well. Cybersecurity Framework for EMS A secure communication protocol and framework has been designed to assure the data integrity and privacy protection of customers [20].

69

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Primary Communication Channel: The primary communication channel between EMM and EDS is Secure Socket Layer (SSL). SSL is a secure communication protocol used to secure a plain TCP socket for secure data transmission. SSL utilizes a X.509 generated certifictaes to establish a session key and securly transmit data [21]. Secondary Communication Channel: The secondary communication channel between EMM and EDS is SMS. To securly transmit energy data, it is encrypted using AES algorithm [22]. AES encryption produces non-printable chrachters that can not be transmitted via SMS, therefore a data data encoding scheme is employed to convert non-printable characters into printable. This AES encrypted & encoded data data is securly transmitted via SMS [23].

Test and Trial Results To test the energy consumption reporting performance of Energy Measuring Module (EMM), three EMMs were deployed at three different locations. The EMMs reported the energy consumption data at interval of 15 minutes. The test was conducted for 24 hours and all three EMMs successfully reported Energy Consumption Data. Different Tamper attempts were generated to check the EMM’s Tamper reporting feature. The Energy Consumption Report results of three EMMs are shown in Fig. 10.

Fig. 10. EMS – Energy Consumption Reporting

Tamper detection feature of the EMM reports a physical tampering attempt made to energy meter, or a power theft attempt. This feature was tested by generating artificial Tampers with the energy meter and these tamper alerts were successfully reported by the three EMMs to the energy data server. The results of Tamper reporting feature are shown in Fig. 11.

Fig. 11. EMS – Tamper Attempts Reporting

70

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 The Energy Consumption Data reporting and Tamper alert reporting accuracy increased over a number of iterations as shown in Fig. 12.

Fig. 12. EMS – Energy and Tamper Reporting Accuracy

Conclusion Energy measurement system has been designed to integrate/upgarde existing power transmission systems of third world with modern Smart grids. A M2M processer based EMM feeds power consumption data to an EDS. EDS is powered by a database and energy consumption trend-analysis can reveal in-efficient power consumptions. EDS can generate detailed reports. The generated energy consumption reports can be sent to a customer through email or SMS. The availability of periodic energy consumption data can be used for the identification of energy theft. EMM can provide alerts in case of tamper attempt with the energy meter. The development of this system can integrate decades old third world transmission line systems with global Smart Grid.

References 1. Electric power transmission and distribution losses. http://data.worldbank.org/indicator/EG.ELC.LOSS.ZS 2. Zhang Haiyang; Li ShanDe, "Design of adaptive line protection under smart grid," Advanced Power System Automation and Protection (APAP), 2011 International Conference on , vol.1, no., pp.599,603, 16-20 Oct. 2011. 3. Yang Mengyao; Wang Wei; Zhao Dianquan; Han Dong, "Research Progress and Development Trend of Smart Transmission Line Technologies," Power and Energy Engineering Conference (APPEEC), 2011 Asia-Pacific , vol., no., pp.1,5, 25-28 March 2011. 4. Crisostomi, E.; Tucci, M.; Raugi, M., "Methods for energy price prediction in the Smart Grid," Innovative Smart Grid Technologies (ISGT Europe), 2012 3rd IEEE PES International Conference and Exhibition on , vol., no., pp.1,7, 1417 Oct. 2012. 5. Ward, A., "The electric car: development and future of battery, hybrid and fuel-cell cars [Book Review]," IEE Review vol.48, no.3, pp.14,14, May 2002. 6. Visweswaran, B.; Kulkarni, A.R., "“Green Luxury” — technology and solutions for energy management," Microelectronics (ICM), 2009 International Conference on , vol., no., pp.138,140, 19-22 Dec. 2009. 7. Nordbø, P.E., "Cyber security in Smart Grid stations," Electricity Distribution (CIRED 2013), 22nd International Conference and Exhibition on , vol., no., pp.1,4, 10-13 June 213. 8. Jain, Abhinandan, Dilip Kumar, and Jyoti Kedia. "Design and Development of GSM based Energy Meter." International Journal of Computer Applications 47 (2012). 9. Sarangle, M. R. G., Khot, U. P., & Modi, J. (2012). GSM Based Power Meter Reading and Control System. International Journal of Advanced Research in Computer Engineering & Technology (IJARCET), 1(4), pp-273. 2012. 10. Tan, A. C., C. H. R. Lee, and V. H. Mok. "Automatic power meter reading system using GSM network." Power Engineering Conference, 2007. IPEC 2007. International. IEEE, 2007. 11. Durrani, S.; Jattala, I.; Farooqi, J.; Shakeel, N.; Murad, M., "Design and development of wireless RTU and cybersecurity framework for SCADA system," Information & Communication Technologies (ICICT), 2013 5th International Conference on , vol., no., pp.1,6, 14-15 Dec. 2013.

71

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 12. Jattala, I.; Durrani, S.; Farooqi, J.; Junjua, G.; Shafique, A.; Hussian, F.; Mahmood, H.; Ikram, N., "Secure automotive telematics system (SATS)," Digital Information Management (ICDIM), 2013 Eighth International Conference on , vol., no., pp.262,267, 10-12 Sept. 2013. 13. Hussian, F.; Durrani, S.; Farooqi, J.; Mahmood, H.; Junjua, G.; Jattala, I.; Ikram, N., "RFID & WSN based integrated maternity ward monitoring system," Multi Topic Conference (INMIC), 2013 16th International , vol., no., pp.43,48, 19-20 Dec. 2013. 14. Jattala, I.; Farooqi, J.; Bin Zakria, O.; Khan, S.A., "Ballistic Fire Control Computer (FCC) For Main Battle Tank (MBT)," Electronics, Communications and Photonics Conference (SIECPC), 2013 Saudi International , vol., no., pp.1,6, 27-30 April 2013. 15. Durrani, Shakeel, Imran Jattala, Dr Rida Ameer, and Nassar Ikram. "Secure Wireless Fax Module." World-Comp2013. 16. Amarnath, R.; Kalaivani, N.; Priyanka, V., "Prevention of power blackout and power theft using IED," Global Humanitarian Technology Conference (GHTC), 2013 IEEE , vol., no., pp.82,86, 20-23 Oct. 2013. 17. Jingyu Wang; Sumei Li; Chungui Liu; Ning Li, "Design and Implementation of Smart Equipment Management System Based on RFID," Wireless Communications, Networking and Mobile Computing (WiCOM), 2011 7th International Conference on , vol., no., pp.1,4, 23-25 Sept. 2011. 18. Long Guo; Xiao-guang Zhang; Yu Wang; Peng-fei Li; Jing-song Li, "Design of service registry for interoperable medical information system," Biomedical Engineering and Informatics (BMEI), 2012 5th International Conference on , vol., no., pp.1068,1071, 16-18 Oct. 2012. 19. Jacobsen, Rune Hylsberg; Torring, Nikolaj; Danielsen, Brian Vestergaard; Hansen, Morten Tranberg; Pedersen, Erik Bojer, "Towards an app platform for data concentrators," Innovative Smart Grid Technologies Conference (ISGT), 2014 IEEE PES , vol., no., pp.1,5, 19-22 Feb. 2014. 20. Chan, A.C.; Jianying Zhou, "On smart grid cybersecurity standardization: Issues of designing with NISTIR 7628," Communications Magazine, IEEE , vol.51, no.1, pp.58,65, January 2013. 21. Sikora, Axel, "Implementation of Standardized Secure Smart Meter Communication," Telecommunications Energy Conference 'Smart Power and Efficiency' (INTELEC), Proceedings of 2013 35th International , vol., no., pp.1,5, 13-17 Oct. 2013. 22. Gangil, G.; Narvey, R., "Advanced Security Algorithm for Power Grid," Communication Systems and Network Technologies (CSNT), 2013 International Conference on , vol., no., pp.409,417, 6-8 April 2013. 23. Brown, J.; Shipman, B.; Vetter, Ron, "SMS: The Short Message Service," Computer , vol.40, no.12, pp.106,110, Dec. 2007.

72

ISBN: 978-969-8680-32-9


Quality and Context-Aware Data Collection Architecture from CrowdSourced Data Akhlaq Ahmad1,2, Imad Afyouni3, Abdullah Murad4, Md. Abdur Rahman5, Faizan Ur Rehman4, 6, Bilal Sadiq4, Mohamed Ridza Wahiddin1 1 Kulliyah of Information and Communication Technology, International Islamic University, Kuala Lumpur Malaysia, 2College of Engineering and Islamic Architecture, Umm Al Qura University, KSA, 3Department of Computer Science, American University of the Middle East, Kuwait, 4KACST GIS Technology Innovation Center, Umm Al-Qura University, KSA, 5College of Computer and Information Systems, Umm Al-Qura University, KSA, 6Dept. of Computer Science, LIG, University of Grenoble Alpes, France,

{aajee, marahman, fsrehman}@uqu.edu.sa,[email protected], {abdullah.murad, bsadiq}@gistic.org, [email protected]

Abstract Mobile crowdsourcing offers great opportunities when considering users’ contexts with the aim of recognizing their interests, locations, history, and activities. This paper introduces a mobile data collection scheme that takes into account users’ spatio-temporal activities, as well as other contextual dimensions towards the design of quality and context-aware crowdsourcing framework. We propose a framework that offers a set of cloud-based services as an incentive to support a very large crowd in a real scenario by capturing their contexts using smartphones. The proposed framework captures data from an individual, validates its quality by different measures to define his context and provides a set of adapted services to end users. The framework allows being in touch with a subset of one’s community of interest. We leverage an approach requiring restricted spatiotemporal zoning model that applies to the scenario where millions of users need to perform some activities in a certain location and at a certain temporal period. We present our implementation details and the general impression of end users about our system by conducting a field test during the event of Hajj 2014. Keywords: Spatio-Temporal Zones, Crowdsourcing. Context-Aware Data Collection.

Introduction Crowdsourcing is a phenomenon in which a large crowd participates in a given task in order to produce a usable content for different services [1]. Crowdsourcing acts as a generic technique to solve problems distributed among the participants [2, 3]. For any particular event, the analysis of crowdsourced data generates useful semantic content in order to adapt the system’s behavior. In a crowdsourcing activity, the crowd should be well defined for helping each participant to clearly know the assigned tasks [4]. A few of the crowdsourcing applications are voting over the web, fund raising campaigns [2], citizen’s opinion about civic issues, public feedback or traffic updates in real-time. Wikipedia, AMT 1 (a crowdsourcing internet market place), E2 (Everything2, a web-based community interlinked by users) are among the best-known platforms for information sharing [5, 6]. Crowdsourcing powers computation for optimum solutions. Crowdsourcing involves a large number of participants that ensures a good quality of generated content. Smartphones with multi-sensing abilities are excellent amongst different tools for crowdsourcing. Built-in sensors within smartphones generate sensory data that is the main source of information. This helps defining users’ contexts thereby opening new dimensions of research [7, 8, 9]. For instance accelerometer sensors can provide the abrupt changes in position in three dimensions that helps to deduce that a user is, for example (not) moving, sitting down, walking, or running. GPS sensors provide longitude and latitude coordinates in order to define the user’s location. Temperature sensors can alert abrupt changes in user ambience to understand certain emergency situation. Devices with built-in sensors enable users to participate easily in a crowdsourcing environment as compared to web-based or traditional resources. Many researchers studied smartphone usage as a successful tool for crowdsourcing. [10, 11, 12, 13]. Crowdsourcing not only depends upon the task design but also on the platform selected and the adequacy of tasks with the relevant mobile devices [14]. A crowdsourcing platform should be intelligent enough to collect the most relevant data for a specific scenario. This relevancy is task- and context-depended. This also introduces a new dimension of spatio-temporal activities that should be taken into consideration to derive relevant data for a given task and a specific user context. Examples of events with a large crowd where moving users perform spatio-temporal activities may be a crowd during or before the start of office hours or football spectators during/after a match. In such events, identifying 1

Amazon Mechanical Turk

73

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 users’ contexts based on their spatio-temporal activities is a real challenge. One of the methods to capture users’ spatio-temporal activities is by video surveillance; however, light intensity, environmental variations, and orientation of human body are challenging factors [15] with regards to the quality of the captured crowd sourced data. Providing advanced crowd sourcing services require context-awareness. For example, crowd sourcing users’ context can fall within the following three categories, fully constrained context, Semi constrained context and fully unconstrained context. If the user context depends on both spatial and temporal dimensions, then it is considered as a fully constrained context; and if context is independent of both of these dimensions, the context is defined as a fully unconstrained context. In case the context depends on only one of these two dimensions, it would be categorized as a semi constrained context. This paper introduces a mobile data collection scheme that takes into account users’ spatio-temporal activities, as well as other contextual dimensions (data relevancy and quality) towards the design of a context-aware crowdsourcing framework. The data relevancy is validated by considering the user’s historic data and plays important role to predict the futuristic [16, 17]. In our work, we propose a framework that derives quality- and context-aware data by first capturing sensory data from large crowd carrying smartphones, and secondly, data explicitly generated by the users either voluntarily or in response to a particular task or survey. The captured data then enters a filtering/relevancy process to ensure quality- and context-awareness in order to suggest a set of services relevant to users’ contexts. The implemented framework includes a high-level crowdsourcing environment and a quality- and context aware user generated data manager. This framework maps the user’s context among the three proposed types of contexts and suggests a set of appropriate services for each. As a use case, we study the Hajj 2 scenario as a crowdsourcing agent through smartphone sensory data. During Hajj, the pilgrims have to go through different spatio-temporal zones 3 thereby requiring specific types of services. This paper is structured as follows. Section 2 discusses the related works. In Section 3, we present the system architecture that includes a high-level crowdsourcing environment and a quality and context-aware data collection mechanism. . Section 4 presents our crowd-sourced data classification and context modeling. Section 5 explains the implementation of the proposed architecture. In section 6 we have discussion about the major challenges and finally, we conclude and discuss our future work in Section 7.

Related Work Crowd-sourced data includes sensory data, personal data, social data, users’ feedback, e-health forms, as well as other kinds of user shared content. Depending upon the nature of the task, different researchers presented their work to meet their goals. Tamilin et. al 2012 presented a context-aware system referred to as “sensoRcivico” for smartphone users to connect them with civil administration in order to participate in public decision making process for civic issues. The implementation carried out by running a prototype connected to a central server, involves only the users whose context matches with a particular multidimensional (geographical, temporal, personal profile, current activity etc.) task. Administration can request users to collect multimedia information and participate in a questionnaire to add their opinions [18]. Emiliano M. et al 2008, presented the CenceMe application to automatically capture the sensory data from builtin sensors within the mobile devices in order to define users’ contexts (e.g., dancing at a party with friends), and then share such activities over the social media such as Facebook. The mobile devices were operating over SOS 4 and JME 5 and had limits of small amount of memory and computational resources as compared to latest smartphones [19]. Korthaus, A et al 2012, proposed a mobile crowdsourcing platform for a heterogeneous network of devices. They highlighted the quality control, privacy risks, and other technical dimensions of social data for large-scale collection; and thereby distribution of personal information in a crowdsourcing environment. The platform selects the mobile workers depending upon tasks, interests of workers, their rating (depending on response times for particular task and participation rates), network load, and free timings. The presented platform minimizes the network and the compensation costs for maximizing the tasks quality [4].

Aakar G. et al 2012, presented mClerk as a mobile crowdsourcing platform for underdeveloped regions with the understanding that mostly mobile devices are not connected to internet. This platform not only supports bidirectional SMS, but can also send small images to anyone with low-end mobile phone. This helps crowd from remote areas to participate and respond to any specific task assigned by civic administration, without any constraints of smartphone usage and Internet availability [20].

2

Fifth pillar of Islam Haram, Mina, Muzdalifah, Jamarat and Arafat 4 Symbian Operating System 5 Java Micro Edition 3

74

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Georgios et. al 2013 consider the user’s location as a primitive type of crowdsourcing. Such information helps sole the in trajectory-related issues and enables an extra dimension for general crowdsourcing tasks and data management trends. Moreover mobile crowdsourcing architectures and systems were discussed addressing location, sensing, power, performance, big data and privacy of large crowds. A number of in-house crowdsourcing prototypes were also overviewed [2]. Most of the work done so far mainly emphasizes different techniques to collect data through different crowdsourcing methodologies. Generally, these works deal with identical data. This has geared us to step in, to deal with large crowd, sourcing different types of data simultaneously. This has motivated us to have to deep our study on different aspects of such scenarios. Although much work has been done on crowdsourcing, very few have addressed the notion of context-awareness while collecting data from a large crowd. Our objective is to make data collection to be context-aware by selecting more relevant users as data sources. To achieve this, we propose an architecture to consider a variety of data sets to be sourced from a large crowd. Mainly it includes sensory data and user generated data. Sensory data covers environmental observations like temperature, humidity, etc. whereas user generated data is further divided into Social data and e-forms. Social data will cover users’ activities over social media like Facebook, Twitter etc. and e-forms include e-health, traffic update, complains etc. that can be unidirectional or bidirectional. Collectively, such datasets would help to select the relevant users’ to ensure the data quality. Data relevancy helps to define users’ contexts in a more authentic way to serve accordingly.

System Architecture High-level Crowdsourcing Environment This section presents our architecture for a high-level crowdsourcing environment, where data is collected, analyzed, and filtered in order to provide appropriate services for mobile users in real-time. Figure 1 shows the high-level crowdsourcing architecture. In this architecture, the User context manager gathers real-time context parameters from built-in sensors available within the smartphone, body sensor networks, as well as the Historic Context available through our cloud-based system. The social network data shared between the users’ social ties is also collected. Body sensor networks provide information related to the human body such as pulse rate, heartbeat, location, and many other parameters depending upon the available set of sensors within the smartphone [21]. The historical data helps in refining the user’s context by studying previously stored contextual data.

Fig. 1. A network of users equipped with smartphones providing a crowdsourcing environment.

Quality- and Context-Aware Data Collection Data quality is one of the most important aspects that need to be taken into consideration. We propose a qualityand context-aware data collection scheme that takes into account users’ spatio-temporal activities, as well as other contextual dimensions towards the design of a context-aware crowdsourcing framework. Crowd-sourced data is collected and goes through a filtering process based on the different contextual dimensions. Depending upon the task and the crowd behavior, the data manager processes, analyzes, and filters data as shown in Figure 2. We explain different primary managers as follows:

75

ISBN: 978-969-8680-32-9


Fig. 2. Architecture of Quality- and context- aware data manager.

Quality Manager: It scans input data to validate it according to a certain predefined quality level (e.g., noisy data). For example, if some inputs have values in between certain limits, it will validate it; otherwise it can be rejected or the relevance component may be alert to change data rating (e.g., from very relevant to less relevant). Environmental Manager: Mostly it uses sensory data available from the built-in sensors with in smartphones. This part takes care of environmental effects over collected data. For example, temperature sensor sends data to the context manager. Environmental manager takes under consideration all environmental conditions (temperature, humidity, etc.) that may affect the sensory data. User Profile Manager: Defining personal profile in details is quite helpful for user profile manager to support context awareness activities. It reads user profile and decides whether collected sensory and/or user-generated data is relevant to that particular user or not. Task Manager: Intakes instructions by administrators to get opinion from crowd accordingly. The task manager also considers other primitive contexts such as spatio-temporal parameters, activities and the personal profile. Updating a task can change overall behavior of the data manager. Relevancy/Rating Manager: This part decides about the relevancy according to users rating for a given service to choose authentic data sources. The rating behavior would let the system know about the integrity and relevance of users as “crowd sources” so that the trust level of collected data is higher. Moreover, collected data is rated based on the assessment given by the quality manager. Spatial-Temporal Manager: The Spatio-temporal parameters include position coordinates within a certain spatial zone along with the time for a particular event or activity; they together define the spatio-temporal context. The spatio-temporal manager captures users’ spatio-temporal dimensions in order to restrict data sources to the relevant participants located within the corresponding area of interest. Thanks to context manager, this is understood that for similar spatio-temporal parameters and for several users but with different POI and profile would be addressed in different way with different set of services. COI Manager: Within a large crowd, many users are linked to each other, and are usually gathered within a Community of Interest (COI). The COI manager takes into account relationships between members of each COI in order to validate data relevancy depending upon the nature of the task. For example, different users within a COI might have a similar reason and feedback for a particular enquiry. Based on that, any input data from a given participant can be compared to other inputs within the corresponding COI so that it can be validated or rejected. Once crowd-sourced data is filtered and passed through all these local managers, so that new refined datasets are derived. Necessary analyzes and actions are taken accordingly, so that context-aware services can be provided. Model Our approach for mobile crowdsourcing intakes real-time sensory data from smartphones with built-in multisensors, historical, as well as personal data in order to define the user’s context. This approach provides users with a set of context-aware services for their specific spatio-temporal activities. This section first presents our crowd-sourced data model, and then discusses the spatio-temporal zoning approach as well as the context modeling in order to achieve a quality- and context-aware data collection.

76

ISBN: 978-969-8680-32-9


(a) Fig. 3.

(b)

(c)

Sensory data and user-generated content are the main sources of crowd data. (a) A smartphone equipped with sensors, (b) a very large crowd equipped with smartphones, (c) user generated contents.

Crowd-sourced Data Classification In this work, we have classified crowd-sourced data into two main categories: sensory data and user-generated content (see Figure 3). Sensory data is derived from built-in sensors within mobile devices. Built-in sensors within smartphones play a vital role for crowdsourcing. Raw sensory data can be analyzed to define a useful context, which further opens new dimensions of research [7, 8, 9]. Latest smartphones generally contain a large number of built-in sensors like GPS, proximity sensor, camera, microphone, accelerometer, barometer, magnetic sensor, infrared, temperature sensor 6 etc. This set of mobile phone sensors individually and collectively can define a particular context to be translated into useful information. For example, if the accelerometer provides the abrupt changes together with high temperature values by temperature sensors, the system can deduce that the area might be under fire and user is running away. To enquire, we read the location of that spatial zone through the GPS sensor for necessary actions. Organizers may also request nearby users (by reading their GPS sensory data) to get the average temperature at that spatial zone. If the reading is similar to that detected earlier, one can deduce that spatial zone is under fire. This interpretation further requires consistency and accuracy of collected data. In our future work we will add Body Sensor Networks comprising sensors that are usually carried by elderly people or a patient under observation without visibility to others. The second type of crowd-sourced data includes: 1) e-forms, surveys, questionnaires, as well as 2) social data. E-forms are generally available within websites or mobile phone applications that can be responded by the users on demand or by using an incentive model so that they are provided with rewards of their choice. Such e-forms are useful to collect health data and road conditions, for instance, of a crowd in order to monitor the environment. Alerts can be generated accordingly for public awareness to deal with any dangerous situation. The other category of user-generated data contains a large domain of platforms over social networks such as Facebook, Twitter, Flicker, YouTube, etc. Through these frameworks, users share their ideas, images, upload audios and videos, and reviews, serving as a good tool for crowdsourcing. However, reliability and privacy of such data is still a big challenge for researchers with the aim of getting relevant knowledge out of it. User-Generated Content usually contains both unidirectional and bidirectional contents. In unidirectional contents users by themselves collect some important data and share with administrative authorities. For example if there is a road accident, people on road may take a picture and send it to the ambulance service and traffic department. Also for a parking area one may update about the parking availability so that others arrange their plan accordingly. However in bi-directional case, user-generated contents managing authorities may involve users to participate in some questionnaires or reports. In civic issues for example, civil administration invites citizens to participate in decision-making for a particular task. In such cases, users may be provided with some e-forms over their mobile phones or through web-based applications, and are requested to respond in order to produce a usable content for different services. In addition, particular areas under certain environmental conditions that is not suitable for human health can be reported to make necessary arrangements by predesigned e-Health forms. Citizens may also register complaints or reports voluntarily for a particular event. For instance, free GPS-based applications (e.g. WAZE 7) are available for users to update traffic conditions, accidents, police traps, etc. Such collected content from users in real-time helps in saving effort and time while performing some traditional or advanced services such as navigating through a certain crowded city. Spatio-temporal Zoning and Context Modeling The user’s context not only depends upon the task required to achieve a common goal, but also on other dimensions such as the spatio-temporal parameters, user profile, historical data, and the crowd-sourced data as. With the aim of integrating the different context dimensions with our data collection framework, we propose to classify the user context within three categories based on the current spatio-temporal constraints: (i) fully constrained context, (ii) semi constrained context, and (iii) fully unconstrained context. If one has to reach the 6 7

http://digitaldimensionoftechnology.com/tag/why-does-a-smartphone-need-eight-cores/ https://www.waze.com/livemap

77

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 office with a certain time constraint for example, this spatio-temporal activity will be considered as a fully constrained context, since both location and time are specified. Another example, when one has to go to a certain bank to perform some financial transaction; this case provides us with a fully spatio-temporal user’s context because the user is bound to go to a specific bank (spatial context) within bank hours (temporal context). Whereas, if one’s activity is semi-restricted to either the spatial or temporal dimension, then it will provide us with a semi-constrained context. This is explained as in case one has to fill the petrol tank within the next hour (temporal context), a semi-restricted context can be derived because the user is free to go to any petrol station (spatial zone). Finally, if an activity is not restricted on neither spatial nor temporal dimension then it will provide us with a fully unconstrained context. For example, in case someone has to buy something from a market will provide us with a fully unconstrained context because one can go to any supermarket at any time to buy the required stuff. In all these three types of context the user needs some services. For example, to visit the bank he would need a navigational map, the city live traffic analysis for time to reach the bank, the bank’s official timings, the dynamic traffic condition, an SMS reminder to reach before the closing time, etc. Next we define the crowd sourcing spatio-temporal context modeling approach. Formally a Spatio-temporal context of given user “ u ” is defined as,

= cstu c ( xu , y u , z u , t , ∆t , A, act u )  (1)

Where xu , y u , z u define the user’s location at given time “ t ”and ∆t is the time duration for which a user remains within the spatial zone “ A ”and act u is the user’s current spatio-temporal activity for example moving to certain Geo Fence within given time frame. The spatio-temporal manager (see Figure 2) is responsible to define set of coordinates ( x, y, z ) within the area “ A ”, the time “ t ”, and the time duration “ ∆t ” as well user’s current activity “ act u ”. In addition, by considering the set of Points of Interest (POI), the context is extended to C = c ( cstu , c poi ) 

( 2)

Where, c poi = c ( p1 , p2 , , pk , f1 , f 2 , , f k ) and p1 , p2 , , pk , f1 , f 2 , , f k are geographic points and friends, family members, physician, group leader etc. Relevancy manager (see Figure 2) intakes cstu and c poi in addition to the profile manager that fetches the historical and personal profile p uprof ( name, language, nationality,...) and refines the user’s context as C = c ( cstu , c poi , p uprof ) 

( 3)

The Environmental manager (see Figure 2) takes care of the environmental parameters such as, temperature, humidity, light etc. The quality of data is taken care by the quality manager that may define a certain domain for the various parameters under consideration. For example if the temperature range is between “ 200 C − 300 C ”, then a particular task should be triggered. Task manager intakes a set of instructions from managing authorities for example to take a public opinion about a civic issue; a set of questionnaire is set. This refined quality and context-aware data will provide us with the user’s context in real-time and can suggest more accurate action to be taken by offering a set of services s {s1 , s2 , s3 ,..., sn } . The Selection of relevant users among a very large crowd to request data should be based on all primitive managers (see figure 2). This filtering will help us to assign a particular task to the selected user(s) thereby avoiding unnecessary raw data collection and involving irrelevant users in order to save resources for their maximum utility. The collective action for a number of users with similar contexts within a large crowd may be integrated uniquely and for different contexts, actions would be different.

Implementation We have implemented our proposed framework using client server architecture as shown in figure 4. The client is an app running on an Android or iOS based smartphone or tablet. The app communicates with the Application Server which extracts information from the crowd in different ways. System Overview The developed prototype has been implemented to be used over a very large crowd, within the Hajj 8 event, while users perform rituals. The high level system architecture for the proposed framework is shown in Figure 4. Users are equipped with smartphones connected to our crowd sourced data collection framework as well as organizers are connected through web portals, and also vehicle tracking service provides spatio-temporal information through special devices attached to the vehicle. For smartphone users, our application serves both

78

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 iOS and Android based operating systems. Front-end communicate with Application Server through HTTP REST API. This API connects front-end to application server and send/receives parameters in JSON 9 format. The direct information by front-end , for example, complaints, audio, video, images, traffic update, sharing location, and the sensory data from smartphones and spatio-temporal data from vehicle is stored in the server continuously .. Built-in mobile sensors provide valuable information about location, ambient temperature, light and sound, accelerometer, and many others. Back-end server, which is Amazon EC2 in our case, has the ability to scale up vertically and horizontally by sharing load on multiple instances for a large crowd. Massive spatiotemporal data from large crowds is forwarded to database server which has a capacity to handle spatial, temporal and as well as spatio-temporal queries. Data is collected and aggregated in Big Data concept. It is assumed that huge volume of diversified data is collected which need to satisfy the efficiency for Online Analytical and Transactional queries.

. Fig. 4. High Level architecture

Our proposed framework is implemented as an iOS/Android Application and supports the following important features for users (see Figure 5). Out of Boundary Service: Users can check whether they are inside or outside certain spatial zone. Our proposed architecture reads user’s GPS coordinates and informs user whether he/she is inside or outside the boundary of that spatial zone. POI (Places of Interest) Service: Users can add their point of interests like their residence; add their favorite restaurant, nearby hospital or money exchange, etc. Our architecture filters opted set of POIs among the whole cloud and returns the nearest POIs with navigational maps. Favorite COI Service: Users are allowed to add/delete/update their COI (community of Interest) like friends/family members/group leader//Physician list and later in case a person is lost our application, will provide him the locations along with navigational map to reach to his COI. In case a user does not have smartphone, the GSM gateway can be used to send SMS to our system containing his approximate location to be communicated to his COI. In case a member of the pilgrim’s COI does not have a smartphone, the SMS service will be sent to him/her about the lost member of his COI. This service allows users’ to add his favorite places too for example his residence, favorite restaurant, mosque etc.

Fig. 5. Screenshot of the crowd source data management app and available set of services.

9

http://json.org/

79

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Update Traffic Service: User can provide traffic update by multimedia service that helps the system to recommend best alternative paths for other users. Any such useful information sent by users to the system is suggested to be rewarded by some credits (e.g., free SMS). E-Health Service: Through this service users can share their health issues with location, and in case of some serious problem, an alert will be communicated to the user’s COI, managing authorities and the nearest hospital for quick remedy. Emergency Services: This service makes contact with fire stations, police, civil defense, traffic police, and hospitals for any kind of emergency and shows optimum path to safe exit from an emergency situation. User is allowed to share his location via free sms service so that recuse operations can be made easy Social Network Service: We use Twitter service that allows users to tweet; search hajj and umrah related tweets and visualize Geotagged tweets with in nearby area. This keeps them aware of all useful information and their tweets can update the rest of the users. This service also includes Hajj messenger service that helps the users to share text, images, audio and video with their COI. Shortest Path Service: This service is actually used in many services. In case a user is looking for any POI or COI, the system returns the shortest path on a navigational map. If a user wants to visit other nearby places, shopping malls etc. the system provides him/her with the shortest path.

Evaluation To validate our proposed architecture, we conducted a field study during the Hajj season of 2014 for a period of 30 days. During this period, 5,000 (3848) pilgrims from 90 countries downloaded the Hajj and Umrah mobile application. Drawing upon the ISO 9241 definition of usability, three factors comprised the evaluation dimensions. These factors are effectiveness (the number of completed tasks), efficiency (the rate of success of task completion), and satisfaction (the level of comfort and acceptability of the system to its users). Usability data was extracted from 1) the system’s log files that capture users experience through buttons clicks and navigation actions, and 2) users’ feedback through online and paper surveys. Among 5,000 users, 125 users participated in filling out the surveys. Results are described per usability factor in the following sections:

Fig. 6. A Comparison. Of Services Used

80

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Effectiveness Effectiveness is defined in our context as the number of times pilgrims use the services provided by the Hajj and Umrah application Figure 6. Our systems log files showed that some services are more effectiveness than others. For examples, the “shortest path service” and “Hajj rituals” were used the most (1455 times and 648 times respectively), while the least used services were the “complaints (14 times)” and “Twitter search map (20 times)”. A summary table of the effectiveness of services is provided below. Efficiency Efficiency is defined as enabling pilgrims to perform rituals and run daily activities faster during their stay in Makkah. Around 89% of our users thought the Hajj and Umrah mobile application was very efficient in helping them to find prayers time in their locations, remember the sequence of particular rituals, and check if they are within a boundary of certain spatial zones. Satisfaction Satisfaction is defined as the user’s comfort and acceptability to using the Hajj and Umrah application. Among the 5,000 users, we have identified 202 super users (those who used the application at least 50 times) and sent them questionnaires to fill out. We have received 108 answered questionnaires. In general, users’ reviews gave us very encouraging feedback that the presented system is overall satisfactory and quite useful. However, two major issues were reported to affect our users’ experience, 1) Internet services were either not available or too slow in some areas, and 2) mobile battery was a bottleneck when using the application. Details can be found in Section 6.

Discussion Although increasing the Internet coverage or duration of battery charge might not be in the hand of a software provider, we have tried to minimize the extent to which they negatively impact the user’s experience of our Hajj and Umrah mobile application. Internet connectivity During Hajj, around 3 million pilgrims gather in small areas at the same time launching millions of Internet service requests simultaneously. The current cell phones towers and infrastructures in areas like Muzdalifah and Arafat are not good enough to handle such tremendous wireless connectivity and data traffic. Many of our users have reported a complete loss of wireless connectivity in these areas and therefore were not able to use some of our services that are Internet dependent including the reported most effective service, the “shortest path service”. Our quick and immediate response to this issue was refining 90% of the services to work offline. For example, we are now providing offline maps to crowd for browsing POIs and other shortest path services. These changes are expected to improve the usability of our application. . Limited battery power Battery power has been documented as a major limitation to the mobile software development industry. During the Hajj season, this problem magnifies even farther due to the fact that pilgrims are always in transition between tents, buses, and trains, all of which have limited access to power outlets. It was no surprise that many of our users reported issues with battery consumption and some preferred to shut down the Hajj and Umrah application in order to extend the duration of the battery charge. One way to increase the battery life is by reducing the need for Internet operations that eat up the battery faster than others. Providing offline maps described in Section 6.1 is also expected to increase the battery life, since it eliminates the need to communicate with the server over the Internet.

Conclusion and Future Work In this paper we have proposed a framework that serves as crowdsourcing platform to collect sensory- and user generated-data through smartphones carried by users. Spatio-temporal manager with-in the context manager classifies the context to be one of three types. Similarly environmental manager, quality manager, relevancy

81

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 manager, user profile manager, spatio-temporal and COI manager refines data to ensures quality and relevancy of collected data to define user’s context with accurate measures. Depending upon the user’s context, we provide the user with a set of suitable services well aware of context for that particular scenario. As a use case, we have implemented our framework as an Android/iOS application that receives sensory data from users’ smartphones during Hajj and Umrah while they perform different spatio-temporal activities. We also take the benefit from real-time social network data for translation of dynamic situations for emergency services. As a part of our research, we also take care of users’ health data that opens new areas of research. We intend to extend our research to model MERS-CoV spreading, which is one of the main concerns for future pilgrimage by selecting more relevant users for collection of health data. This would include adding multi-modal sensory devices, video monitoring, gesture-based interactions and wireless sensors network. We expect to furnish our research with more accurate and convincing results with more focus for preserving user’s privacy.

Acknowledgement We extend our gratitude to King Abdul Aziz City for Science and Technology (KACST), Kingdom of Saudi Arabia for funding this research project through NSTIP grant number 11-INF1683-10, 11-INF1700-10 and 13INF2455-10.

References 1. Estellés-Arolas, E., & González-Ladrón-de-Guevara, F. (2012). Towards an integrated crowdsourcing definition. Journal of Information science, 38(2), 189-200. 2. Chatzimilioudis, G., Zeinalipour-Yazti, D., "Crowdsourcing for Mobile Data Management," Proceedings of the IEEE 14th International Conference on Mobile Data Management (MDM), vol.2, pp.3-4, 2013. 3. Zambonelli, F., "Pervasive urban crowdsourcing: Visions and challenges." In Proceedings of the IEEE International Conference on Pervasive Computing and Communications Workshops (PERCOM Workshops), pp. 578-583, 2011. 4. Korthaus, A., Dai, W., "Crowdsourcing in Heterogeneous Networked Environments - Opportunities and Challenges", Proceedings of the 15th International Conference on Network-Based Information Systems (NBiS) pp.483, 488, 26-28, 2012. 5. Yue P.; Blevis, E., "A survey of crowdsourcing as a means of collaboration and the implications of crowdsourcing for interaction design," Proceedings of the 2011 International Conference on Collaboration Technologies and Systems (CTS),, pp.397-403, 2011. 6. Doan, A., Ramakrishnan, R., & Halevy, A. Y. (2011). Crowdsourcing systems on the world-wide web. Communications of the ACM, 54(4), 86-96. 7. Rahman, M. A., Kim, H. N., Saddik, A. E., and Gueaieb, W., “A Context-aware Multimedia Framework toward Personal Social Network Services”, Journal of Multimedia Tools and Applications, Springer, vol. 61, no. 3, pp. 1-31, 2012. 8. Ganti, R.K., Fan Y., and Hui L.,"Mobile crowd sensing: current state and future challenges", I EEE Communications Magazine, 49, no, 11 (2011): 32-39. 9. Burke, J.A., Deborah E., Mark H., Andrew P., Nithya R., Sasank R., and Mani B. S., "Participatory sensing", Center for Embedded Network Sensing (2006). 10. Kanhere, S. S., Participatory sensing: Crowdsourcing data from mobile smartphones in urban spaces. In Proceedings of the 12th IEEE International Conference on Mobile Data Management (MDM), 2011 (Vol. 2, pp. 3-6). 11. Chatzimilioudis, G.; Konstantinidis, A; Laoudias, C.; Zeinalipour-Yazti, D., "Crowdsourcing with Smartphones," Internet Computing, IEEE, vol.16, no.5, pp.36, 44, Sept.-Oct. 2012, doi: 10.1109/MIC.2012.70 12. Yang, D., Xue, G., Fang, X., & Tang, J. Crowdsourcing to smartphones: incentive mechanism design for mobile phone sensing. In Proceedings of the 18th annual international conference on Mobile computing and networking (pp. 173184). ACM-2012. 13. Rai, A., Chintalapudi, K. K., Padmanabhan, V. N., & Sen, R., Zee: zero-effort crowdsourcing for indoor localization. In Proceedings of the 18th annual international conference on Mobile computing and networking (pp. 293-304). ACM2012. 14. Vincenzo D. M., Eddy M., Stefano M., Crowdsourcing to Mobile Users: A Study of the Role of Platforms and Tasks, Proceedings of the 1st VLDB Workshop on Databases and Crowdsourcing (DB Crowd 2013), pages ..--... 15. Melfi, R., Kondra, S., Petrosino, A., “Human activity modeling by spatio temporal textural appearance”, Pattern Recognition Letters, 34(15), 1990-1994. 16. Bamis, A., Savvides, A., "Lightweight Extraction of Frequent Spatio-Temporal Activities from GPS Traces," Proceedings of the IEEE 31st Symposium on Real-Time Systems (RTSS), pp.281-291, 2010

82

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 17. Rahman, M. A., Kim, H. N., Saddik, A. E., and Gueaieb, W., “A Context-aware Multimedia Framework toward Personal Social Network Services”, Journal of Multimedia Tools and Applications, Springer, vol. 61, no. 3, pp. 1-31, 2012. 18. Tamilin, A., Iacopo C., Emmanuel S., Alfonse O., and Nicola C., Context-aware mobile crowdsourcing, Proceedings of the 14th International Conference on Ubiquitous Computing (Ubicomp 2012), pp. 717-720. 2012. 19. Emiliano M., Nicholas D. L., Kristóf F., Ronald P., Hong L., Mirco M., Shane B. E., Xiao Z., Andrew T. C., “Sensing Meets Mobile Social Networks: The Design, Implementation and Evaluation of the CenceMe Application”, SenSys’08, November 5–7, 2008, Raleigh, North Carolina, USA. ACM 978-1-59593-990-6/08/11 20. Aakar G., William T., Edward C., Ravin B., mClerk: enabling mobile crowdsourcing in developing regions, CHI 12 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, ISBN: 978-1-4503-10154 doi>10.1145/2207676.2208320. 21. Vermesan, O., and Peter F., eds. Internet of Things: Converging Technologies for Smart Environments and Integrated Ecosystems, River Publishers, 2013.

83

ISBN: 978-969-8680-32-9


Simulation of PMSG based Wind Energy System with Multi Level NPC Converter Sajid Hussain Qazi1,2, M.W Mustafa1, Irfan Ahmed Bajkani1,2 1

Faculty of Electrical Engineering, Universiti Teknologi Malaysia, Malaysia Department of Electrical Engineering Mehran U.E.T S.Z.A.B Campus Khairpur Mir’s, Pakistan [email protected]

2

Abstract The size and power ratings of commercial wind turbines have increased in the last decade. This has motivated the improvement for new generator-converter wind turbine configurations with high power density, efficiency and low maintenance. Where permanent magnet synchronous generators (PMSG) have been at the forefront as compared to the classical doubly fed induction generator (DFIG). In this paper a new WECS configuration is explored based on three-phase PMSG with multi-level NPC converter. The proposed converter configuration allows to elevate the low voltage at generator side to a medium-voltage at the grid side. Each three-phase terminal of the machine is connected to a full bridge diode rectifier, followed by a dc-dc boost converter to elevate the voltage level and control the MPPT. The grid side converter is a multi-level Neutral Point Clamped (NPC) inverter operating at medium-voltage level. The performance of the proposed configuration is then compared with conventional grid side inverter. Complete system verified by simulation using MATLAB/Simulink platforms. Keywords: PMSG, NPC, WECS, MPPT

Introduction Wind energy conversion systems (WECS) have been among the fastest growing renewable energy systems of the last decades, mainly due to technology development, cost reduction, environmental awareness and increasing cost of fossil fuels [1]. Regardless of significant progressions and stimulus to the environment, the costs of wind power is continue to be greater than the conventional low-carbon alternative such as natural gas. Therefore, in order to improve wind turbines behaviour and to make them cost-efficient and reliable than traditional power generation systems, plentiful research is to be done. According to the reference [2, 3], there are many kinds of variable speed generators used for wind energy conversion system, from them doubly fed induction generator (DFIG) is more generally used than permanent magnetic synchronous generator (PMSG) today, PMSG has some downsides. Electromagnetic field with the flexible structure is its pre-requisite, which leads to the high standard of the fabrication as well as of the procedure. Furthermore, power inverters also need to know the variable speed of the generator. Beside all, PMSG have some experts advantages. Particularly, PMSG is direct drive, has slow rotation speed, does not have rotor current, and can be used without gearbox. The high efficiency and low maintenance will reduce the cost that is the most concern to invest [4]. The search for lower maintenance and higher power rating, predominantly for the growing offshore penetration, has made the gearless direct drive with large pole number synchronous generators an attractive solution. In particular, the permanent magnet synchronous generator (PMSG), due to the cost reduction of magnets, higher power density and no need for slip rings [5]. Wind energy conversion systems (WECS) normally operate in low voltage ratings at generator and grid side (≅690V), for reaching higher power rating, two-level back-to-back voltage source converter topologies connected in parallel. In case of PMSG based WECS, since they do not consume reactive power, diode rectifiers at generator side followed by dc-dc boost stage and a grid tied two-level voltage source converter have also been adopted in industry as a cost effective and higher-power-density solution [6]. Continuing research for the development of wind power technology revealed that the performance and efficiency of wind energy conversion system also depend on the efficiency of inverter device, which is facing some tough issues to meet the new grid codes requirement. PMSG based WECS can be connected to the grid side inverter through a dc link element. The back-to-back two level VSC is widely used in variable-speed wind turbines [7]; in this configuration high voltage system is in series with active switches and the issue of voltage sharing between devices make this system insecure. There are many other topologies like, back-to- back multilevel VSCs such as three-level neutral-point-clamped converter [8, 9] and multi modular cascaded H-

84

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 bridge converter are in use [10]. As compared to two level VSCs the multilevel converters allows to use lowvoltage devices and improved power quality can be achieved. On the other hand latest researches explored the use of back-to-back CSC with medium voltage level for high-power PMSG- based wind turbines [11]. In this paper a comprehensive design and evaluation of the control system of multi-level NPC converter system for wind power system based of direct-drive variable-speed synchronous machine (gear less). Following are the salient features of proposed direct drive, variable speed, NPC-based wind power configuration when compared with wind power systems based of fixed-speed, squirrel-cage induction machine based and doubly-fed induction machine. • • • •

Significantly reduced maintenance cost due to elimination of the gearbox; Turbine-generator mechanical speed can be controlled in a prominently wider range and have high capturing of the wind energy; Due to high-poles this configuration can logically accommodate low-speed, which makes significant reduction in size, volume, and weight; and The converter system between turbine and electrical grid is like a buffer, it thus minimizes unwanted dynamic interactions between the two subsystems, because of fluctuation in wind speed and electrical faults.

Description of System Topology The proposed system is shown in fig-1. The wind turbine is directly connected to 3-∅ PMSG without gear box because of high number of poles; the output of PMSG is then connected to diode bridge rectifier because PMSG do not require any magnetizing current. The diode bridge supplies unidirectional flow of power to the grid, on the other hand it is cost effective design and simple. Second phase of converter is dc-dc boost converter which raises the voltage level that enable MPPT scheme by controlling power drawn from generator. Interleaved or multichannel boost converters can be used with PMSG depending on the power level [12]. Third part of converter is based on medium-level voltage NPC inverter in connection with power grid. Three voltage levels per phase breeding by NPC. In the proposed topology IGBT based NPC is exploited, The NPC is usually connected to 2.3, 3.3, 4.16 and 6.6 kV medium voltage grids, and can be found up to tens of megawatt [13].

Fig. 1. PMSG based Wind Turbine

PMSG Model WECS utilizing PMSG because of its advantageous such as better reliability, lower maintenance and more efficient [14]. The model of PMSG is established in the d-q synchronous frame as shown in Fig 2.

Fig. 2. Reference Frame of PMSG

85

ISBN: 978-969-8680-32-9


The equation for current of PMSG are given in equation 1 and 2, 1 = �usd + pωr Lq iq− R s isd �

did dt

diq dt

=

Ld 1

Lq

�usq − pωr (Ld id + ψm ) − R s isq �

Electromagnetic torque of PMSG is, Te = 1.5ρ[λiq + �Ld − Lq �id iq

4 2

3

Lq, Ld : Rs : iq, id :

d and q axis inductances Stator winding resistance d and q axis currents

vq, vd :

d and q axis voltages

ωr :

Angular velocity of the rotor

λ :

Amplitude of the flux induced by the permanent magnets of the rotor in the stator phases

p : Te :

Number of pole pairs Electromagnetic torque

In the proposed system 3-∅ Round rotor type PMSG is used and other parameters are shown in Table-1. Table 1. Parameters of PMSG

Rated Voltage of stator Rated frequency of stator Stator phase resistance Armature inductance d-axis inductance q-axis inductance Number of poles

5 KV 50 Hz 0.01ohm 0.03H 5.5mH 3.75mH 50

Pitch Angle Controller The control of aerodynamic system plays a significant role in regulating the mechanical power. Pitch angle controller is based on the principle which is changing the blades angle at the revolutions over the maximal generator speed as well as protecting the generator before overloading at high wind speeds [15]. When speed of wind is below the nominal value optical angle is approximately zero and when speed grows, angle increases. The value of turbine torque and the performance coefficient of turbine depend on pitch angle controller. In this controller, illustrated in Fig 3, the speed of the rotor which is the input is added with its reference value to have the output value of the pitch angle of the blades, which changes the performance coefficient of the turbine.

Fig. 3. Pitch Angle Controller of Wind Turbine

86

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Maximum Power Point Tracking (MPPT)

In the generator-side inverter, MPPT produces the ωref for the Differential filter. According to the relationship between output power (kW) and ω (rad/s) as shown in Fig 4, the coefficient of wind turbine achieves the maximum for the tip speed, when the pitch angle β=0. There is specific point to get the maximum output for every wind speed. Hence, in order to control the maximum power in every wind speed, the MPPT tracks the continuous line and optimal line in Fig 5. The tip speed ratio is kept at constant value for all maximum power points, while the relationship between the wind speed and the wind turbine generator speed is explained as follows:

Ωn = λ n

Vn R

4

Where Ω𝐧 is the optimal rotation of wind turbine generator at the wind speed Vn. The MPPT control strategy is based on monitoring the wind turbine generator output power using measurements of the wind turbine generator output voltage and current as well as directly modelling the dc/dc converter duty cycle, which is followed by the comparison of among output power values.

Fig. 4. Relation Between Power and Wind Speed

Fig. 5. Wind Generator Power Curve at Different Wind Speed

Generator Side Converter Generator side converter controller is designed to catch maximum power from available wind power. According to [16] in order to catch maximum power, the optimum value of the rotation speed is adjusted. The tip speed ratio λ is taken into account due to the equation being addressed as: ωref =

λopt .v R

5

87

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 Here, 𝛚𝐫𝐫𝐫 is the blades angular velocity reference and 𝛌𝐨𝐨𝐨 tip speed ratio optimum. In generator side controller three phase diode rectifier is connected with boost rectifier maintaining constant 1700-volt dc-link voltage to the grid side converter. Grid Side Converter The grid tied NPC inverter is controlled by traditional voltage oriented control VOC) as shown in Fig 6. In VOC, active and reactive power can be controlled by q and d components of grid currents [17]. To control dclink capacitor voltage vdc active power is used. The reactive power is set according to grid requirements. As mentioned NPC inverter is IGBT based so the switching of IGBT is done by phase decomposition followed by a IGBT and diode for each level. All three phases decomposed in same manner with neutral point of phases connected together and at the output one level (3-phases) increased voltage to be supplied to the grid via LC filter to diminish harmonics from the inverter output.

Fig. 6. NPC Controller Scheme

Simulation Results A detailed switching model of the wind power system of Fig. 1 and the controllers are developed in the MATLAB/SIMULINK environment. Based on the parameters of PMSG given in Table-1, the proposed controller was simulated in at constant wind speed of 12 m/s. In the simulation, wind turbine provided by MATLAB/SIMULINK is used to drive the system. The turbine model provides optimized speed to control system after receiving the wind speed. Turbine and electromagnetic torque of system is drawn in figure-7 and figure-8. Voltage output waveform of PMSG is displayed in figure-9.

Fig. 7. Tubine Torque

Fig. 8. Electromagnetic Torque of PMSG

Fig. 9. Voltage Output of PMSG

88

ISBN: 978-969-8680-32-9


Fig. 10. Voltahe Output of NPC Converter

Fig. 11. Curent Waveform of NPC Converter

Fig. 12. DC Link Voltage

Fig. 10 and Fig. 11 showing the voltage and current waveforms of NPC converter which is then supplied to the grid through LC filter to mitigate harmonics portion from the waveform. From waveform of dc-link voltage (Fig. 12) it is clear that at the initial condition voltage is around 500-volts but as the system runs at normal condition dc-link voltage reaches its constant value of 1700-volts. Frequency of the system is approximately same as grid frequency (50-hz), drawn in Fig. 13.

Fig. 13. Frequency of The System

Validation of Proposed System Wind turbine using PMSG is further simulated using conventional 3-arm 6-pulse grid side inverter and compared with the proposed methodology. Results shown that using NPC converter at grid side will give less THD as compared to conventional method. Figure 14 (a) showing the line to line voltage of conventional method with its total THD in Figure 14 (b), and the output of proposed system and its THD shown in Figure (a and b). Total harmonic distortion shown above for the two methods validates the proposed system 0.09% compared to 5.69% of conventional method. This shows the verification of proposed work.

Conclusion In this paper, a system is designed for IGBT based multi-level NPC connected direct driven wind energy conversion systems was proposed. The strategy was developed for controlling current of diode bridge rectifier while extracting the maximum power from wind. To ensure stable dc-link voltage generator power feedforward method is employed, and therefore, improves system efficiency and dynamics performance.

89

ISBN: 978-969-8680-32-9


(a)

(b) Fig. 14. (a) Three Phase Voltage (Conventional) (b). THD of Conventional Method

(a)

(b) Fig. 15. (a) Three Phase Voltage (Proposed) (b) THD of Proposed Method

References 1. L. Vattuone, S. Kouro, G. Estay, and B. Wu, "Open-end-winding PMSG for wind energy conversion system with dual boost NPC converter," in Industrial Technology (ICIT), 2013 IEEE International Conference on, 2013, pp. 1763-1768. 2. J. Chen, H. Wu, M. Sun, W. Jiang, L. Cai, and C. Guo, "Modeling and simulation of directly driven wind turbine with permanent magnet synchronous generator," in Innovative Smart Grid Technologies-Asia (ISGT Asia), 2012 IEEE, 2012, pp. 1-5.

90

ISBN: 978-969-8680-32-9

Regular Papers-Proceedings of IMTIC’15 3. A. Binder and T. Schneider, "Permanent magnet synchronous generators for regenerative energy conversion-a survey," in Power Electronics and Applications, 2005 European Conference on, 2005, pp. 10 pp.-P. 10. 4. C. N. Wang, X. K. Le, and W. C. Lin, "Modelling and Simulation of Autonomous Control PMSG Wind Turbine," in Computer, Consumer and Control (IS3C), 2014 International Symposium on, 2014, pp. 1144-1147. 5. J. M. Carrasco, L. G. Franquelo, J. T. Bialasiewicz, E. Galván, R. P. Guisado, M. A. Prats, et al., "Power-electronic systems for the grid integration of renewable energy sources: A survey," Industrial Electronics, IEEE Transactions on, vol. 53, pp. 1002-1016, 2006. 6. X. Xin and L. Hui, "Research on multiple boost converter based on MW-level wind energy conversion system," in Electrical Machines and Systems, 2005. ICEMS 2005. Proceedings of the Eighth International Conference on, 2005, pp. 1046-1049. 7. J. A. Baroudi, V. Dinavahi, and A. M. Knight, "A review of power converter topologies for wind generators," Renewable Energy, vol. 32, pp. 2369-2385, 2007. 8. A. Yazdani and R. Iravani, "A neutral-point clamped converter system for direct-drive variable-speed wind power unit," Energy Conversion, IEEE Transactions on, vol. 21, pp. 596-607, 2006. 9. S. Alepuz, S. Busquets-Monge, J. Bordonau, J. A. Martínez-Velasco, C. A. Silva, J. Pontt, et al., "Control strategies based on symmetrical components for grid-connected converters under voltage dips," Industrial Electronics, IEEE Transactions on, vol. 56, pp. 2162-2173, 2009. 10. C. H. Ng, M. A. Parker, L. Ran, P. J. Tavner, J. R. Bumby, and E. Spooner, "A multilevel modular converter for a large, light weight wind turbine generator," Power Electronics, IEEE Transactions on, vol. 23, pp. 1062-1074, 2008. 11. J. Dai, D. Xu, and B. Wu, "A novel control scheme for current-source-converter-based PMSG wind energy conversion systems," Power Electronics, IEEE Transactions on, vol. 24, pp. 963-972, 2009. 12. B. Wu, Y. Lang, N. Zargari, and S. Kouro, Power conversion and control of wind energy systems vol. 77: John Wiley & Sons, 2011. 13. S. Kouro, J. Rodriguez, B. Wu, S. Bernet, and M. Perez, "Powering the future of industry: High-power adjustable speed drive topologies," Industry Applications Magazine, IEEE, vol. 18, pp. 26-39, 2012. 14. A. Cultura and Z. Salameh, "Modeling and simulation of a wind turbine-generator system," in Power and Energy Society General Meeting, 2011 IEEE, 2011, pp. 1-7. 15. E. Muljadi and C. P. Butterfield, "Pitch-controlled variable-speed wind turbine generation," Industry Applications, IEEE Transactions on, vol. 37, pp. 240-246, 2001. 16. A. M. De Broe, S. Drouilhet, and V. Gevorgian, "A peak power tracker for small wind turbines in battery charging applications," Energy Conversion, IEEE Transactions on, vol. 14, pp. 1630-1635, 1999. 17. M. Malinowski, M. P. Kazmierkowski, and A. M. Trzynadlowski, "A comparative study of control techniques for PWM rectifiers in AC adjustable speed drives," Power Electronics, IEEE Transactions on, vol. 18, pp. 1390-1396, 2003.

91

ISBN: 978-969-8680-32-9

Extension to Partner Based Grid Workflow Extension to Partner Based Grid Workflow Scheduling Heuristic Scheduling Heuristic

Maryam Bashir, Jawad Ashraf

IIT, Kohat University of Science and Technology, Kohat Pakistan Maryam Bashir and Jawad Ashraf

[email protected], [email protected]

Institute of Information Technology, Kohat University of Science and Technology, Kohat, Pakistan [email protected] and [email protected] http://www.kust.edu.pk

Abstract. Optimizing workflow makespan in an advance reservation Grid environment is a challenging task. With the help of readily available information Partner based Dynamic Critical Path for Grid (PDCPG) proposed a technique in which partner jobs, jobs with at least one common child, are scheduled together to reduce the communication cost between parent and child jobs. In this paper we further extend PDCPG by introducing a new resource selection technique for the most critical partner job while considering earlier completion of its most critical child job. The notion of this technique is to increase the probability of accommodation of partner jobs on fewer resources. Another amendment is for less critical partner jobs which involves route calculation for resource mapping. The proposed strategy phenomenally outperformed other heuristics for workflow makespan, especially in a congested situation of computing and network resources. Keywords: Grid Computing, Scheduling, Scientific Workflow, Advance Reservation

1

INTRODUCTION

Introduction

Grid is an infrastructure in which user can access multiple resources, like computing, network, storage and data resources which are owned by different organizations to execute large scale scientific applications. Sometimes these scientific applications consist of multiple jobs with inter dependencies, which means output of one job is required by another job as an input. These dependencies apply the constraints on the execution order of jobs, thus a WorkFlow (WF) is formed, usually modelled as Directed Acyclic Graph (DAG). Some WFs are discussed in EMAN, GridPhyN, e-Protein, LEAD, WIEN2K, Montage [1]. A WF can be compute intensive and/or data intensive. In compute intensive WF computational needs of individual jobs are high, whereas in data intensive WF data needs are high. Furthermore, multiple resources, residing on different geographically dispersed sites, are needed to complete a WF and availability of all resources at a required time is critical. For improving the efficiency of the execution of the WF, scheduling is a key factor to look into [1]. Since a simple case

92

ISBN: 978-969-8680-32-9

2

Extension to Partner Based Grid Workflow Scheduling Heuristic

of DAG/WF scheduling problem is NP-Hard [13], heuristics are used to cope with the problem. Different WF scheduling techniques consider different metrics like improving or optimizing makespan (execution time of a WF) by minimizing the communication and computation cost [8]. Unavailability of resources at requested time may increase the computation and communication cost of jobs and thus can degrade the Quality of Services (QoS). Advance Reservation (AR) [10] is a technique used to reserve needed resources in future time to guarantee the availability of resources which helps in improvement of QoS. Further improvements can be achieved by reserving both computing and communication resources, especially in case of data intensive WFs [14]. It is also suggested in [14] that OBS is suitable for the Grids AR environment. In OBS, control packet is sent for setting up a path and for reserving bandwidth in advance. Then packets of data are sent in burst. Partner Based Dynamic Critical Path for Grids (PDCPG) [2] is proposed for the above mentioned environment with the objective to minimize the execution time of a WF. PDCPG is a partner based WF scheduling algorithm. Jobs are said to be partners if they have at least one common child. Partner jobs are conditionally scheduled on a single resource to minimize communication time with their common child job, which in turn can improve the makespan of a WF. Better results have been observed from PDCPG when partner jobs are scheduled together. Based on this observation, proposed technique, will be referred to as PDCPG2, attempts to select a resource for a job on which probability of accommodation of its partner jobs, if any, is higher while considering the earlier completion time of its most critical child job as well. If a partner job cannot be accommodated with other partners, according to a certain condition, then it would be scheduled on a resource which gives a minimum arrival time of its output file to the resource where its most critical partner job is scheduled. This technique may make the probability higher for child jobs to start earlier on partners’ resource. Thus, this local optimization can improve the makespan of a WF. The results of extensive experiments show that the proposed technique phenomenally outperformed two heuristics, PDCPG and Dynamic Critical Path for Grids (DCPG), for almost all settings of computing and network loads and communication and computation cost of WFs. The rest of the paper is organized as follows. The next section discusses the previous proposed heuristics for Grid WF scheduling. Section 3 describes Grid WF terminology and Grid resource model used in this paper and defines the problem as well. Section 4 describes the proposed approach and Section 5 discusses the results. Section 6 concludes the paper.

2 Related Work Related Work Many DAG scheduling heuristics and meta heuristics have been proposed for homogeneous multiprocessing environment, where speed of processing elements are same, and heterogeneous multiprocessing environment, where speed of pro-

93

ISBN: 978-969-8680-32-9


3

cessing elements are different [1]. Some proposed solutions used local optimization techniques and other used global optimization methods. Local optimization means optimizing the schedule of a considered job while in a global optimization schedule of a complete workflow is considered for optimization. Heuristics proposed for a job and resource selection may fall into following categories: Individual task scheduling, List scheduling, Workflow based scheduling, Cluster based scheduling, Duplication based scheduling and Meta heuristics [1]. Only few of the List scheduling heuristics are briefly discussed here. Heterogeneous Earliest Finish Time (HEFT) [12]: Jobs are assigned priority based on a rank which is calculated by taking average execution time for each task and average communication time between resources of a job and its child job. The longest path length from a job to the last job in the WF gives the rank of a job. Thus, jobs are scheduled in descending order of rank on a fastest resource which may improve the overall makespan of a WF. DCPG [9]: This is a modified form of Dynamic Critical Path (DCP) [5] scheduling technique. DCP was proposed for a homogeneous environment while DCPG was proposed for a heterogeneous environment like Grid. The critical path, calculated on the basis of communication and computation cost, is the longest path from starting jobs to the ending jobs of a WF. Jobs on a critical path, some time referred to as most critical jobs, are given priority for scheduling. A critical path of a WF may change at every step of scheduling because of scheduling decision. Resource which gives possible earliest completion time of a job and possible earliest starting time for its most critical child job is selected. When mapping is done then execution time and data transfer time of a job is updated according to the situation and this may change criticality of remaining jobs. PDCPG [2]: It was proposed for AR Grid environment. Jobs are selected according to DCPG job selection method. PDCPG proposed a partner based resource selection technique. Jobs are said to be partners if they have at least one common child job. While selecting a resource for a job, its scheduled partner job’s resource is considered according to a condition (see Line 22 in Algorithm 1). If the condition is met then the partners resource is selected, otherwise it is assigned to a fastest resource. If its partners are not scheduled or it has no partners then it is assigned to a resource which may finish it earliest. Thus, PDCPG improves makespan by reducing communication time between parents and child jobs because of this partner based scheduling technique. The probability of adjustment of partner jobs on a resource can be increased, which can help in better performance in terms of WF makespan. Further, in AR

94

ISBN: 978-969-8680-32-9

4


environment the future load of a network is readily available, thus can be used to improve WF makespan even further. The proposed technique addresses these two points.

3 Terminology Problem Definition: Terminology and and Problem Definition Before giving the problem definition, the WF model and Grid resource model used for the proposed work is mentioned. Workflow Model:Grid workflow is represented in the form of Graph G = (V, E), where V is a set of v nodes/jobs and E is a set of e edges. An edge connects two jobs, i.e., parent (predecessor) job with its child (successor) job, and it means parent job will be executed before its child job.A job j has a set of x partner jobs P T Rj = {ptr1 , ptr2 , . . . ptrx }, where P T Rj ⊂ V . Job size of job j is denoted by JSj and measured in Million Instructions (M I). Jobs can take multiple input files, generated by their parents and can produce one output file. It is assumed that only starting jobs need some input files which exist on some resources before submission of WF. Jobs’ output files are added to the set of files F = {f1 , f2 , f3 , . . . fy } and data size of file f is denoted by DSf which is measured in Mega Bytes M B, as integer number. Grid Resource Model: Grid has set of n resources R = {r1 , r2 , . . . rn } and a set of m network links L = {l1 , l2 , . . . lm } . Each r ∈ R has multiple CPUs having the same processing speed denoted by P Sr and is measured in Million Instruction per Second M IP S. A user is assumed to be attached with r0 ∈ / R, from where he submits a job. It is assumed that all links and resources support AR and the adopted network is OBS. Information about a reservation is stored in Utilization Profiles of resources and network links. A resource’s utilization Profile upr stores the information about the number of reserved CPUs of resource r ∈ R and a link’s utilization profile upl maintains information about reserved bandwidth of link l ∈ L [3]. T is the size of all utilization profiles and limit within which a reservation can be made. Binary Capacity Availability Vector of utilization profiles is calculated and used as in [14]. 1 Granularity = v − ej

v−ej X i=1

estETji , ADT Tji

(1)

Granularity of the workflow is the average of the ratio of computation cost and communication cost of all jobs and exit jobs (having no child jobs, denoted by ej) are not included in this calculation (see equation 1). Problem Definition: Given the above Grid resource model and WF model, schedule all jobs of a WF, such that its makespan is minimised, where makespan is the difference between submission time of a WF and completion time of its last job.

95

ISBN: 978-969-8680-32-9


5

Algorithm 1 Proposed Resource Selection Pseudo-code 1: 2: 3: 4: 5: 6:

function Resource Selection(M) for each r in R do Find expCTM on r maxSTccM = expCTM for each ptr in PccM do Find expCTptr in upr within s and d, where s = AESTptr and d = expCTM + ADT TM − ADT Tptr 7: if (s ≤ expCTptr ≤ d) 8: maxSTccM = max(expCTptr , maxSTccM ) 9: else 10: Find expCTptr in upr within s and T 11: OutputArrptr = AESTptr + estETptr + ADT Tptr 12: minP T = min(expCTptr , OutputArrptr ) 13: maxSTccM = max(minP T, maxSTccM ) 14: end if 15: end for 16: Find expCTcc after maxSTccM on r 17: minexpCTcc = min(expCTcc , minexpCTcc ) 18: end for 19: Select r which gives minexpCTcc 20: end function 21: function Resource Selection(ρ) 22: if (expCTρ ≤ CTM + ADT TM − ADT Tρ on rM ) 23: Schedule ρ on rM 24: else 25: Find OutputArrρ on rM from r ∈ R after expCTρ on r 26: Schedule ρ on r which gives min OutputArrρ on rM 27: end if 28: end function

4

Proposed Solution

Proposed Solution

In PDCPG2, DCPG’s job selection technique is adopted. As mentioned earlier, PDCPG2 attempts to accommodate all partner jobs on a single or a few resources. A new resource selection technique is introduced for the most critical partner jobs, which is expected to be scheduled ahead of all its partners. Furthermore, PDCPG’s method to adjust a partner job on its already scheduled partner’s resource has been extended as well. A brief description of the technique is as follows. The new addition is related to the most critical partner jobs and their most critical child job. The most critical partner job is scheduled on a resource on which earliest starting time of its most critical child job is achieved. The mechanism to calculate the expected completion time of this most critical child job is explained in the description section of the Algorithm. 1. If a job is a less critical partner then its availability is checked on its scheduled partner’s resource based on the condition given in [2] (see Line. 22 in Algorithm. 1). If the condition is met then it is scheduled otherwise the resource which is expected to give minimum arrival time of its output file, after its CT on the resource under consideration, to its most critical partner’s resource, is selected. These details are also given in the description section of Algorithm. 1. In the next section notation for the Algorithm 1 is given, and description follows it.

96

ISBN: 978-969-8680-32-9

6


Notation AESTj is the Absolute Earliest Starting Time, i.e., lower bound for starting time of a job j, ALSTj is the Absolute Latest Starting Time, i.e., upper bound for starting time of a job j. estETj is the estimated Execution Time of a job j and is calculated by getting the execution time of job j on the fastest resource. ADT Tj is the Absolute Data Transfer Time, i.e., data transfer time for the output file of a job j by using requested bandwidth. AESTj , ALSTj , estETj and ADT Tj , for all j ∈ V , are used to calculate criticality of jobs, as in [9]. maxSTj is the maximum Starting Time of a job j. expCTjr is the expected Completion Time of a job j on resource r which is being considered for mapping, but not yet selected for mapping. M is the most critical partner job to be scheduled. cc is the most critical child job of M. P is the list of partners of M having common child cc. P is sorted in descending order of criticality. ρ is a less critical partner job. OutputArrρ is the arrival time of output file of partner ρ on the resource where M is scheduled. Description of Algorithm 1 For M, the resource which gives the minimum of maxSTccM is selected as follows. Find expCTM on each resource r ∈ R and set maxSTccM to expCTM . Afterwards, find the estimated availability (ahead of its partner’s actual schedule) for all ptr ∈ PccM in the utilization profile of the same resource from s to d slots. The purpose of the range s to d is to keep the search space smaller to avoid a possible increase in runtime complexity of algorithm while getting a reasonable estimation of the availability. Set maxSTccM by choosing the maximum of expCTptr or maxSTccM . If upr slots are not available between the range s to d then find expCTptr from s to T slots and calculate the estimated arrival time of the output file of ptr, which is OutputArrptr , on resource r(See line 11). Set minP T as minimum of expCTptr and OutputArrptr , which gives the possible availability of output of ptr on the resource r. Set maxSTccM to the maximum of minP Tptr and maxSTccM . After getting maxSTccM , find expCTcc on each resource r. Set minexpCTcc by choosing the minimum of expCTcc and minexpCTcc . At the end select a resource r which gives the minimum of expCTcc . For a less critical partner job ρ, the resource is selected as follows. If the condition (see line 22) for expCTρ is met on the resource on which M is scheduled then schedule ρ on it. Otherwise, find expCTρ on each resource r in R and then find OutputArrρ , after expCTρ , from the resource r to rM . The resource which gives minimum OutputArrρ to rM is selected for ρ.

5 Evaluation Evaluation The simulation is done using Gridsim [11]. AR for network and computing resources is implemented as proposed in [14] and network architecture is OBS. A user submits his workflow for AR of bandwidth and processing element, with

97

ISBN: 978-969-8680-32-9


7

the objective of the earliest completion time. Bandwidth of all links is set to 100 Mega bits per second(Mbps). For all experiments requested bandwidth is set to 10 Mbps and requested CPUs is assumed to be one. Hundred instances of simulation for each WF are created for each combination of computing and network resources configuration and granularity, and average of makespan (in seconds) has been taken from it. The three configurations of computing and network resources, which are related to load, are high, medium and low. High network load means 50 to 80% bandwidth has already been reserved on a link. Similarly, for medium load, it is 30 to 50% and for low load it is 5 to 30%. For resources, the settings for high, medium and low load varies corresponding to a workflow. The loads are set in such a way so that availability of a resource for the accommodation of partner jobs and/or a complete worklflow becomes harder on high load, less harder on medium load and easier in case of low load. Three different high, medium and low granularity configurations consist of job length and data file sizes ranges. Job length ranges from 40, 000 to 60, 000 Million Instructions (M I) and for some workflow it ranges from 60, 000 to 90, 000 M I. Data sizes have three different configurations according to job length. When job length ranges from 40, 000 to 60, 000 M I, for low granularity data size ranges from 60 to 90 MB, in medium granularity data size ranges from 40 to 60 MB and high granularity contains data size from 10 to 20 MB. When job length ranges from 60, 000 to 90, 000 M I then for low granularity data size ranges from 30 to 50 MB, for medium granularity data size ranges from 60 to 90 MB and data size ranges from 90 to 120 for high granularity. In the remaining section the results for each workflow are analysed. e-Protein [7] In e-Protein WF there are 15 jobs, at the third last level there are 5 partners and these partners are child of different jobs. DCPG might scatter these partners on different resources closer to their parents. PDCPG2 might schedule these partners together on one resource. Thus PDCPG2 performed better than DCPG, especially in case of low granularity, in which 7 out of 9 load settings showed very encouraging improvements. The reason for bigger improvements is the increase in the sizes of makespan because of high data demands 1a). The greatest of the improvement, i.e, 37% is also achieved in low granularity. Other mentionable big improvements range from 10% to 31% (see Figure.1b). It is worth mentioning over here that gaining such improvements for smaller WF is very promising, because in AR it is harder to achieve greater improvement for smaller WF [1]. Since, PDCPG and PDCPG2 performed closely, it is safe to assume that the PDCPG2’s gain over DCPG is because of the settings rather than its modified resource selection technique. Job60PSPLIB [4] Job60PSPLIB WF has 60 jobs and there are 13 jobs which have 3 parents and 12 jobs which have 2 parents. These jobs can increase communication cost if their parents are scheduled on different resources. Therefore, opportunities to gain more improvements are greater in this WF as compared to

98

ISBN: 978-969-8680-32-9

8

Extension to Partner Based Grid Workflow Scheduling Heuristic 1800 1600 1400 1200 1000 800 PDCPG2 DCPG

600

PDCPG 400 200

High Granularity

4% 0% 11% 2% 0% 0% @1% 0% 0% 0%

8% 2%

8% 2%

LowResLowNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResMedNet

MedResLowNet

LowResHighNet

LowResLowNet

LowResMedNet

LowResHighNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResMedNet

LowResLowNet

7% 14% 14% 19% 31% 12% 0% @2% @2% 2% 3% 0%

HighResHighNet

LowResMedNet

Low0Granularity LowResHighNet

MedResLowNet

MedResHighNet

HighResLowNet

HighResHighNet 4% 0%

HighResMedNet

LowResLowNet

LowResMedNet

Medium0Granularity LowResHighNet

MedResLowNet

MedResHighNet

MedResMedNet

HighResLowNet

HighResMedNet

HighResHighNet

4% 3% 0% 16% 7% 0% 0% @2%

(b)

Low Granularity

Average makespan (in seconds) of heuristics for e-Protein workflow. High0Granularity

vs0DCPG vs0PDCPG

MedResMedNet

HighResLowNet

Medium Granularity

MedResMedNet

(a)

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResHighNet

HighResMedNet

0

5% 29% 10% 9% 21% 21% 6% 37% 13% 2% 9% 1% @2% 0% 0% @2% 5% @4%

Performance of PDCPG2 against other heuristics for e-Protein workflow.

Fig. 1: e-Protein Workflow

e-Protein WF. That is why huge improvements can be observed in the results. Furthermore, the improvement is greater for low granularity and high load settings, and it decreases as the granularity increases and load decreases. The effect of the path finding technique in PDCPG2 is noticeable especially in high load settings, where upto 79% improvement has been observed (see Figure.2b). Consistent high improvements have been observed in a very encouraging way in low and medium granularities for all load combinations, which range from 21% to 79%. High improvements are also because of the greater number of jobs in WF. Makespan of DCPG and PDCPG increased upto 6500 and PDCPG2 makespan ranges upto 1300. Makespan increases with the increase in load and decrease in granularity and vice versa (see Figure. 2a). Another noticeable aspect of these results is that the PDCPG’s partner based technique couldn’t perform differently from DCPG, which helps in concluding that the combinations of settings are not favourable for PDCPG’s partner based technique but the enhancement of the technique found the better schedules.

99

ISBN: 978-969-8680-32-9


9

7000

6000

5000

4000

3000

PDCPG2 DCPG

2000

PDCPG

1000

High Granularity

(b)

LowResLowNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

LowResLowNet

50% 38% 50% 38%

LowResMedNet

78% 78%

LowResHighNet

4% 9%

MedResLowNet

13% 14%

HighResLowNet

31% 32%

MedResMedNet

5% 8%

MedResHighNet

14% 14%

HighResMedNet

42% 42%

LowResLowNet

12% 15%

HighResHighNet

21% 21%

LowResMedNet

54% 53%

Low0Granularity LowResHighNet

MedResHighNet

MedResMedNet

4% 3%

HighResLowNet

5% 5%

HighResHighNet

LowResHighNet 2% 7%

HighResMedNet

4% 2%

LowResLowNet

4% 5%

Medium0Granularity LowResMedNet

4% 8%

MedResLowNet

0% 2%

MedResHighNet

HighResLowNet

1% 4%

MedResMedNet

HighResMedNet

HighResHighNet 4% 7%

Low Granularity

Average makespan (in seconds) of heuristics for Job60PSPLIB workflow. High0Granularity

vs0DCPG vs0PDCPG

MedResMedNet

HighResLowNet

Medium Granularity

MedResLowNet

(a)

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResHighNet

HighResMedNet

0

79% 78%

52% 50%

39% 38%

62% 61%

45% 45%

28% 28%

Performance of PDCPG2 against other heuristics for Job60PSPLIB workflow.

Fig. 2: Job60PSPLIB Workflow

Job120PSPLIB [4] Job120PSPLIB WF has the similar layout of partner job’s as in case of Job60PSPLIB WF. The same sequence of improved percentage in all granularities has been observed as like in Job60PSLIB. PDCPG2 outperforms the DCPG and PDCPG in high load combinations in all granularities and a path finding effect is also noticeable, especially in low granularity with high load settings. 66% improvement has been achieved in high load settings. In tight situations where DCPG and PDCPG makespans reach 7770, PDCPG2 makespan remained within 2667 (see Figure.3a), thus gave huge improvements (see Figure.3b). As in case of Job60PSPLIB WF, consistent high improvements are also noticeable in JOB120PSPLIB results which range from 23% to 66% in all load combinations (see Figure.3b). Almost 18 combinations showed outclass performance, which is phenomenal. It can also be observed from results of Job60PSLIB, e-Protein and Job120PSLIB WFs, with the increase in number of jobs, makespans and improvements also increased. As in case of Job60PSPLIB, old partner based technique couldn’t perform well in these results as well, emphasising on the fact that enhanced partner based technique has a great potential.

100

ISBN: 978-969-8680-32-9

10


9000 8000 7000 6000 5000 4000 PDCPG2 DCPG

3000

PDCPG 2000 1000

High Granularity

LowResLowNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

LowResLowNet

LowResMedNet

MedResLowNet

LowResHighNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResMedNet

Low Granularity LowResHighNet

MedResLowNet

MedResMedNet

HighResLowNet

HighResMedNet

HighResHighNet

LowResLowNet

LowResMedNet

Medium Granularity LowResHighNet

MedResLowNet

MedResMedNet

MedResHighNet

HighResLowNet

HighResMedNet

HighResHighNet

Low Granularity

Average makespan (in seconds) of heuristics for Job120PSPLIB workflow. High Granularity

vs DCPG vs PDCPG

MedResMedNet

HighResLowNet

Medium Granularity

MedResHighNet

(a)

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResHighNet

HighResMedNet

0

16% 9% 8% 25% 10% 9% 32% 9% 8% 59% 31% 33% 62% 36% 18% 62% 41% 29% 66% 64% -9% 59% 57% 52% 62% 57% 50% 16% 7% 8% 23% 10% 8% 25% 6% 7% 57% 24% 50% 61% 32% 18% 58% 34% 18% 62% 60% 7% 57% 53% 42% 53% 50% 40%

(b)

Performance of PDCPG2 against other heuristics for Job120PSPLIB workflow.

Fig. 3: Job120PSPLIB Workflow

LIGO [6] LIGO is a 90 jobs WF and there are 41 pairs of partners, 4 groups of 4 partners, 5 groups of 5 partners and couple of groups of 6 to 9 partners. It can be observed that the performance of PDCPG2 is not as consistent as in case of previously discussed WFs. This inconsistency could be because of the layout of the WF. There are 6 such partners in the WF which have 2 common child jobs and same is the case with few other smaller groups of partners. This layout could create an imbalance situation for partners’ adjustment on a single or few resources. The number of marginally improved or worsen performance is little more than the performances of huge or big improvements. Overall, low granularity settings yield most of the huge or big improvements, especially against DCPG. A huge improvement of 65% against DCPG and big improvement of 28% is achieved in high load scenario (see Figure. 4b). Other improvements against DCPG range from 10% to 28%. Despite a difficult layout of WF, from the partner based adjustment perspective, the number of improved performances are remarkable. Furthermore, the improvements become more valuable, since most of these are achieved in case of larger sizes of makespans (see Figure. 4a). An-

101

ISBN: 978-969-8680-32-9


11

7000

6000

5000

4000

3000

PDCPG2 DCPG

2000

PDCPG

1000

High Granularity

LowResLowNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

10% 514% 2% 57%

LowResLowNet

28% 1%

LowResMedNet

22% 14% 17% 53%

LowResHighNet

65% 28%

MedResLowNet

11% 56%

MedResMedNet

16% 3%

HighResLowNet

26% 16%

MedResHighNet

56% 51%

HighResMedNet

15% 52%

LowResLowNet

16% 9%

HighResHighNet

LowResMedNet

57% 3%

LowResHighNet

9% 12%

MedResLowNet

HighResLowNet

52% 18%

Low0Granularity

MedResHighNet

HighResHighNet

12% 15% 8% 11% 5% 55%

HighResMedNet

LowResLowNet

LowResMedNet

10% 0% 2% 51%

Medium0Granularity LowResHighNet

20% 16%

MedResLowNet

MedResHighNet

MedResMedNet

HighResLowNet

HighResMedNet

HighResHighNet

52% 54% 55% 8% 7% 9%

(b)

Low Granularity

Average makespan (in seconds) of heuristics for LIGO workflow. High0Granularity

vs0DCPG vs0PDCPG

MedResMedNet

HighResLowNet

Medium Granularity

MedResMedNet

(a)

MedResHighNet

HighResMedNet

LowResLowNet

HighResHighNet

LowResHighNet

LowResMedNet

MedResLowNet

MedResMedNet

HighResLowNet

MedResHighNet

HighResHighNet

HighResMedNet

0

19% 2%

2% 53%

14% 2%

Performance of PDCPG2 against other heuristics for LIGO workflow.

Fig. 4: Workflow LIGO

other interesting aspect of the these results is the performance against PDCPG in high load settings in case of low and medium granularity (see Figure. 4b). It seems that in high load settings, PDCPG struggles to fit partner jobs in few resources as compared to PDCPG2.

6 Conclusion Conclusion Proposed technique showed phenomenal improvement for all tested workflows’ makespan, especially in a high congestion situation of computing and network resources. One promising aspect of the results is the improvement in case of smaller workflows, which is hard to achieve in AR environment. Both new additions to PDCPG, i.e., resource selection method for most critical partner job and routing feature in resource selection for a less critical partner jobs, caused the huge improvement in the higher resource and network load cases. The better of the performance is observed in the case of 120 jobs workflow, where the improvement is huge when other heuristics gave large makespans, which makes the improvement even more valuable. There are 16 instances of settings for 120

102

ISBN: 978-969-8680-32-9

12


job workflow in which improvements of 30% to 66% are gained. Many other instances of settings for smaller workflows also gave big improvements, which range from 15% to 78%. The proposed technique also performed very well in case of LIGO workflow, which seemed a harder situation for the partner based technique to perform well.

References 1. Ashraf, J.: Partner-Based Scheduling and Routing for Grid Workflows. Ph.D. thesis, Department of Computer Science, University of Leicester (2012) 2. Ashraf, J., Erlebach, T.: A new resource mapping technique for grid workflows in advance reservation environments. In: Proceedings of High Performance Computing and Simulation (HPCS). pp. 63 –70 (2010) 3. Christodoulopoulos, K., Doulamis, N., Varvarigos, E.M.: Joint communication and computation task scheduling in grids. In: CCGRID ’08: Proceedings of the 2008 Eighth IEEE International Symposium on Cluster Computing and the Grid. pp. 17–24. IEEE Computer Society, Washington, DC, USA (2008) 4. Kolisch, R., Sprecher, A.: Psplib - a project scheduling problem library. European Journal of Operational Research 96(1), 205–216 (1997) 5. Kwok, Y.K., Ahmad, I.: Dynamic critical-path scheduling: An effective technique for allocating task graphs to multiprocessors. IEEE Transactions on Parallel and Distributed Systems 7(5), 506–521 (1996) 6. LIGO: http://www.ligo.caltech.edu, last accessed: June, 2014 7. O’Brien, A., Newhouse, S., Darlington, J.: Mapping of Scientific Workflow within the E-Protein Project to Distributed Resources. In: UK e-science all-hands meeting, AHM 2004, Nottingham, UK, September 2004. pp. 404–409 (August 2004) 8. Pandey, S., Buyya, R.: Scheduling of scientific workflows on data grids. In: Proceedings of the Eighth IEEE International Symposium on Cluster Computing and the Grid. pp. 548 –553 (2008) 9. Rahman, M., Venugopal, S., Buyya, R.: A dynamic critical path algorithm for scheduling scientific workflow applications on global grids. Proceedings of the 3rd IEEE International Conference on e-Science and Grid Computing (e-Science 2007) pp. 1–8 (2007) 10. Smith, W., Foster, I., Taylor, V.: Scheduling with advanced reservations. Proceedings 14th International Parallel and Distributed Processing Symposium pp. 127–132 (2000) 11. Sulistio, A., Yeo, C.S., Buyya, R.: Visual modeler for grid modeling and simulation (gridsim) toolkit 2659, 1123–1132 (2003) 12. Topcuoglu, H., Hariri, S., Wu, M.Y.: Performance-effective and low-complexity task scheduling for heterogeneous computing. IEEE Transactions on Parallel and Distributed Systems 13(3), 260–274 (2002) 13. Ullman, J.D.: Np-complete scheduling problems. Journal of Computer and System Sciences 10(3), 384–393 (1975) 14. Varvarigos, E., Sourlas, V., Christodoulopoulos, K.: Routing and scheduling connections in networks that support advance reservations. In: Broadband Communications, Networks and Systems, 2008. BROADNETS 2008. 5th International Conference on. pp. 536–543 (Sept 2008)

103

ISBN: 978-969-8680-32-9

A Fast Encryption/Decryption Scheme for

A Fast Encryption/Decryption Scheme for BiometricImages ImagesUsing usingMultiple MultipleChaotic ChaoticMaps Maps Biometric Jawad Ahmed1, Soong Oun Hwang2 1 Ahmad Seong Engineering, Oun Hwang2 Hongik University, 1 Department of Jawad Electronics and and Computer South Korea Department of Electronics and Computer Engineering 1

Graduate School, Hongik University, South Korea

2 Department of [email protected] and Information Communication, Hongik University, 2

Department of Computer and Information Communication South Korea

Engineering, Hongik University, South Korea [email protected] [email protected], [email protected]

Abstract. To achieve the confidentiality of digital data, many encryption algorithms have been proposed so far. However, many of these traditional algorithms are not suitable for the real time processing of large biometric data. In order to achieve fast encryption/decryption time which are required in practical systems, we propose a new chaotic-based biometric image encryption scheme. The proposed scheme utilizes Bernoulli shift map and Logistic map for the confusion and diffusion stages, respectively. The plaintext image is firstly permuted block-wise using random indices, which are generated through the iteration of Bernoulli shift map. To add diffusion and mixing properties, Logistic map has been used due to its complex behaviour. In the last step of the proposed scheme, we increase the security against various attacks by using the Gray substitution box. Unlike the traditional encryption algorithms, the proposed image encryption scheme is significantly fast and proven to be immune against information extracting by testing through many statistical and security parameters.

1 Introduction Introduction With the evolution of technology, threats to important confidential information have became more crucial [1]. That is the reason why most of the researchers got attention to the field of information security. The subject of information security deals with developing new and robust techniques for securing important multimedia contents. Images are widely used in different scenes of life and can be easily stored or transmitted to distant places on the Internet. However, Internet is an insecure channel and images are always vulnerable to eavesdropping [2]. During transmission, the confidentiality of images are required in most of applications like video-on-demand, Internet security camera and telemedicine [3]. Due to the close relationship between chaos and cryptography, chaotic maps have been extensively applied during the last two decades for the design of efficient image encryption algorithms. To protect images from eavesdroppers, some nonlinear dynamic properties of chaotic maps can be used as a source of diffusion

104

ISBN: 978-969-8680-32-9

2

in images [4]. In [5], Chen et al. generalized a two-dimensional chaotic cat map to 3D for the design of a secure image encryption. The three-dimensional map is applied on image pixels to shuffle the positions and gray values for enhancing the ability against statistical attacks. To increase the speed of image encryption, Mao et al. extended a two dimensional baker map to three-dimensional map [6]. The concept of external 80-bit secret key for two Logistic maps were given in [7] by Pareek et al. To make the algorithm more secure against various kinds of attacks, Pareek et al. carried out eight different types of operations and modified the secret key after each block of sixteen pixels. In [8], faster diffusion of image pixels was achieved by using standard chaotic map through simple sequential add-and-shift operations. In 2009, Sahar et al. proposed a novel technique for encrypting color images by using coupled nonlinear chaotic map [9]. To generate the initial conditions for chaotic map, Sahar et al. used 240 bit-long secret key which increases the key space against attackers [9]. Many techniques for image encryption have been proposed so far, but some of these algorithms are vulnerable to different type of attacks, as mentioned in [10]. Problem Statement: Many traditional or modern cryptosystem such as Rivest Shamir Adleman (RSA), Data Encryption Standard (DES) and Advanced Encryption Standard (AES) have been designed to encrypt textual data. These traditional encryption algorithms can be used to encrypt an image directly. However, due to the cost of computational complexity, traditional algorithms are not suitable for the real time encryption of biometric data. Generally, the size of biometric images is much larger than those of texts and hence, much time is required for encryption and decryption. Due to the aforementioned reasons, there is a strong need to design and implement new encryption methods, especially for high correlated and large biometric data. New algorithm should be designed in such a way that it fulfil the criteria of security and real time operation.

2 Fundamental Knowledge Fundamental Knowledge In this paper, we use two chaotic maps, Bernoulli shift map and Logistic map for confusion (permutation) and diffusion process (XOR), respectively. To enhance the security feature, we also use the Gray Substitution Box (S-Box) [11] in the proposed scheme. To get a better understanding of the proposed scheme, we briefly discuss how the Bernoulli shift map, Logistic map and Gray S-Box can play a key role in encryption algorithms. 2.1

The Bernoulli Shift Map

The Bernoulli shift map (also known as dyadic map, bit shift map, sawtooth map and doubling map) is shown in Fig. 1. Mathematically, Bernoulli map can be written as iterated shift map:

105

ISBN: 978-969-8680-32-9

3

Fig. 1: The Bernoulli shift map xn+1 = 2xn mod 1 =

2xn , 2xn − 1,

0 ≤ xn < 12 1 2 ≤ xn < 1

(1)

Equation 1 exhibits a deterministic chaotic behaviour by using stretch and cut mechanism, as discussed in [12]. The most general form of Bernoulli map can be modelled as [13]: xn mod 1, (2) xn+1 = a where ∀ xn ∈ [0, 1]: xn+1 also ∈ [0, 1]. Both xn and xn+1 are known as the states of the Bernoulli map and a ∈ [0, 1] is the control parameter. 2.2

The Logistic Map

In 1976, Robert May proposed a simple model (known as Logistic map) to describe the dynamics of insect population. The continuous form of Logistic map is shown in Eq. 3: dx = rx(1 − x), (3) dt where r is a positive parameter representing the rate of population growth. The discrete version in the form of recurrence relation can be written as: xn+1 = rx(1 − xn ), 0 ≤ xn ≤ 1.

(4)

In the discrete version of Logistic map r ∈ (0, 4). Equation 4 can be used for illustration purpose to show how a chaotic behaviour arises from a very simple system. The parameter r in the Logistic map plays a vital role in the stability and chaotic nature of the map. When the value of r ∈ (0, 3), xn reaches a fixed point after some iterations indicating the stable state as shown in Figs. 2(a-d).

106

ISBN: 978-969-8680-32-9

4

When r ∈ [3, 3.6), the iterative sequence for xn swings between some values showing the periodic states. This periodic states of Logistic map are shown in Figs. 2(e-f). Figures 2(g) and 2(h) highlight the chaotic nature of a Logistic map for the range of [3.6, 4) which is a hidden nonlinear property in a simple formula. Thus the Logistic map has different kinds of regular behaviour in the range of (0, 3.6) and it also has a chaotic behaviour in contrast to linear systems when r ∈ [3.6, 4). The range of r of [3.6, 4) is more useful for implementing an encryption algorithms. We utilize the chaotic range for implementing our image encryption scheme in Sec. 3. The chaotic range (r ∈ [3.6, 4) for the Logistic map seems very small but in fact a small change in initial condition generates different outputs after 22nd iteration. These types of effects are shown in Figs. 3(a) and 3(b) for slightly different r and x, respectively. Figure 3 exhibits that the orbit for the Logistic map is too large. 2.3

Gray Substitution Box

In [11], Tran et al. presented a new substitution box, known as Gray S-BOX. To enhance the security of AES, the proposed Gray S-Box has been implemented in existing AES. The results were then compared and found worthwhile for the AES. The Gray S-Box can increase the security of any encryption scheme against algebraic and interpolation attacks. It has been proven that Gray S-Box satisfies all the necessary requirements like strict avalanche criterion, nonlinearity, and differential uniformity which are essential for encryption process. Due to the aforementioned reasons, we also use the Gray S-Box in our proposed scheme. Details of “Gray augmentation” and Gray matrix can be found in [11].

3Description Description of the Proposed Scheme of the Proposed EncryptionEncryption Scheme In this section, we present a new encryption scheme based on Bernoulli shift map, Logistic map and Gray S-Box for large biometric images. For the sake of generality, we consider biometric images of size M × N pixels. Each step of the encryption process is discussed in more detail below. 1. The biometric input image P of size M ×N is first broken into non-overlapping blocks of size K × K. 2. Random indices Λ is obtain by iterating the Bernoulli map B times. The value of B is calculated using Eq. 5, which is formulated through the size of image and block. Mathematically, B can be written as: M ×N . (5) K ×K 3. The image divided into blocks in step 1 are now randomly permuted though pseuedo-random permutation indices Λ to obtain a permuted image Pˆ . 4. Sum all the pixel values of Pˆ which is obtained in step 3 to get a value S: B=

S=

m X n X

Pˆ (i, j).

(6)

i=1 j=1

107

ISBN: 978-969-8680-32-9

5

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 2: Showing the plot for Logistic Map against number of iterations for differ108 ISBN: 978-969-8680-32-9 ent values of r (x0 = 0.1).

6

(a) Orbit of the Logistic map with x0 = (b) Orbit of the Logistic map with r = 0.1000. 3.8000.

Fig. 3: Showing the plot for Logistic map against number of iterations for slightly different values of r and x0 . 5. The value S is used in the key generation phase. Here key is the initial condition x0 for the Logistic map: x0 =

Mod (S, 256) . 256

(7)

In order to change the ciphertext image completely in an unpredictable manner for one bit change in the plaintext image, calculation of x0 is an important phase. The value of x0 is made dependent on both S and Modular operation to get higher values of Number of Pixel Change Rate (NPCR) and Unified Average Change Intensity (UACI). 6. A matrix ψ is generated by iterating the Logistic map M × N times using the initial condition x0 . 7. To get a random matrix ∆ of size M ×N with values in [0, 255], ψ is multiplied by a value of 1014 under the modulo 256 operation. The step is summarised in Eq. 8. ∆ = Mod (ψ × 1014 , 256). (8) 8. The random matrix ∆ is bit-wise XORed with the each pixel of permuted image Pˆ to obtain Υ . 9. To get the ciphertext image C, substitute the values of Υ with the elements of Gray S-Box. The substituted version of Υ can be called the resultant ciphertext C. To get the plaintext image P from the cipertext image C, steps from 1 to 9 are applied in reverse order, which is also known as decryption process.

4Experimental Experimental ResultsResults To demonstrate the effectiveness of the proposed scheme, several experimental results are presented in this section. The proposed algorithm is implemented in

109

ISBN: 978-969-8680-32-9

7

MATLAB R2014a using biometric iris and fingerprint gray-scale images of size 256 × 256. In the experimental analysis, the selected value of K is 16. Intuitively, when the size of image is 256 × 256 and K is 16, obviously B is 256. 4.1

Biometric Image Encryption

Figures 4(a) and 4(b) show the plaintext (original) iris and encrypted images, respectively. The encrypted version of iris has no similarity with the plaintext version. In [14], Elashry et al. have highlighted that when the histogram of an encrypted image is uniform, the encryption algorithm is considered good. From Figs. 4(c) and 4(d), it can be seen that the histogram of encrypted iris image is almost flat as compared with the histogram of plaintext iris image. Similar results are obtained for the fingerprint image, as shown in Fig. 5. By visual inspection and histogram analysis, it is noticeable that the encrypted image does not reveal any significant information about the plaintext iris and fingerprint images. However, only these analyses are not sufficient to prove that the proposed encryption scheme is secure and robust. Some statistical analysis is also essential to prove the security of proposed encryption scheme. 4.2

Security Analysis

In our previous work [15], we have underlined various security parameters which can be used for the analysis of an image encryption scheme. These security parameters are correlation coefficient (C.C), entropy (Hm ), maximum deviation (MD ), irregular deviation (ID ), deviation from uniform histogram (DP ), Mean Square Error (M SE), Number of Pixel Change Rate (N P CR) and Unified Average Change Intensity (U ACI). How these parameters reflects the security and how it can be used in the evaluation of an image encryption algorithm can be found in [15]. These parameters are applied on both encrypted iris and fingerprint images and the results are listed in Table 1. The results depicted in Table 1 prove the security and validity of the proposed scheme against various statistical attacks. 4.3

Time Analysis

Due to high computational cost and time complexity, many traditional encryption schemes can not be employed as an image encryption algorithm as outlined in [16, 17]. For computational time analysis, the proposed scheme is implemented in MATLAB R2014a on a PC with 2.0 GHZ CPU and 3.0 GB memory. Table 2 reveals the lower computational complexity and hence proves a lightweight encryption/decryption when compared to traditional AES algorithm.

5 Conclusion Conclusion In this paper, we have presented a lightweight scheme for encryption and decryption of digital biometric images. The proposed scheme combines strong confusion and diffusion properties, as the basic requirements for secure images, using

110

ISBN: 978-969-8680-32-9

8

(a) Plaintext iris

(b) Encrypted iris

(c) Histogram of plaintext iris

(d) Histogram of encrypted iris

Fig. 4: Encryption and histogram illustration: iris image

111

ISBN: 978-969-8680-32-9

9

(a) Plaintext fingerprint

(b) Encrypted fingerprint

(c) Histogram of plaintext fingerprint

(d) Histogram of encrypted fingerprint

Fig. 5: Encryption and histogram illustration: fingerprint image

112

ISBN: 978-969-8680-32-9

10

Table 1: Security evaluation of proposed scheme Security Parameter

Iris

Fingerprint

C.C

0.0045

0.0036

Hm

7.9973

7.9972

MD

50335

55144

ID

37754

12028

DP

0.0438

0.0411

M SE

38.6655

40.0952

N P CR

99.6078

99.6201

U ACI

33.7114

33.5695

Table 2: Processing time (sec) for the proposed scheme Process

AES

AES

Proposal

Proposal

Iris

Fingerprint

Iris

Fingerprint

Encryption

0.82

0.81

0.06

0.05

Decryption

1.16

1.13

0.18

0.17

113

ISBN: 978-969-8680-32-9

11

chaotic maps and Gray S-Box. The sensitivity of Logistic map to the initial conditions and control parameters proves the security of the proposed scheme. Gray S-box is applied after the diffusion process to change the image pixels in a complex way. We analysed the scheme through performing time complexity and various differential and statistical tests. The results of all tests are in favour of proposed scheme confirming its higher security as well as less computational cost.

Acknowledgements This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2012002139) and funded by the Ministry of Education (2014054174). This research was also supported by the MSIP (Ministry of Science, ICT and Future Planning), Korea, under the Global IT Talent support program (NIPA2014-H0905-14-1004) supervised by the NIPA (National IT Industry Promotion Agency).

References 1. Daya, B.: Network security: History, importance, and future. University of Florida Department of Electrical and Computer Engineering (2013) 2. Etemadi Borujeni, S., Eshghi, M.: Chaotic image encryption design using Tompkins-Paige algorithm. Mathematical Problems in Engineering 2009 (2009) 3. Huang, R., Rhee, K., Uchida, S.: A parallel image encryption method based on compressive sensing. Multimedia Tools and Applications (2012) 1–23 4. Fridrich, J.: Symmetric ciphers based on two-dimensional chaotic maps. International journal of bifurcation and chaos 8(06) (1998) 1259–1284 5. Chen, G., Mao, Y., Chui, C.K.: A symmetric image encryption scheme based on 3D chaotic cat maps. Chaos, Solitons & Fractals 21(3) (2004) 749–761 6. Mao, Y., Chen, G., Lian, S.: A novel fast image encryption scheme based on 3D chaotic baker maps. International Journal of Bifurcation and Chaos 14(10) (2004) 3613–3624 7. Pareek, N.K., Patidar, V., Sud, K.K.: Image encryption using chaotic Logistic map. Image and Vision Computing 24(9) (2006) 926–934 8. Wong, K.W., Kwok, B.S.H., Law, W.S.: A fast image encryption scheme based on chaotic standard map. Physics Letters A 372(15) (2008) 2645–2652 9. Mazloom, S., Eftekhari-Moghadam, A.M.: Color image encryption based on coupled nonlinear chaotic map. Chaos, Solitons & Fractals 42(3) (2009) 1745–1754 10. Wang, X., Liu, L.: Cryptanalysis of a parallel sub-image encryption method with high-dimensional chaos. Nonlinear Dynamics 73(1-2) (2013) 795–800 11. Tran, M.T., Bui, D.K., Duong, A.D.: Gray S-Box for advanced encryption standard. In: Computational Intelligence and Security, 2008. CIS’08. International Conference on. Volume 1., IEEE (2008) 253–258 12. Wiggins, S., Ottino, J.M.: Foundations of chaotic mixing. Philosophical Transactions of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences 362(1818) (2004) 937–970

114

ISBN: 978-969-8680-32-9

12 13. Ye, R., Ma, Y.: A secure and robust image encryption scheme based on mixture of multiple generalized Bernoulli shift maps and Arnold maps. International Journal of Computer Network and Information Security (IJCNIS) 5(7) (2013) 21–33 14. Elashry, I.F., Faragallah, O.S., Abbas, A.M., El-Rabaie, S., Abd El-Samie, F.E.: A new method for encrypting images with few details using Rijndael and RC6 block ciphers in the electronic code book mode. Information Security Journal: A Global Perspective 21(4) (2012) 193–205 15. Ahmad, J., Ahmed, F.: Efficiency analysis and security evaluation of image encryption schemes. IJENS 23 (2012) 18–31 16. Chen, Z.: A lightweight encryption algorithm for images. In: Affective Computing and Intelligent Interaction. Springer (2012) 235–241 17. Behnia, S., Akhshani, A., Mahmodi, H., Akhavan, A.: A novel algorithm for image encryption based on mixture of chaotic maps. Chaos, Solitons & Fractals 35(2) (2008) 408–419

115

ISBN: 978-969-8680-32-9

I MT I C’ 15

ACCEPTEDPAPERS

Accepted Papers - Proceedings of IMTIC'15

Analyzing the Influence of Prosodic Features on Emotional States Recognition in Regional Languages of Pakistan Syed Abbas Ali, Maria Andleeb Siddiqui and Anas Khan N.E.D University of Engineering & Technology, Karachi. Faculty of Electrical Engineering, Usman Institute of 2 3 Technology, Karachi. [email protected], [email protected] and [email protected]

Abstract Emotion is complex states of feelings reflect the processing of inner state of human brain, resulting in physical and psychological changes which effect behavior and thought of the speaker. This paper introduces the corpus of emotional speech and emotional states recognition system based on observed values of prosodic features for recognizing the supervised emotional speech in regional languages of Pakistan: Punjabi, Pashto, Sindhi, Balochi and Urdu with four different emotional states (Happiness, Neutral, Anger and Sadness). The objective of this paper is to analyze the influence of prosodic features on recognizing the emotional states of speaker in regional languages of Pakistan. Initially, 55 samples have been collected randomly from both male and female native speakers for experimental evaluation in four different emotions using standard recording environment. Experimental results show the implication of prosodic features intensity and formant in recognizing the emotional states sad and happy respectively. Whereas, prosodic feature pitch shows significant accuracy in recognition of emotional states neutral and angry. Keywords: Index Terms-Prosodic features, Emotional states, Speech emotion, regional languages of Pakistan.

Introduction The speech emotion recognition is one of the challenging tasks due to following reasons. First, the extraction of the appropriate speech features for distinguishing among different emotion having considerable variability in spoken utterances of different sentences due to rate and style of the spoken utterances[1].Second, dependency of certain speech emotion on speaker culture and environmental conditions. Most of the research focused in this direction based on the assumption that no culture difference among speakers. Whereas, classification of multi-lingual speakers have been investigated in [2]. Third, detection of the longterm and short-term speech emotion recognition varies with the passage of time [3]. This paper focusing on the first issues related to speech features which are used to categorize the content of emotion from spoken utterances. Speech features can be further categorized into four groups: spectral, prosodic, TEO-based and qualitative features [4]. Speech not only conveys linguistic information but also contains paralinguistic component, which added the linguistic information by providing the cues for emotional states of speakers called prosody. In linguistic, prosody is defined as supra segmental properties of speech which comprises of the following correlates: loudness/intensity, pitch, rhythm/duration and formant perceived from spoken utterances of the speakers [brown 2005]. Speech features that capturing the supra segmental aspect of speech consider as prosodic features otherwise called acoustic features [5]. Speech physically apprehended the prosodic features as set of acoustic parameters which combines variation in energy and period of speech segment during pitch of speech. Prosodic features are used to convey added sense in natural speech to provide emotion in spoken word of speaker such as; happiness, comfort, sadness and anger. Prosodic features are treated as major correlates of vocal emotion for discriminating and identifying the emotion from spoken utterances and emotion present in daily life conversation respectively [6]. Speech emotion research community believed that the prosodic speech features provide considerable emotional contents of spoken utterances and commonly used in speech emotion recognition [1,7,8]. Three prosodic features: formant, intensity and pitch are considered here for parameterization of speech signal. Pitch is the perceptual property of speech based on prosodic feature having pitch contour which relates to an acoustic correlates and fundamental frequency F0 used to detect speech emotion from spoken utterances [9]. Intensity is one of the important property of speech used to encode the prosodic information contain the emotion contents of spoken utterances in term of speech signal energy measured by sensing the sound pressure of speaker [10]. Formant is defined as structure of the speech which provides resonant frequency of vocal track filter and defines speech articulation for parameter estimation of vocal track based on autoregressive modeling techniques (LPC) [11,12].Despite of the significant exploration of many speech features, the identification of the finest speech feature for speech emotion recognition is still a challenging task. This paper is an attempt to analyze the influence of prosodic features (Intensity, Pitch and Formant) on recognition of four different emotional states (Happiness, Anger, Neutral and Sadness) in regional languages of Pakistan. Rest of the paper is organized as follows. In section 2, we discuss the research methodology for estimation of prosodic features. Corpus collection and recording specification is presented in section 3. In section 4, briefly discuss

116

ISBN: 978-969-8680-32-9


the hardware specifications and demonstrative results of recognition system. Finally, the conclusion is drawn in section 5.

Methodology for prosodic Features Estimation The presented framework for analyzing the influence of prosodic features on emotional states recognition is based on the combination of prosodic features estimation and emotional states recognition system built on observed values estimated from prosodic features. In the first phase of experiment, the prosodic features (Pitch, formant, Intensity) were extracted from speech samples spoken by native speakers with four emotional states (Happy, Angry, Sad, Neutral) by implementing an algorithm using MATLAB tool kit. The values of prosodic features: pitch and formant were estimated using auto-correlation [13] and LPC method [14] respectively. Whereas, the reference estimated values for intensity is calculated from PRAAT software [15] and implemented in MATLAB tool kit. Formant Estimation Using LPC Method: LPC command is used to develop the predictor filter making use of speech emotion samples and order of filter. Estimating the values of formant speech signal taken from native speakers were calculated by keeping ‘fs’ equals to 44.1khz and predictor filter are used to measure the predictor polynomial. Polynomial roots were calculated using predictor polynomial and these poles is used to determine the formant frequencies of speech signal. The range of the first formant for recognition of happy emotion in regional languages of Pakistan were found in between 400=45dB, 16 bit recording format and sampling rate of 48 KHz based on standard ITU recommendations. The microphone sensitivity, impedance, stereo and cable length are 54dB±2dB, 2.2W and 1.8m respectively. The selection of a carrier sentences were based on the following criterions. First, sentence should be semantically neutral, secondly provide consistency with any situation in speech emotion and finally meaning of sentences easily identified for each languages. Based on the previous studies [16, 17], the selected carrier sentences are: “It will happen in seven hours” and “let’s go home”. These carrier sentences were spoken by their native speakers in different languages of Pakistan (Urdu, Pashto, Punjabi, Sindhi, and Balochi) with different emotional states (Angry, Sad, Happy, and Neutral). Urdu ‫يہ سات گھنٹے ميں ہو گا‬ Pashto ‫كيږي ساعتو اوو دابھپه‬ Punjabi ‫ستايہ ہو‬ َ ‫ئگا ِوچگھنٹے‬ Sindhi ‫ستھي ٿيندو ۾ کن کال‬ Balochi ‫بئيتہاکالکہفتھي‬ For second sentence in Urdu Language: Chalo Ghar Chaloǁ ―‫ چلــو گھـــر چلــو‬ǁ

Hardware Specifications and Results In the second phase of experiment, emotional states recognition system has been built based on mean observed values estimated from prosodic features using algorithm implemented in MATLAB tool kit for specific emotion. Developed system consist of GUI environment built in MATLAB tool kit for showing the output of the demonstrative experiment and PIC microcontroller 16F877A for interfacing result prosodic features estimation with hardware. Input speech samples is taken from Microsoft Windows 7 built-in sound recorder and applied in an algorithm implemented in MATLAB for extracting the mean values of different prosodic features for specific emotion. Once the emotion is verified, its value is stored in a buffer from where it was sent to developed system using serial communication as shown in Fig.6.

Fig. 6. Flow diagram of serial communication

The controller is programmed in such a manner that on receiving different character it works for different emotion, these characters send to the controller using MATLAB serial and buffers in which value of verified speech sample is stored. Once the controller initializes and received a particular character it glow different color LED for demonstrating different emotion also a string is displayed on the LCD which is at the start of the displaying “select audio for detection” and after verification of emotion display the type of emotional

119

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 state. Fig.7 displays the demonstrative experimental results of the emotional state recognition system based on observed values of the prosodic features for specific emotion by displaying the recognized emotion on LCD and glowing the LED on specific emotion.

Fig. 7. Demonstrative Experiment Result

Conclusion This paper introduced supervised speech emotion recognition system and emotional states corpus in regional languages of Pakistan: Pashto, Sindhi, Urdu, Punjabi and Balochi with four different emotional states (Happiness, Anger, Sadness and Neutral) using prosodic features (Intensity, Pitch and Formant). This research presented the initial study for analyzing the influence of prosodic features on emotional states recognition in regional languages of Pakistan. Experiments have been conducted on MATLAB tool kit using 55 speech samples taken randomly from both male and female native speakers. Experimental study of analyzing the influence of prosodic features on emotional states recognition in term of ranges of mean values demonstrated following observations: The prosodic feature intensity for developed system detecting the emotional state sad approximately between the ranges of values 72400 and 260> Pitch>200 respectively. Developed emotional states recognition system was based on above observations to detect different emotions using observed values of the prosodic features for specific emotion.

References: 1. Banse, R., Scherer, K.: Acoustic profiles in vocal emotion expression. J. Pers. Soc. Psychol. 70 (3), 614–636 (1996). 2. Hozjan, V., Kacic, Z.: Context-independent multilingual emotion recognition from speech signal. Int. J. Speech Technol. 6, 311–320 (2003). 3. Kleinginna Jr., P.R., Kleinginna, A.M.: A categorized list of emotion definitions, with suggestions for a consensual definition, Motivation Emotion. 5 (4), 345–379 (1981). 4. Ayadi, M. E., Kamel, M. S., Karray, F.: Survey on Speech Emotion Recognition: Features, Classification Schemes, And Databases. Pattern Recognition. 44, 572-587 (2011). 5. Moreno, P. J.: Emotion recognition from speech using prosodic features. Ph.D. thesis, University of Oulu Graduate School, Finland (2014). 6. Karathapalli, S.R., Koolagudi, S.G.: Emotion recognition using speech features, Springer Science+ Business Media New York (2013). 7. Bosch, L.: Emotions, Speech and the ASR Framework. Speech Communication. 40, 213–225 (2003). 8. Busso, C., Lee, S., Narayanan, S.: Analysis of emotionally salient aspects of fundamental frequency for emotion detection. IEEE Trans. Audio Speech Language Process. 17 (4), 582–596 (2009). 9. Utane , A.S., Nalbalwar, S.L.: Emotion recognition through Speech. International Journal of Applied Information Systems (IJAIS). 5-8 (2013). 10. Shen, P., Changjun, Z., Chen, X.: Automatic Speech Emotion Recognition Using Support Vector Machine. In: International Conference on Electronic and Mechanical Engineering and Information Technology, (2011).

120

ISBN: 978-969-8680-32-9


11. Bozkurt, E., Erzin, E, Erdem, C.E., Tanju Erdem, A.: Formant Position Based Weighted Spectral Features for Emotion Recognition. Science Direct Speech Communication. (2011). 12. Makhoul, J.: Spectral analysis of speech by linear prediction. IEEE Transactions on Audio and Electroacoustic. 21 (3), 140–148 (1973). 13. Huang, X., Acero, A., Hsiao-wuen hon.: Spoken Language Processing: A guide to Theory, Algorithm, and System Development. Prentice Hall, January (2001). 14. Snell. R.C., Milinazzo. F.: Formant location from LPC analysis data. IEEE Trans. Speech and Audio Processing. 2 (1),129-134 (1993). 15. Software tool PRAAT, http://www.fon.hum.uva.nl/praat/ 16. Ali., S.A., Zehra, S.: Performance Evaluation of Learning Classifiers for Speech Emotions Corpus using Combinations of Prosodic Features. International Journal of Computer Applications. 76 (2), 35-43 (2013). 17. Ali., S.A., Zehra, S.: Development and Analysis of Speech Emotion Corpus Using Prosodic Features for Cross Linguistic. International Journal of Scientific & Engineering Research. 4 (1), 1-8 (2013).

121

ISBN: 978-969-8680-32-9


Development of Gesture based Wireless Appliances Control System for Hospital Ward Tayyab Ahmad Ansari, Ahmed Hassan, Muhammad Faris Faculty of Engineering Science and Technology, Hamdard University, Karachi, Pakistan [email protected], [email protected], [email protected]

Abstract This paper outlines the design and development approach of a wireless gesture based system to control electrical appliances of hospital ward. The proposed system consists of a network of flex sensors as medium of input. These sensors are mounted on a glove that works as a medium to control and operate appliances. Through this system a patient can control numerous appliances of immediate needs e.g. adjusting bed level, switching light(s) and/or fan(s) on/off etc. The core aim of study involved behind this system is to provide more and more assistance to patients admitted in hospital and to work towards the advancement of facilities at the disposal of a patient in a particular ward. This system is principally targeted towards speech and/or hearing impaired or homebound patients. With the availability of this system, the life of a patient/user will be much more easy, comfortable and safe without too much dependence on personal attendant or helper. Finally, it has been tested to 15 patients with various movement gives satisfactory result.

Keywords: Home Automation, Gesture control, Wireless control

Introduction A Smart system is an assistive tool to facilitate in some basic daily activities safely, easily and comfortably [1]. A Patient in a hospital ward faces various difficulties regarding operating electrical appliances and assistance in case of emergencies and many others needs. An assistive technology is necessary for taking care of home or ward bound and on-bed patients. A smart system can solve almost all of the problems that are faced by the patients in a hospital. The purpose of study is to design a wireless assistive system that operates electronic devices using flex sensor. Through this research patient will stay at a hospital ward more easily, safely and comfortably without any attendant or nurse. Currently, it provides controlling of four basic parameters namely bed level adjustment, light and fan on/off control and alarm. Furthermore, with the advancement in this system, it can be designed to control television, telephone or many other devices. This flex sensor embedded glove can also be used as a medium for controlling assistive robots, room cleaning devices, operating computer and playing games. The primary objectives of this system are  To design a gesture monitoring glove using flex sensors.  By using the glove, design an assistive system for disabled, on-bed and speechless patients to control electrical appliances based on a particular gesture input.  It must be economical, cost effective, fully embedded and have minimum weight.

Related work Smart Home system is currently one of the most researched area worldwide. Researchers have employed numerous wireless techniques and controlling mechanisms to implement more and more efficient systems. Ramlee et al [2] designed a wireless Bluetooth and Android based smart home system for disabled people. They used software application in mobile phone through android's operating system to control the electrical appliances switches using Bluetooth. Lingling Li et al [3] designed a system to control and monitor home environment using ZigBee and Embedded Web Technology. In this system, 32bits-ARM9 processor is the brain of the system, and the Embedded Web Server Boa is designed by using Embedded Linux Operating System. It exchanges information between the Server and remote users through Common Gateway Interface Technology. And ZigBee wireless module exchanges information between information appliances and the Server. Through this, Smart Home Control System can be achieved. Chunjing Tao et al [4] designed a Environment control systems for disabled people; it take input using pneumatic switches, big key switches and Touching panel and control using infrared, X-10 type systems and radiofrequency.

122

ISBN: 978-969-8680-32-9


Jae-Myung Yoo et al [5] performed a study on a sensing system to control artificial arms. In this they used flex sensors for detecting signal in sensing system. In this system, flex sensors measures signal from different muscles in the arm. They differentiate the movements of the arm in to four different motions that are flexion, extension, pronation and supination. In the end, they test the system by performing experiments on healthy persons has various movement in their arm. Tanaka, T. et al [6] design a measurement device to measure doctor’s finger motion to control remote catheter. Through this study, they solved problems faced by a doctor during operation like exposure of hands and face to x-ray. These problems are solved by controlling remote catheter using measurement of finger motion. They used acceleration sensor and flex sensor for catheter insertion and rotation. Natasha Alves et al [7] designed and tested a switch for severely disabled people which is controlled by means of Mechanomyogram (MMG). Signals are generated from Frontalis muscle by movement of an eyebrow. The Signal is then isolated by applying wavelet transform algorithm and then used to control switch function. They also enhanced the system by reducing noise signals due to involuntary movements. Cued performance test was conducted on eleven patients out of whom one was severely disabled for manual switch operation.

Biomechanics of Hand A human hand contains 27 bones and comprises of carpometacarpal, intermetacarpal, metacarpophalengeal and interphalangeal joints. There are five fingers are attach to the hand which are referred to as digits from one to five. First digit is called thumb. Hand is unique part of human body. It can hold, grasp, pick, move and touching etc. The most important function is to use in communication and expression of emotions through gesture. It also provides large ranges of movements are abduction, adduction, extension, flexion and opposition as shown in fig. 1 and 2.

Fig 1: Movements of the thumb [8].

Fig. 1. Movements of Fingers [8] The range of motion [9] of metacarpophalengeal (MCP) and interphalangeal joint (IP) of the thumb is

The range of motion [9] of metacarpophalengeal (MCP) and distal interphalangeal joint (DIP) and proximal interphalangeal joint (PIP) of the fingers is

123

ISBN: 978-969-8680-32-9


System Design Approach System comprises of two units; Control Input unit and Actuation unit. Control Input unit contains a network of sensors, main control unit (for encoding user input) and wireless transmitter circuit. While the Actuation unit comprises of Receiver circuit, another control Unit (for decoding received data and generate control signals) and actuator unit. The most important and foundation step towards the implementation of the system is to devise a way to obtain control input by the user. It is necessary that the input mechanism must be both easy to operate and comfortable to wear/carry. Secondly, since the system is designed for patients and users with limited mobility, so an efficient wireless data transmission and receiving mechanism with efficient range and reliable accuracy must be included to ensure error free and comfortable operation of the system. Lastly, a fast, low cost and efficient Controller for streamlining and synchronizing all the processes. The selection of suitable controller is vital as it determines the overall performance of the system. The details of the above mentioned system components are discussed in detail as follows: Control Input Mechanism Following are various methods [10, 11] for sensing and/or extracting biomechanical signals from the user:  EMG (Electromyography)  MMG (Mechanomyography)  Load cell  Strain gauge tactile sensor  Flex Sensor  Sliding fiber optic cable  Wearable conductive fiber For the current system, we have decided to employ Flex Sensors as sensors for control input. It is found that use of flex sensor is a much better solution for detecting signal because these sensors are reliable, cost effective, economical solution. Flex sensors are resistive sensors. These sensors show changes in resistance with respect to change in bending angle. This system contains 5’’ long flex sensors that vary resistance from 10kΩ to 50kΩ. Δθ α ΔR Where ΔR is the change in resistance and Δθ is the change in bending angle. The Following fig. 3 highlights the final resistance of the Flex sensor at a particular bending angle. Relation b/w Bending Angle and Resistance 30 28 26

Bending Angle

24 22 20 18 16 14 12 10

0

10

20

30

40

50

60

70

80

90

Resistance

Fig. 2. Relation between bending angle and Resistance [10] Wireless communication system

124

ISBN: 978-969-8680-32-9


Many digital wireless transmitter and receiver systems are popular among embedded system designers. Following are most widely used systems for short range and acceptable data rates:  ZigBee  Bluetooth  Radio Controlled Transmitter and Receiver Radio controlled transmitter and receiver circuit is used in the system. The main advantage of this system is its low cost and availability. Normally such types of circuits are used in radio controlled (RC) toys [12]. These toys have a remote control unit (transmitter) and a device or toy (receiver). Remote Control unit has a maximum of 6 control inputs and correspondingly we have 6 outputs at the receiver. Different types of operating ranges are available. In this system, a fully functional controller is used which has four controls that can control appliances of the patient ward. Control signal is fed in to transmitter that transmits to the receiver. By using combinations of available control inputs, we can extend the no. of available control signals from 4 to 16. Main Controller Nowadays, with the advancement of Semiconductor fabrication technologies, it is possible to design a microcontroller based system which is cost efficient, lightweight and most importantly low power consuming. A large no. of companies have introduced single chip microcontrollers in the market aiming to reduce the size of the system and at the same time enabling more and more complex systems design possible. Some examples are as under:  8051 family (Atmel Corp.)  Advanced Virtual RISC (AVR) family (Atmel Corp.)  PIC Controller (Microchip Technology Inc.) For the under seen system, we are using AVR Atmega32 microcontroller from Atmel Corp. as the system main controller. This controller features all the necessary peripherals as per the requirement of the system. These are highlighted as below: 1. 6-channel ADC (To digitize Flex Sensor Analog Input) 2. UART (for serial data communication between transmitter and receiver modules) 3. Maximum 32 Input/output Pins (for interfacing LCD and to generate control outputs for actuators)

Hardware design Control Input Unit The block Diagram of Control Input Unit is shown in fig. 4: Input

Flex Sensors

Signal Conditioning

AVR ATMEGA32L

Transmitter

Output

Fig. 3. Block Diagram of Control Input Unit The whole unit is mounted on a glove as shown in fig. 7. The glove is worn by the patient/ user. The flex sensor continuously sends signals to signal conditioning unit as shown in fig. 5.

To controller Input

Fig. 4. Signal conditioning circuit Flex sensor is a resistive sensor. Change of resistance occurs with respect to bending angle of the sensor. Which means every gesture performed by the patient generates a unique control input which can be utilized. By simply reading gestures of each individual finger separately, we can have five distinct signals. By using

125

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 combinations, we can increase the amount of gestures as per requirement. For the current system we are using only 4 individual finger gestures. The formula for voltage divider circuit in fig. 5 is

The Following fig.6 highlights the final resistance of the Flex sensor at a particular bending angle. Relation b/w Output Voltage and Resistance 4

Output Voltage

3.5

3

2.5

2

1.5

0

10

20

30

40

50

60

70

80

90

Resistance

Fig. 5. Relation between Resistance and Output Voltage

Fig. 6. Gesture glove with embedded Flex Sensors (designed in BMI Lab, Hamdard University) Actutating Unit The block Diagram of Actuating Unit is shown in fig. 8: Patient Bed Adjusting motor Input (output of transmitter) RECIEVER

AVR ATMEGA32

Light RELAYS Fan Alarm or call for nurse

Fig. 7. Block diagram of gesture based recognition system In this unit, the received signal is fed into the microcontroller. The microcontroller decodes the received signal using predefined algorithm and generates appropriate control signals according to the received instruction. The control signals are sent to the target appliances using appropriate isolation circuit (Relays in present case) to protect sensitive controller circuitry.

System in operation and simulation in Labview The following Flowchart describes the working of system that illustrates in fig. 9 which is very much similar to R.A.Ramlee et al design home automation system [13].

126

ISBN: 978-969-8680-32-9


Start

Microcontroller scans the input (from Signal conditioning unit) No Is the finger flexed? Yes

Microcontroller collects the signal from receiver and activates specific Relays

Microcontroller collects and processes the signal & send to

Relays open or close the appliances Transmitter sends signal to receiver

Fig. 8. System flow chart In addition to Hardware implementation, Virtual demonstration of the system is also prepared and presented using LabVIEW. LabVIEW is a graphical programming language provides virtual environment of any required instrument [14]. LabVIEW has two windows front panel and block diagram. Control panel contains input and shows output. And block diagram contains complete programming of any system. In virtual demonstration of our system, front panel of the system takes input using binary button (flex sensors) and gives output using led’s (appliances) as shown in fig. 10.

Fig. 9. LabVIEW demonstration (Front panel) And block diagram comprises of sub VI (Virtual Instrument). The VIs consist of signal conditioning VI, transmitter VI and receiver VI. The input signals is fed in to Signal conditioning VI performs switching functions, Transmitter VI transmits the signal to Receiver VI and finally Receiver VI sends the signal to appliances of a ward as shown in fig. 11.

Fig. 10. LabVIEW demonstration (Block diagram)

127

ISBN: 978-969-8680-32-9


Result and Conclusion Finally, it has been tested to 15 patients with various movement. Each patients performs 50 trials in which of 28% trials were missed due to flexibility of patient’s fingers.

Since this system is easy to learn and handle; it is a useful tool for those patients who have to stay at home or in hospital wards alone and for those who are partially paralyzed. Through this system, patients can stay at home or hospital ward independently, as it eliminates the need of an attendant or nurse.

Future Advancement The number of appliances that can be controlled by a particular user can be enhanced by some of the many ways listed as under:  By connecting wireless communication system or videoconferencing system in the glove in order to contact their family.  By developing a smart wireless monitoring of vital signs. These vital signs include parameters like pulse rate, temperature, blood pressure, etc. of the user.  By using impact sensor to detect any accident to the user. These advancements may also contribute in early recovery of patients.

References 1. 2. 3.

4.

5. 6. 7. 8. 9. 10.

11. 12. 13. 14.

Warriach, E.U.: State of the Art Embedded Middleware Platform for A Smart Home. International Journal of Smart Home. Vol.7, (No.6): p. 275-294 (2013) Ramlee, R.A.: D.H.Z. Tang, and M.M. Ismail. Smart home system for Disabled People via Wireless Bluetooth. in System Engineering and Technology (ICSET), 2012 International Conference on (2012). Li, L., et al.: Design of Smart Home Control System Based on ZigBee and Embedded Web Technology, in Artificial Intelligence and Computational Intelligence, J. Lei, et al., Editors. Springer Berlin Heidelberg. p. 67-74 (2012) Tao, C., X. Zhang, and X. Wang.: Research of Environmental Control systems for Disabled people, in 7th AsianPacific Conference on Medical and Biological Engineering, Y. Peng and X. Weng, Editors. Springer Berlin Heidelberg. p. 476-479 (2008) Jae-Myung, Y. and A. Yong-Myung.: A study on a sensing system for artificial arm's control. in SICE-ICASE. International Joint Conference (2006) Tanaka, T., G. Shuxiang, and X. Nan.: Development of a doctor's finger motion measurement device for a remote catheter operating system. in Mechatronics and Automation (ICMA). IEEE International Conference on. (2013) Alves, N. and T. Chau.: The design and testing of a novel mechanomyogram-driven switch controlled by small eyebrow movements. Journal of NeuroEngineering and Rehabilitation, 2010. 7(1): p. 1-10. Hall, S. J.: Basic biomechanics. New York, NY, McGraw-Hill. (2012) www.eatonhand.com/nor/nor002.htm Ahmed, S.F., S.M.B. Ali, and S.S.M. Qureshi.: Electronic Speaking Glove for speechless patients, a tongue to a dumb. in Sustainable Utilization and Development in Engineering and Technology (STUDENT), IEEE Conference on. (2010) Bhatti, N.P., et al.: Electronic hand glove for speech Impaired and Paralyzed Patients. EJR Magazine. p. pp. 59-63 (May 2009) Available: http://electronics.howstuffworks.com/rc-toy1.htm R.A.Ramlee, et al.: Bluetooth Remote Home Automation System Using Android Application. The International Journal of Engineering and Science, 02(01): p. 149-153 (2013) National Intruments: http://www.ni.com/labview

128

ISBN: 978-969-8680-32-9


Analyses of Intercultural Challenges in Offshore Software Development Outsourcing Relationships based on Continent and Company Size Muhammad Ilyas Azeem1, Siffat Ullah Khan1 and Abdul Wahid Khan1,2 1 Department of Computer Science and IT, University of Malakand, Pakistan. 2 Institute of Engineering & Computing Science, University of Science & Technology, Bannu, Pakistan. [email protected],[email protected], [email protected]

Abstract Offshore software development outsourcing (OSDO) is a strategy for software development adopted by many organisations around the world. The increasing popularity of this approach has been facilitated by a number of technical and commercial factors. These include low development cost, high software quality, round-the-clock development, access to well skilled workers and modern communication technologies etc. Besides these driving factors there are many issues, including temporal, geographical and intercultural differences that impedes the realization of OSDO benefits. The objective of this research is to identify the intercultural challenges faced by vendors in OSDO relationships. We have used a systematic literature review (SLR) process for the identification of these challenges from vendors’ perspective. We have further analyzed these challenges based on continents and company size. Keywords: Intercultural challenges, Systematic literature Review, offshore software development outsourcing.

Introduction Offshore software development outsourcing is a strategy for software development that is dispersed across different cultural, temporal and geographical locations around the world. “Offshore software development outsourcing describes the transfer of information system services (business process) from one country to another” [1]. Using this strategy client organisations obtain services from remote organisations called vendor or service provider. In India software outsourcing has become one of the most successful strategies for its businesses and India is known as a leader in delivering high end software outsourcing services [1, 2]. The worldwide use of offshore software development outsourcing is because of the advantages it provides to the software outsourcing clients and vendors. Cost is always a primary factor behind the adoption of offshore software outsourcing. For the last decade outsourcing of software projects to remote countries, that offer lower development costs and higher availability of resources, have beenincreased[3]. Client organisations benefit from offshore software development outsourcing by reducing their expenses and getting high quality software products [4]. A recent study [5] mentioned two main reasons for the adoption of offshore software development outsourcing: Firstly, increased personnel requirements due to increasing size of software projects, secondly increasing maintenance costs and limited availability of skilled workers at onshore sites. Effective time-zone utilization enables client organisations to develop large software products in short schedules. This can provide exactly round the clock development which can lead towards improved performance [6]. Similarly access to the latest technology, available at vendor organisations is another major reason for offshore software outsourcing [7]. Offshore software outsourcing is not a risk free activity and a number of projects failures have been reported. Geographical distance introduces physical distance among the client and vendor organisations which makes face to face meeting difficult [8]. Time difference between two sites make efficient communication difficult and increases coordination costs [9]. Communication is the backbone of offshore software outsourcing but it is negatively affected by geographical, temporal and cultural distance [10, 11]. OSDO also faces problems related to coordination and control when dividing and assigning the project work [12]. In addition to the non-technical issues in Global Software Development (GSD), one of the most technical issues is the integration of software components developed by different teams at different locations [13]. Besides the aforementioned problems cultural distance is another challenge in OSDO relationships [10], [14]. The importance of understanding culture and the key role it plays in the software industry has substantially increased over the last twenty years [15]. Cross-cultural incompatibilities and miscommunication hinder the effectiveness of software outsourcing [4]. Sixty nine percent of all the outsourcing projects failed either completely or partial due to intercultural incompatibilities among the client and vendor organisations and poor relationship management [4]. Similarly team corporation among members may be reduced due to cultural and language difference resulting in misunderstanding [9]. These

129

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 intercultural challenges are needed to be overcome to realize the benefits of OSDO. Thus it is important to study OSDO relationships from intercultural perspective. We have published initial results of the systematic literature review (SLR) in the journal of IET Software [16]. This is the extended version of the paper published in IET Software [16]. Similarly we have other publications from this research project [17, 18]. This paper presents further analyses of all the intercultural challenges based on continents and company size, to find what has been done and what to be done by generating various inferences. Our long term goal is to develop Intercultural Challenges Mitigation Model (ICCMM) to assist OSDO vendors in addressing intercultural challenges in OSDO relationships. In this paper the following research questions (RQs) have been answered. RQ1. How are the intercultural challenges, identified through the SLR, related to the company size? RQ2. Do the intercultural challenges, identified through the SLR, vary from continent to continent?

Background Culture is difficult to define and this is the reason that there are limited definitions of culture availablein global software engineering research community [4]. Culture is a very vast topic and is defined in many ways in the literature [19]. A well-known Sociologist and Anthropologist Sir Edward Tylor [20] defines culture as “Culture is that complex whole which includes knowledge, belief, art, morals, law, customs and any other capabilities and habits acquired by man as a member of society.” From this definition it is clear that culture has a great impact on almost all aspects of our life. It affects activities involve in software development outsourcing projects as they require a level of cooperation and coordination that cannot ignore the impact cultural diversity plays and the barriers and misunderstandings it can and does create [21]. Vendors need to understand the culture of their target clients and its impact on software development lifecycle and their relationships [14], [22]. Developers must focus on the culture and linguistic requirements of the client’s countries and local groups including developers, customers, clients, and end-users [23]. Because language differences between development sites can result in misinterpretation and unconveyed information [24]. OSDO encompasses people from various geographical locations and cultural backgrounds. This increases the need for research on the impact of national and cross cultural issues [25]. In our previous study we identified culture as a critical factor for vendor organisations in OSDO relationships [14, 26]. The significance of culture in OSDO cannot be ignored. This is because cultural issues can “make and break an offshore project” [27]. The following studies mentioned the importance of intercultural challenges in the OSDO relationships: Ita et al,[10] in their work on a “Process Framework for Global Software Engineering Teams”, identified culture as an important issue to be addressed and recognized to successfully manage distributed teams across different cultures. If cultural diversity is not given proper attention and measures taken to address them it can have serious negative repercussions on the operation of global teams. Surveys conducted at US and Jordan reveals that culture plays an important role in affecting software piracy, and individual behavior in general. Results of the study show that tendency towards software piracy for gender varies from culture to culture [28]. A recent study [29] reveals that intercultural challenges play significant role in communication process in outsourcing relationships. Communication problems that arise in outsourcing relationship include differences in corporate culture and differences in linguistic and rhetorical choices. Issues causing these problems include differences in education and training.Bannerman et al, [30] identified four coordination challenges namely temporal, geographical and socio-cultural distance in global software development. Socio-cultural challenge has impact on the effective coordination due to inconsistent work practices and reduced cooperation arising from misunderstanding. Gopal et al, [12] used a structured survey approach for collecting data from leading Indian software outsourcing firms in their research on coordination and performance in GSD. The result suggests that cultural differences between the vendor team and the client will lead to some misalignment in expectations about the project and less mutual understanding, and lower software quality. Abufardeh and Megal,[23] focused on the impact of cultural and linguistic aspects of global software development. They found that multi-lingual and multi-cultural aspects are very critical to the success of the GSD projects. These aspects affect directly the quality of the software, as well as the GSD process itself. It is argued that cultural influences must be thoroughly studied, fully understood and should be considered throughout the software development lifecycle. Koh et al,[31] study indicates that conflict is inevitable in teams, and cultural misunderstandings often contribute to more disagreements and disputes, as members fromdifferent cultures handle conflict differently. Similarly language differences and differences in communication styles can lead to misunderstanding. The literature discussed above described the impact of intercultural challenges on the different aspects of offshore software development outsourcing projects. The work in this paper is an extension to the SLR findings previously conducted on the identification of the intercultural challenges in outsourcing relationship

130

ISBN: 978-969-8680-32-9


from vendor’s perspective [16]. The results of this paper will help the vendor organisations to address the intercultural challenges that they may face in OSDO relationships. The findings may assist client organisations in addressing the identified challenges that may lead to successful outcomes of the outsourced project.

Research Methodology We have performed a systematic literature review (SLR) for identifying the intercultural challenges from the literature, following the guidelines by Kitchenham and Charters [32].Initially we developedSLR protocol for conducting the study and was published [17, 18]. Through implementation of the SLR protocol a total of 11 intercultural challenges were identified amongst which ‘language and language proficiency’, ‘national culture’, and ‘organisational culture’ were marked as critical intercultural challenges for vendors in OSDO relationships. Further details of these intercultural challenges and the various steps in the SLR methodology are given in our previously published paper in IET software[16].

Results In this section, we discuss the results and analyse the identified intercultural challenges for each of the Research Questions as mentioned in Section 1. The details are given in the following subsections.

Intercultural challenges for OSDO vendors based on company size Our sample size is 60 papers; however, only 20 papers have mentioned the company size as shown in Table 1. Using company size definition provided by Australian Bureau of Statistics [33], we divided papers of these companies into three categories: SMALL (0 to 19 employees), MEDIUM (20 to 199 employees), and LARGE (200 + employees). All the intercultural challenges have been reported in Large-sized companies according to our findings. Amongst these 11 intercultural challenges in total, 3 challenges have been cited in ≥30% of the articles. These three intercultural challenges are ‘national culture’ – 55%, ‘language and language proficiency’ – 51% and ‘organisational culture’ – 33%. It is worth noting that the intercultural challenge ‘language and language proficiency’ has the highest percentages (64%) for large companies. Our analysis indicates that large-sized companies should focus on the ‘language and language proficiency’ to maintain there long lasting relationships with their clients. Previous literature also mentioned ‘language and language proficiency’ as a critical intercultural challenge for large-sized companies [34, 35]. We found two intercultural challenges for medium-sized companies in the literature. These intercultural challenges have been cited in ≥50% of the articles. These two intercultural challenges have highest percentage (100%) of occurrence in the medium-sized companies. These intercultural challenges are ‘language and language proficiency’ and ‘communication style’. Our results indicate that the aforementioned intercultural challenges have a significant impact on medium-sized companies. We suggest medium-sized companies to address the aforementioned intercultural challenges to succeed in their relationships with their clients in OSDO. For small-sized companies we found four intercultural challenges in the literature as shown in Table 1. Two intercultural challenges among these intercultural challenges has been cited in 50% of the articles. ‘Organisational culture’ and ‘difference in mutual understanding’ have a highest percentage (100%) of occurrence for small sized software development outsourcing vendors. Intercultural challenges ‘communication style’ and ‘national culture’ have 50% occurrence for small sized software outsourcing vendors. Small-sized companies needs to address these four intercultural challenges to compete with their other counterparts in OSDO. Similar to our results literature also identified ‘organisational culture; and ‘communication style’ as critical intercultural challenges for small-sized companies [36].Our results shows that ‘language and language proficiency’ is a critical challenge for both large and medium size companies but for small company size its frequency is 0.However due to small sample size for small, medium and large sized companies, as mentioned in Table 1, it is difficult to generalize these findings. Further it needs empirical validation in the industry, which we plan to do in future.

131

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 We found only one significant difference amongst the intercultural challenges i.e. ‘communication style’ based on company size. We argued that the intercultural challenge ‘communication style’ has a great impact on large sized software development vendors in outsourcing relationships. We have used Linear-by-Linear Chi-Square test for the identification of statistically significant difference amongst the various company sizes. As the data was of ordinal nature, the linear by linear association Chisquare test was used in order to find whether there is any statistically significant difference among the identified intercultural challenges based on company size. The linear by linear association test is preferred when testing the statistically significant difference between ordinal variables because it is more powerful than Pearson chi-square test [37]. The following hypotheses were examined. Table 1 Distribution of intercultural challenges based on company size

Null hypothesis (H0): There is no difference among the company sizes used to reveal the intercultural challenges. Alternative hypothesis (H1): There is a significant difference among the company sizes used to reveal the intercultural challenges. If the value of ‘p’ is greater than 0.05 for an intercultural challenge then we will consider null hypothesis (H0) otherwise alternative hypothesis will be considered. If there is no significant difference for any particular intercultural challenge it means that there is no big difference in citation reported in different company sizes for that challenge. If there is statistically significant difference for a particular intercultural challenge it means that this challenge is not equally cited by different company sizes e.g. ‘communication style’ is reported with a frequency of 50%, 100% and 35% in small, medium and large size company respectively which shows a clear difference in the citation of this challenge by the three company sizes. This is the reason that the Chi-Square test produces the value of ‘p’ less than 0.05. Table 1 reveals statistical significance for only one intercultural challenge (‘communication style’) as the value of ‘p’ is less than 0.05. We therefore reject null hypothesis (H0) and accept alternative hypothesis (H1) for this intercultural challenge. The alternative hypotheses for this case is formulated as follow: 1. The intercultural challenge ‘communication style’ is revealed mainly with medium size company than with other company sizes. Intercultural challenges for OSDO vendors across various continents Our results show the number of articles reporting studies related to different continents. We have grouped the papers found through SLR into different continents, i.e. ‘Asia, Europe’, ‘North America’, ‘Not mentioned’, and ‘Mixed’, where the original study was conducted. By the term ‘Mixed’ we mean studies conducted across multiple continents like ‘North America and Asia’, ‘Europe and Asia’, ‘Europe, Asia and North America’, ‘Europe, Asia and South Africa’ and ‘Europe and North America’. Our aim is to find whether these factors differ from continent to continent. We suggest that understanding the similarities and differences in these factors can contribute to the body of intercultural challenges knowledge in offshore software development outsourcing relationships. It should be noted that in the SLR we did not put any continents boundaries on our search. However, we found the papers across different continents. Moreover the category ‘not mentioned’-18.3%, as shown in Table 2, in our study shows the papers where the location of the study has not been reported. Other categories in our analysis are ‘Europe’- 10%, ‘North America’- 11.6% and ‘Asia’-16.6% have been reported on intercultural challenges in the context of outsourcing. This may be a reason that most of the vendor’s in the GSD industry are from Asian countries and clients organisation from North America.

132

ISBN: 978-969-8680-32-9


Comparison of the intercultural challenges identified in three continents indicates that there are more similarities than differences among the intercultural challenges. We did not find any significant differences among the three continents as shown in Table 2. Our findings show that ‘language and language proficiency’ (60%, 57%) is the most common intercultural challenge across the two continents Asia and North America. This indicates that most vendor organisations in Asia and clients organisations in North America face ‘language and language proficiency’ problems. Asian countries like China and India have high interaction distance due to language proficiency issues [38]. ‘Organisational culture’ (66%) and ‘national culture’ are the most common intercultural challenges in Europe. Intercultural challenges (‘communication style’, ‘national culture’ and ‘difference in social behavior’) and (‘organisational culture’ and ‘work ethics’) are the second and third most commonly cited in Asia respectively. It means Asian vendor organisations may face problems regarding ‘communication style’, ‘national culture’ and ‘difference in social behavior’ of their clients. They need to educate their staff members about the ‘national culture’, ‘organisational culture’ and ‘work ethics’ etc. of their target clients in order to overcome these intercultural challenges in OSDO relationships. In complement to our results previous studies mentioned the aforementioned intercultural challenges critical for Asia e.g. [39]. (‘Language and language proficiency’, ‘response to time’, ‘difference in mutual understanding’ and ‘difference in social behavior’) and ‘communication style’ are the second and third most commonly cited intercultural challenges in Europe respectively. Similar to our analysis some of the previous studies mentioned these challenges critical for Europe e.g. [15], . Similarly (‘national culture’ and ‘communication style’) and (‘organisational culture’, ‘response to time’ and ‘difference in mutual understanding’) are the second and third most commonly cited intercultural challenges in North America respectively. Previous literature also mentioned these challenges as critical for North America e.g. [40]. The following hypotheses were examined. Null hypothesis (H0): There is no difference among the continents used to reveal the intercultural challenges. Alternative hypothesis (H1): There is a significant difference among the continents used to reveal the intercultural challenges. Our finding indicates that there are more similarities than differences in intercultural challenges for different continents. We do not find any alternative hypothesis for the intercultural challenges based on continents. Table 2 Distribution of intercultural challenges based on continents

Summary and Discussion Through the SLR we have identified various intercultural challenges faced by vendor organisations in OSDO relationships. Intercultural challenges present some of the key areas which need management’s attention to develop OSDO initiatives. We plan to find the practices for addressing these challenges in future through empirical study. In order to decide the importance of a factor, we have used the following criterion: 

If an intercultural challenge is cited in the literature with a frequency percentage of >=30% then we treat that factor as a critical factor in this explorative study

We have used the similar criterion in our previous research [14], [26], [41, 42] . However, software outsourcing practitioners can define their own criteria in order to decide the criticality of listed intercultural challenges in OSDO relationships. For RQ1, using the criterion for intercultural challenges, we have

133

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 identified 4 critical intercultural challenges for each company size i.e. small, medium and large respectively. These include  ‘Organisational culture’, ‘difference in mutual understanding’, ‘communication style’ and ‘national culture’ is identified as critical intercultural challenges for small sized companies.  ‘Language and language proficiency’, ‘national culture’, ‘communication style’, and ‘response to time’ are critical challenges for medium sized companies.  For large sized companies ‘language and language proficiency’, ‘national culture’, ‘communication style’ and ‘difference in social behavior’ are identified as critical intercultural challenges.  ‘Language and language proficiency’ is identified as critical for both large sized and medium sized companies.  ‘National culture’, ‘communication style’ and ‘organisational culture’ are critical for all the company sizes i.e. small, medium and large. We found only one significant difference among the intercultural challenges across different company size. That intercultural challenge is communication style as shown in Table 1. In order to address RQ2, we identified no significant difference among the 11 intercultural challenges across various continents. The critical challenges identified in various continents are discussed as follow:  The intercultural challenges identified as critical for Asia are ‘language and language proficiency’, ‘national culture’, ‘communication style’ and ‘difference in social behavior’.  ‘National culture’, ‘language and language proficiency’, ‘organisational culture’, ‘response to time’, ‘difference in mutual understanding’ and ‘difference in social behavior’ are identified as critical for Europe.  Only one intercultural challenge, ‘language and language proficiency’ is identified as critical for North America.  ‘Language and language proficiency’ is critical for Asia, Europe and North America. It means that OSDO organisations across these continents should focus on this intercultural challenge.  ‘National culture’ is critical for both Asia and Europe. Comparison of the intercultural challenges identified in three continents indicates that there are more similarities than differences between the intercultural challenges in Asia and Europe. Table 2 shows that 10 intercultural challenges are cited in Asia, 7 intercultural challenges in North America and 7 intercultural challenges in Europe.

Conclusion and Suggestions for Future Work We have analysed, based on company size and continents, the identified intercultural challenges faced by vendors in outsourcing relationships. We identified 4, 2 and 11 intercultural challenges for small, medium and large sized companies respectively as shown in Table 1. Similarly we identified 10, 7 and 6 intercultural challenges for Asia, Europe and North America respectively as shown in Table 2.We suggest that focusing on these challenges can help OSDO vendors to strengthen their relationships with the client organisations in offshore software development outsourcing. From the findings of this study, we have identified the following goals that we plan to follow in future:  Validate these intercultural challenges using empirical studies with practitioners working in outsourcing industry as conducted by other researchers [43, 44].  Analyze the challenges/risks in OSDO relationships from clients’ perspective.  Conduct empirical studies to determine practices for addressing the identified critical intercultural challenges. Our ultimate aim is to develop an Intercultural Challenges Mitigation Model (ICCMM). This paper contributes to only one component of the ICCMM, i.e. the identification of the intercultural challenges via SLR. The eventual outcome of the research is the development of ICCMM to assist OSDO vendors in managing their relationships with client organisations efficiently.

References 1. 2. 3.

Kumar, S.A. and A.K. Thangavelu. Factors Affecting the Outcome of Global Software Development Projects: An Empirical study. in International Conference on Computer Communication and Informatics (ICCCI -2013). 2013. Coimbatore, INDIA: IEEE. Li, F., et al. Self-Adapting Task Allocation Approach for Software Outsourcing Services. in Service Operations and Logistics, and Informatics (SOLI), 2012 IEEE International Conference on. 2012. Suzhou: IEEE. Wende, E. and T. Philip. Instant Messenger in Offshore Outsourced Software Development Projects: Experiences from a Case Study. in System Sciences (HICSS), 2011 44th Hawaii International Conference on. 2011. Kauai, HI.

134

ISBN: 978-969-8680-32-9


4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.

Shah, H., et al. Studying the Influence of Culture in Global Software Engineering: Thinking in Terms of Cultural Models. in 4th international conference on Intercultural Collaboration (ICIC '12). 2012. Bengaluru, India. Nidhra, S., et al., Knowledge transfer challenges and mitigation strategies in global software development—A systematic literature review and industrial validation. International Journal of Information Management, 2013. 33: p. 333-355. Cho, J., Globalization and Global Software Development. Issues in Information Systems, 2007. 8(2). Khan, S.U. Software Outsourcing Vendors’ Readiness Model (SOVRM), PhD thesis, in School of Computing and Mathematics. 2011, Keele University UK. p. 381. Jaanu, T., M. Paasivaara, and C. Lassenius. Effects of four distances on communication processes in global software projects. in Empirical Software Engineering and Measurement (ESEM), 2012 ACM-IEEE International Symposium on. 2012. Lund. Bannerman, P.L., H. Emam, and J. Ross. Scrum Practice Mitigation of Global Software Development Coordination Challenges: A Distinctive Advantage? in 45th Hawaii International Conference on System Sciences. 2012. Maui, HI. Ita, R., et al., A Process Framework for Global Software Engineering Teams. Information and Software Technology, 2012. 54(11): p. 1175-1191. Palacio, R.R., et al., Selective availability: Coordinating interaction initiation in distributed software development. IET Software, 2012. 6(3): p. 185-198. Gopal, A., et al., Coordination and Performance in Global Software Service Delivery: The Vendor’s Perspective. IEEE Transaction on Engineering Managment, 2011. 58(4): p. 772 - 785. Ilyas, M. and S.U. Khan. Software integration model for Global Software Development. in Multitopic Conference (INMIC), 2012 15th International 2012. Islamabad: IEEE. Khan, S.U., M. Niazi, and R. Ahmad, Barriers in the Selection of Offshore Software Development Outsourcing Vendors: an Exploratory Study Using a Systematic Literature Review. Information and Software Technology, 2011. 53(7): p. 693-706. Casey, V., Imparting the Importance of Culture to Global Software Development. ACM inroads, 2010. 1(3): p. 5157. Khan, S.U. and M.I. Azeem, Intercultural Challenges in Offshore Software Development Outsourcing Relationships: An Exploratory Study Using a Systematic Literature Review. IET Software, 2014. 8(4): p. 161-173. Azeem, M.I. and S.U. Khan. Intercultural Challenges in Offshore Software Development Outsourcing Relationships: A Systematic Literature Review Protocol. in 5th Malaysian Conference in Software Engineering (MySEC). 2011. Johor Bahru, Malaysia: IEEE. Azeem, M.I., et al., Intercultural Challenges in Offshore Software Development Outsourcing Relationships: A Systematic Literature Review Protocol with Preliminary Results. International Journal of Advanced Science and Technology, 2012. 46: p. 175-185. Zakour, A. Cultural Differences and Information Technology Acceptance. in 7th Annual Conference of the Southern Association for Information Systems. 2009. Tylor, S.E.B., Primitive Culture: Researches Into the Development of Mythology, Philosophy, Riligion, Art, and Custom. Vol. 1. 1871, London: John Murray, Albemarle Street: Bradbury, Evans, and Co., Printers, Whitefriars. Herbsleb, J. and D. Moitra, Global Software Development. IEEE Software, 2001. 18(2): p. 16-20. Khan, S.U., M. Niazi, and R. Ahmad, Empirical Investigation of Success Factors for Offshore Software Development Outsourcing Vendors. IET Software, 2012. 6(1): p. 1-15. Abufardeh, S. and K. Megal. The Impact of Global Software Cultural and Linguistic Aspects on Global Software Development Process (GSD): Issues and Challenges. in New Trends in Information Science and Service Science (NISS), 4th International Conference. 2010. Gyeongju. Verner, J.M., et al., Risks and risk mitigation in global software development: A tertiary study. Information and Software Technology, 2014. 56(1): p. 54-78. Niazi, M., et al., Establishing trust in offshore software outsourcing relationships: an exploratory study using systematic literature review. IET Software, 2013. Khan, S.U., M. Niazi, and R. Ahmad, Factors Influencing Clients in the Selection of Offshore Software Outsourcing Vendors: An Exploratory Study Using a Systematic Literature Review. The Journal of Systems and Software, 2011. 84(4): p. 686-699. Gupta, U. and V. Raval, Critical Success Factors for Anchoring Offshore Projects. Information Strategy: The Executive's Journal, 1999. 15(2): p. 21-27. Setterstrom, A.J., P. John M., and A. Hassan. An Exploratory Examination of Antecedents to Software Piracy: A Cross-Cultural Comparison. in 45th Hawaii International Conference on System Sciences. 2012. Maui, HI. Raju, R., Intercultural Communication Training in IT Outsourcing Companies in India: A Case Study. IEEE TRANSACTIONS ON PROFESSIONAL COMMUNICATION, 2012. 55(3): p. 262-274. Wen, J., et al., Systematic Literature Review of Machine Learning Based Software Development Effort Estimation Models. Information and Software Technology, 2012. 54: p. 41-59. Koh, C., D. Joseph, and S. Ang. Cultural Intelligence and Collaborative Work: Intercultural Competencies in Global Technology Work Teams. in International Workshop on Intercultural collaboration (IWIC'09). 2009. Palo Alto, California, USA: ACM. Kitchenham, B. and S. Charters, Guidelines for performing Systematic Literature Reviews in Software Engineering. 2007, Keele University UK. p. 65. Trewin, D., Small Business in Australia: 2001. Australian Bureau of Statistics report 1321.0. 2002, http://www.ausstats.abs.gov.au/Ausstats/subscriber.nsf/0/C639A01ED725ADABCA256C54000336D1/$File/13210 _2001.pdf, June 2008. Sarker, S. and S. Sarker, Exploring Agility in Distributed Information Systems Development Teams: An Interpretive Study in an Offshoring Context. Information Systems Research, 2009. 20(3): p. 440-461.

135

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 35. Huang, H. and E.M. Trauth. Cultural Influences and Globally Distributed Information Systems Development: Experiences from Chinese IT Professionals. in Proceedings of the ACM SIGMIS CPR '07. 2007. St. Louis, Missouri, USA: ACM. 36. Boden , A., et al. Knowledge Management in Distributed Software Development Teams – Does Culture Matter? in 4th International Conference on Global Software Engineering. 2009. Limerick, Ireland: IEEE. 37. Martin, B., An Introduction to Medical Statistics. 3rd edition ed. 2000: Oxford Medical Publications. 38. Teagarden, M.B., J. Meyer, and D. Jones, Knowledge Sharing Among High-Tech MNCs in China and India: Invisible Barriers, Best Practices and Next Steps. Organizational Dynamics, 2008. 37(2): p. 190-202. 39. Ramingwong, S. and A.S.M. Sajeev. Influence of Culture on Risks in Offshore Outsourcing of Software Projects: A Quantitative Study on Mum Effect. in IEEE. 2010. 40. Ali-Babar, M., J. Verner, and P. Nguyen, Establishing and Maintaining Trust in Software Outsourcing Relationships: An Empirical Investigation. The Journal of Systems and Software, 2007. 80(9): p. 1438–1449. 41. Niazi, M., D. Wilson, and D. Zowghi, Critical Success Factors for Software Process Improvement: An Empirical Study. Software Process Improvement and Practice Journal, 2006. 11(2): p. 193-211. 42. Niazi, M. and M. Ali-Babar, De-motivators for software process improvement: an empirical investigation. Software Process Improvement and Practice Journal, (Perspectives on Global Software Development: special issue on PROFES 2007), 2008. 13(3): p. 249–264. 43. Niazi, M. An instrument for measuring the maturity of requirements engineering process. in The 6th International Conference on Product Focused Software Process Improvement, LNCS, Oulu, Finland, June 13-16. 2005. 44. Niazi, M., K. Cox, and J. Verner, A Measurement Framework for Assessing the Maturity of Requirements Engineering Process. Software Quality Journal: in press for publication, 2008. 16(2): p. 157-298.

136

ISBN: 978-969-8680-32-9


SDR Design of Reconfigurable Transceiver for Rapid Implementation on DSP Platform Syed Shabeeh Ul Husnain1, Muhammad Mustaqim2, Bilal Muhammad Khan3 and Bilal A. Khawaja4, Member IEEE Electronic and Power Engineering Department, PN-Engineering College (PNEC), National University of Sciences and Technology (NUST), Karachi, Pakistan 1 [email protected],[email protected], [email protected] [email protected]

Abstract In this paper, a software defined radio (SDR) baseband system design is presented. This design is implemented for real time operation on low-cost DSP platform. The modeled baseband system will be reconfigurable using software blocks with added advantage of hardware prototyping and reprogramming ability of Simulink. Texas Instrument’s digital signal processor (DSP) C6000 platform is used in this implementation. By using Simulink and DSP blocks together, rapid prototyping and faster implementation of the system is achieved. Keywords: Software defined radio (SDR), Digital signal processor (DSP), Simulink, mathematical model.

Introduction Modern modeling techniques (using mathematical models of different systems) have enabled the designers to rapidly design a system. The functionality of the system can be changed by connecting different code blocks. One of such powerful tool known to us is Matlab and its Simulink block sets [2]. Software defined radio (SDR) [1-8] technology offers broad range of advantages over conventional radio systems like flexibility, programmability, reusability [10,21]. It also replaces traditional implementation methods for radio components of transmitters and receivers. Design adaptability, multi-functional modes of operation, reconfigurability, selectivity in frequency bands, air interfaces, and waveforms are the main deliverables of SDR technology [1]. Thus, an SDR must use a reprogrammable platform like digital signal processor (DSP) to build an open architecture with coding, modulation, decoding and demodulation that can be implemented in the software [2]. The use of digital signal processing for baseband operations provides design power and flexibility to the designers allowing wide variety in choosing coding and modulation algorithms/techniques for the design [3]. In SDR implementation, the conventional hardware circuitry to implement typical communication system blocks has been replaced by mathematical models [1]. This type of implementation has voided the hardware components at the baseband and pass band levels. So, now a radio can be implemented if there is an availability of a processing platform and the RF front-end. Unlike conventional radios, the RF front-end and baseband processing hardware decide the working specifications of the system. This technology is a revolutionary advancement in the area of communications. Most of the software defined components in an SDR especially waveform application part requires signal processing techniques and mathematical manipulation of digital data. So, a DSP platform is ideal for implementation [4]. The SDR discussed in this paper is modeled on Matlab Simulink and is implemented and tested on Texas instrument’s C6000 series digital signal processor [5-6]. Conventionally, a digital communication system consists of a signal/message source having infinite bandwidth with a certain data-rate [7]. The SDR technology defines the parameters and functions of the transceiver in the software, including frequency of the carrier, bandwidth of modulation, coding and space/time/frequency/code agility. These were seldom selectable in the previous radio generations and were defined using hardware blocks [8-9]. The implementation presented in this paper uses convolution encoding and 16-level amplitude modulation (QAM) in which each symbol is defined by 4-bits. The similar scheme is followed at the receiving end where initially, the symbols are de-mapped, then channel decoder corrects any errors and then the source decoder adds the extracted redundancy. In this way, the received data is available to use. This process is depicted by the model presented in Fig. 1 [1], [9-10].

137

ISBN: 978-969-8680-32-9


Transmitter

Receiver

Fig. 1. Typical communication system blocks

The modeled system in Simulink can be linked with the supported DSP hardware using available Simulink and Texas Instrument blocks. The whole process is so powerful that it generates the firmware for the properly configured hardware [11-13]. This way of using the tools provide faster implementation benefits to the designer. The rest of the paper is structured as follows. The modulation scheme used for implementation is discussed in section 2. Section 3 shows the selected SDR blocks that were used for Simulink modeling. Finally, the design and simulation results are discussed in Section 4. Section 5 draws the conclusion on the basis of the results. References are listed in section 6.

Modulation Scheme: 16-QAM The 16-QAM scheme makes use of the fact that the baseband data in terms of bits can be mapped onto both amplitude and phase of the carrier waves. The main benefit of the QAM scheme as compare to phase shift keying technique is that higher data-rates can be achieved using smaller bandwidth [14]. Two orthogonal waves are used called I and Q channels and one symbol is formed by grouping four-bits together. The number of symbols in the constellation diagram is governed by 2M, where M = 4 in 16-QAM [15]. The Fig. 2 from [7] shows the I & Q constellation diagram for the complex envelope of the symbols in 16-QAM.

Fig. 2. Square constellation for 16-QAM

In Fig. 2, the first two boxed bits shows the quadrant wise gray codes [15] reflecting the phase change. Similarly, the remaining bits in Fig. 2 are gray coded from each other within the quadrant reflecting amplitude changes. The symbol ‘d’ refers to the distance between the symbols. The signal and mapping coordinates for this modulation scheme are governed by the relationships given in Eq. (1) - (5):

138

ISBN: 978-969-8680-32-9


(1)

(2)

(3)

(4)

─

Where

represents the energy of the signal in the carrier waves.

─

, and

─

(5)

The constellation for 16-QAM maps according to the values from equation (5) in I & Q plane.

Implementation of SDR on Simulink The SDR implementation is done using Simulink blocksets. Simulink as a tool provides a very convenient modeling environment for the implementation of dynamic systems [17-18]. Its advantages include the ability to simulate linear, nonlinear, discrete-time, continuous-time, and multi-rate systems [16]. In Simulink available blocksets and user-defined functions are used to make models of the desired system [17-18], which are based on the mathematical descriptions. A radio baseband system may be designed, constructed and operated using an SDR architecture which is comprehensive, and in which used components are clearly assigned and related to their functions [10].

139

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 Similarly, in this paper the implemented SDR is using higher order 16-QAM modulation / demodulation blocks [14] for channel coding/decoding, convolution encoder and Viterbi decoder is used [10, 13]. It is important to understand that the main part of the SDR is the baseband processing stage which is in bitformat i.e. it is digital in nature for input and output signals [1, 10]. In real hardware for providing digital input to the system, DIP switches are being used and for the output, LED block is being utilized [11]. Hence, in this way SDR implementation has been carried out in Simulink. Simulink Models The SDR model in Simulink has two main parts i.e. transmitter and the receiver [1]. To develop a certain model in Simulink, it is necessary to identify the blocksets from which required blocks can be selected for use [17, 19]. Moreover, it is also important to identify the parameter details of the blocks to make them work in the desired manner. To correctly configure the blocks, mathematical and functional description understandings are the prerequisites. The target blocksets used are identified as communication, signal processing, Simulink general blockset and embedded target for TI C6000 and math’s functions [11, 15, 17, 20]. The designed model presented in Fig. 3 is an SDR baseband processing stage employing random integer generation source which acts as raw data bits input for the transmitter block. The channel encoding is achieved through convolution encoder block. The modulated complex data symbols are passed through additive white Gaussian noise (AWGN) channel (between transmission and reception). At the receiver stage, 16-QAM demodulator extracts bits from the received symbols then Viterbi decoder corrects the errors in the received bits [1, 5, 10, 13]. The error-rate calculator at receiver side in the model estimates the bit error rate (BER) as shown in Fig. 3.

Fig. 3. Simulink SDR model

(a)

(b) Fig. 4. (a): Transmitter Block in the Simulink SDR Model (b): Receiver Block in the Simulink SDR Model

It is important to note that some additional blocks are also present in the model because a model should have balanced data types and correct signal dimensions [13]. The model in Fig. 3 is simulated for different bitrates till an acceptable BER is achieved.

140

ISBN: 978-969-8680-32-9


After successful modeling and BER of the order of , now a model is to be built which can be used in real time on a DSP platform [10, 13, 17]. This is done by successively replacing and adding blocks from the model of Fig. 3 with the DSP kit’s input / output supported blocks in Simulink embedded target for TI C6000 blockset [11]. The new model now contains the SDR transceiver that can be implemented on the DSP kit.

Fig. 5. DSP Implemented Model

In Fig. 5, digital inputs from the DIP switches of the DSP kit are sent to the transmitter block of SDR. The input data is buffered and after passing it through the processing blocks to make it useable by the convolution encoder, it becomes an input to the modulator. The rest of the system configuration is same as in Fig. 3 because the used blocks can be implemented directly on hardware with hardware translation of Simulink [17]. On hardware, the real time input is processed and is available on the output LEDs for validating the whole operation of the transceiver. DSP Hardware Implementation Computationally demanding SDR processing blocks require high data rate processing platform [21]. Specialized design and integration of the software tools have enabled a quick and handy model’s hardware translation on the DSP platform [17-18]. The software radio must be easily programmable through software so reconfiguration of the physical hardware (with the capability to handle data-rates) can be done rapidly. This is done by resource allocation in hardware since there are many important differences between software and hardware radios. The real time target workshop support is used to generate hardware code for the DSP platform [13, 17]. The most important factor to accomplish this support of Simulink for DSP is the use of a hardware oriented model to fulfill the requirements [1, 10, 21]. In such a model, the blocks of the kit are used as they can be used in the real physical world. Else Simulink will generate errors in building the model for the hardware or at least the code would not run according to the logic it is intended for. The model presented in Fig. 5 is translated to a hardware understandable firmware using built-in Matlab functions [17].

Results The overall modeled system discussed in this paper has different blocks with certain parametric configurations. It is critical to get the logical results from one block in order to step up the second. Mainly, correct configurations are required for the good channel encoding and constellation for the QAM mapper. The two main tasks are the generation of a Simulink model plus a real time DSP implementation with the satisfactory real time experimental results.

Simulation Results The results obtained from the model simulation in Simulink are validating different operations of the transceiver such as modulation and BER. The constellation diagram is presented in Fig. 6 at the receiver end after passing through AWGN channel with lower SNR for analysis of the transmitter stage [14]. In comparison to Fig. 2 earlier, and as presented in [15], the constellation diagram in Fig. 6 shows the scattered symbols in I and Q plane. The results are acceptable since the positions of symbols in the specific coordinates are not overlapping. Furthermore, the ISI (inter symbol interference) is also acceptable since the eye diagrams of In-phase and Quadrature-phase are open and a clear rhombus shape pattern can be seen in Fig. 7 indicating lower ISI.

141

ISBN: 978-969-8680-32-9


Fig. 6. 16-QAM constellation at demodulator Input

Fig. 7. Eye Diagram at demodulator Input

The BER results of the modeled system are presented in Fig. 8 (a). It can be seen that the which is considered as within the achieved BER for the current systems is 9.259 x acceptable range for the presented system's configurations.

(a)

(b)

Fig. 8. Shows (a) BER calculation (b) Tx and Rx data

The display used at the transmitter and receiver ends to see the input and output baseband data shows good reception of the transmitted bits at the receiver end as can be seen in Fig. 8 (b). The simulated I and Q signals after the AWGN channel are presented in Fig. 9.

142

ISBN: 978-969-8680-32-9


Fig. 9. I and Q signal waveforms before (Up) and after (Down) passing AWGN Channel

Experimental Results There is a built-in Joint Test Action Group (JTAG) emulator on the DSP platform [11] through which realtime data exchange (RTDX) and the code debugging is performed [11]. The input from the DIP switches are recorded in a buffer and a bit stream is formed. The model was shown in previously in Fig. 5. This bit stream is baseband processed in the DSP platform and output bit stream is presented on the LEDs. The time taken by the output to appear on the LEDs is approximately 28 seconds. According to the processing platform, this time delay can further be improved by the hardware level optimization. The functionality of all the modules in the model was verified by analyzing the input and output bit streams using RTDX and from I/O present on the DSP platform. In Fig. 10, the DIP switch position shows the input to the transmitter block which after all the processing is output to the LEDs verifying the systematic operation of the blocks in the system.

Fig. 10. On kit emulation results

Conclusion and Future Work The design approach for SDR and its implementation over DSP hardware is simple and result oriented. A Simulink model oriented for hardware can be made workable on the real time hardware with good & satisfactory results. By using the modeling blocks wisely, rapid changes can be made and tested both in the simulation and emulation on the hardware. Unlike other available platforms and approaches, our proposed approach provides cost effective, less time to market and easily upgradable software defined radios for commercial & academic purpose. Further, adding custom blocks to carry out special function is easier in Simulink than to use any other OS (operating system) platform requiring C++ & python programming skills. The digital signal processor platforms with high frequency built-in analog-to-digital and digital-to-analog converters can be utilized for the direct processing of real time RF signals

143

ISBN: 978-969-8680-32-9


References 1. J.Mitola,III Mitre corp., McLean,VA, “Software radio architecture: a mathematical perspective,” IEEE Journal on Selected Areas in Comm., vol. 17, no. 4, 1999. 2. Grégory, E.N, M.S and François. V, “Transaction Level Modeling Of SCA Compliant Software Defined Radio Waveforms and Platforms PIM/PSM” Conference & Exhibition on Design, Automation & Test, 2007, Europe. 3. J. H.Varde, N.B.Gohil, J.H.Shah, “A General System Design & Implementation Of Software Defined Radio System”, JIKRECE November ’12 to October ’13, Vol.2, Issue 2. 4. R. C. Reinhart, S. K. Johnson, T. J. Kacpura, C. S. Hall, C. R. Smith, J. Liebetreu, “Open Architecture Standard for NASA’s Software-Defined Space Telecommunications Radio Systems”, Proceedings of the IEEE, Vol. 95, No. 10, 2007. 5. Ali E. Moussati, K. Ghoumid, J. Zaidouni, E. Signaux, “Hardware Implementation For Turbo Code-OFDMA Using Software Defined Radio”, 978-1-4673-4766-2/12, 2012 IEEE. 6. S.K Vasudevan, Sivaraman R, Z.C.Alex” Software Defined Radio Implementation (With simulation & analysis)”, IJCA (0975 – 8887) Volume 4–#8, 2010. 7. Chapter 6 Passband DataTransmission by: Po-Ning [email protected] 8. J.Mitola III, “Wiley Encyclopedia of Telecommunications”, by John Wiley & Sons. April 2003. 9. National Instruments,” Case Study on Architecture of SDR for Communications Test”, Publish Date: Apr 02, 2013. 10. J. Mitola, “The Software Radio Architecture”, IEEE Comm. Mag. May 1995. 11. Rulph Chassaing, “Digital Signal Processing and Applications with the C6713 and C6416 DSK”, John Wiley & Sons, Inc. 12. H. N. Abdullah, H. A. Hadi, “Design And Implementation Of Fpga Based Software Defined Radio Using Simulink Hdl Coder”, University Of Al-Mustansiryah, College Of Engineering, Electrical Engineering Department. 13. Okhtay A., S. G. Bilén” Developing A Rapid Prototyping Method Using Matlab/Simulink/Fpga Development Environment To Enable Importing Legacy Code”, Proceedings Of The SDR ’08 Technical Conference And Product Exposition, SDR Forum, Inc. 14. Xiaoloung L., Simulink-based simulation of quadrature amplitude modulation (QAM) system, Proceedings of the IAJC-IJME International Conference 2008. 15. R. Swain, A. K. Panda,” Design of 16-QAM Transmitter and Receiver: Review of Methods of Implementation in FPGA” in IJES Vol. 1, Issue 9, 2012. 16. E. Mayo and J. Bryan, “Model-Based Design of Communication Systems,” Mathworks Presentation, 2008. 17. Neil Ding, Zhenggao Xu Zhiqiang, Wang Stella Sun “Model-Based Design of Embedded Signal Processing Systems with Simulink®”, September, 2008 The MathWorks, Inc. 18. Signal Processing and Communications with MATLAB and Simulink by: Giorgia Zucchelli Application Engineer – MathWorks. 19. Simulink Tutorial on Digital Modulation Methods”, MathWorks Inc. 20. MATLABCENTRAL: http://www.mathworks.com/Matlabcentral 21. S. S., J. H. Reed, P. Athanas, R. Boyle, “A soft radio architecture for reconfigurable platforms”, IEEE Comm. Mag.z, 2000. 22. Joseph Mitola, “Software Radio Architecture Evolution: Foundations, Technology Trade Offs, and Architecture Implications”, IEICE TRANS. COMMUN., VOLE83-B, June 2000.

144

ISBN: 978-969-8680-32-9


Implementation of PLC Based VFD Fed Three Phase Induction Motor via OPC Server Mosaib Jan, Mukhtiar Ahmed Mahar, Abdul Sattar Larik, Syed Sohail Ahmed Shah Mehran University of Engineering and Technology, Jamshoro, Pakistan [email protected], [email protected], [email protected] Isra University, Hyderabad, Pakistan [email protected]

Abstract This paper addresses the obstacle of data communication between various Windows APIs through the implementation of the OPC Server. One such approach highlighted in this paper is the communication between NI LabVIEW and PLC programming application. This paper intends to explore the method through which LabVIEW as a HMI can be used to manipulate the data into the PLC. The techniques of sharing variables across Microsoft Windows and using them in LabVIEW are discussed along with the advantages of using standard interface protocol. Using the discussed knowledge, the later part of the paper is dedicated in making a practical implementation of such system and use it to remotely control a VFD fed three phase induction motor through an industrial PLC. Keywords: OPC Server, Shared Variables, VFD, Windows API, NI

Introduction With the accession of sophistication in Control System, it is of prime importance to inter-link two or more application as to exchange data over common platform such as Windows. OPC Server acts as the standard interface protocol between Microsoft Windows APIs for data exchange. OPC Server not only allows reliable inter device compatibility but also provides easy and cheap expansion of industry. An OPC Server can therefore be used as an interface link between LabVIEW which is an excellent HMI and a PLC programming software [1]. When such setup is achieved it provides swift and mere means of remote controlling industrial electrical devices such as motors with only few clicks. The paper discusses the conventional system in contrast with the system employing OPC Server. Afterwards a system has been designed which utilizes the Siemens PC Access acting as a standard OPC Server for remote control the speed of Three Phase Synchronous Motor. In order to control speed, Delta VFD S-Series has been used which controls the frequency in according to the step input provided by PLC Siemens S7-200. The I/O control of PLC is achieved through LabVIEW.

Design Components: LabVIEW LabVIEW, acronym for Laboratory Virtual Instrument Engineering Workbench, is interactive development software made by National Instruments or NI. NI is a software company based in America. Their product LabVIEW is G-Language development platform which means that all the instructions are written graphically by means of various function blocks and icons which are then interconnected through wires. LabVIEW workspace typically consist of two windows i.e. front panel and block diagram. Due to the resourcefulness, extensive open source libraries and hardware support LabVIEW holds strong reputation among Engineers. One such important feature of LabVIEW is that it can be used to import, export and manipulate the shared variables, making it perfect for automation system [2].

Working with Shared Variables Shared Variable is a common code that can be used by multiple executables such as drivers. By sharing the data of one application, it enables us to use it into some other windows application. Operating System of PC also use shared variables. In Windows, it is known as .dll or dynamic link library. Once a variable is shared as .dll, several windows application can utilize it which not only results in time but also memory space

145

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 saving. For an example it can be seen in fig 1 that the print screens of both WordPad and Notepad look alike. Rather than making print function for different applications, print .dll is shared data and is being used by several programs.

Fig. 1.

Two distinct applications of windows utilizing same shared print variable

LabVIEW comes with built-in libraries and programs which can access shared variables and manipulate that data according to program stored into LabVIEW. Through LabVIEW Data logging and Supervisory Control (DSC) Module, a user can have unlimited shared variable. LabVIEW provides extensive device support but shared variables ensures the communication with drivers that are not integrated in LabVIEW. OPC Server OPC is an open system interface standard, which allows simple and standardized data exchange between PLC applications, field devices, PC applications and HMIs. OPC Foundation came into 1996 and is creator of OPC interface. In past OPC stood for the acronym OLE for Process Control which was later in 2011 changed to Open Platform Communications. The main intention of creating OPC interface was to design standard communication link between industrial automation devices and since then it has become a prime choice of industries [3, 4]. A plant or an industry typically needs to extract and manipulate the data from PLC through HMI. Due to vast variety of devices, each application has its own driver and each vendor uses its own communication link to interface trends, reports, and HMIs to PLC. As in result whole data exchange is proprietary based. It create a communication barrier between different automation devices because whenever an industry wants to expend, they are unable to interface their new devices with their old. So each time an industry ends up going back to the service vendors and spend more money. As illustrated in fig 2, the network compatibility of devices with each other decreases with every expansion. For an example when a plant is initially running with PLC-1 and PLC-2 which are both linked to HMI-1. Now upon expansion 1 i.e. installing additional PLC-3 and HMI-2, PLC-1 and PLC-2 are unable to communicate directly to HMI-2 hence leaving us with the only option of linking HMI-1 and HMI-2 directly. Similarly on expansion 2 i.e. installation of Computer Maintenance Management System (CMMS), not surprisingly our existing system is again unable to communicate with CMMS due to the unavailability of network drivers, consequently we once again end up setting an operator on CMMS whose sole job is to manual feed the data into CMMS by taking the data from HMIs visually. Now most certainly this is not desirable and also it has decrease the process and production efficiency. This is the stage where OPC set foots in. As shown in Fig 3, due to standard interfacing, not only all the devices are talking to each other without any communication barrier but also it has eliminated expensive vendor interfaces and drivers. Therefore it enables the end users to freely choose hardware and software application as per their needs without having to consider the availability of proprietary drivers. OPC being the standard interface link, can be used to convert the proprietary data into OPC format, which can then be used to setup trends, develop alarm reports and controlling motors etc.

146

ISBN: 978-969-8680-32-9


Fig. 2. Incompatibility of automation devices with future expansion

OPC architecture comprises of a Server and a Client network. Clients are typically devices such as HMI, Digital Reporting System, and CMMS which request data from the OPC Server. In response to that, OPC Server look into the memory location of PLC for desired data variable and make it available for Client. In order for that to happen it is imperative that a data tag has been established in the field device (PLC). Since each data tag has unique program memory of PLC it permit the OPC Server to look for those data tags specifically and pass onto the current status of variables to the OPC client. OPC Server comes with different specifications, among which OPC DA is of prime significance. OPC DA allows Windows APIs to communicate and exchange real time data with each other. It also allows a programmer to read and write data into the OPC Server and Client. Due to this feature, OPC DA provides very swift communication between Windows applications and field devices. A user can effortlessly create visual programs as on LabVIEW and use it to control and command PLC and take real time decision on the basses of acquired data.

Fig. 3. Devices compatibility after OPC Server

147

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 One such approach is shown in this paper. NI has its own OPC Server application which can be downloaded and also many PLCs come with vendor provided OPC Server application where a user can easily create tags (variables), assign it on LABVIEW program and the PLC is all ours to control. In case of Siemens SIMATIC S7-200, Siemens PC Access serves as an OPC Server and maximum of eight PLCs can be connected with unlimited tags. On the other side LabVIEW provides excellent platform for HMI and can easily act as an OPC Client.

Making it work One of the cardinal component of many plants and industries is AC motor. Due to its robustness, cheapness and simple maintenance, AC motors have superiority over other motors. Many applications of AC motor require accurate speed control and for that reason, VFDs are revolutionizing today’s most active and efficient plants [5, 6]. Variable Frequency Drives or VFD is the most efficient electronic device which can alter the speed of AC motors by varying supply frequency. nr = ns (6) In case of three phase AC motor, the rotor is magnetically locked with stator field and rotor speed nr is equal to the stator field synchronous speed ns. As shown in Eq (1), the rotor speed can be varied by changing the supply frequency fs and number of poles (p). Since most of the time three phase AC motors have compact design, it is impractical to change the number of poles on load. In that event changing the frequency is most viable option but is not that straightforward because it has been observed that decrease in supply frequency will cause in strengthening in stator flux which will further cause motor current to increase. So we deduce that for any VFD it is essential to decrease the supply voltage corresponding the decrease in supply frequency thus keeping v/f ratio constant. It is worth mentioning that VFD not only provides with the capability to control speed but also essential features like:  Energy Saving  Controlled Starting Current  Less dependency on mechanical drive components  PLC control. For the purpose of practical prototype, Delta VFD S-Series is used along with Siemens SIMATIC S7-200. Like many VFDs, Delta VFD S-Series comes with built in PLC and Ethernet frequency control. For the purpose of controlling frequency via PLC, it has multi step inputs labelled as M terminals. Once the parameters of VFDs are set for PLC control, terminals from M0 to M5 on VFD can be used to assign functions such as Start, Stop, Reverse, Forward and Multi step frequency control. As long as our prototype is concerned, we intend to assign variables to these terminals as shown in table 1. Here we have dedicated terminal M0 and M1 for controlling motor dynamics, M2 for parameters reset, and the terminals M3, M4, M5 are responsible for controlling the frequency and thus speed output. In Table 1 it can be seen that these control terminals of VFD are directly fed through the PLC terminals from Q0.0 to Q0.5. It is also worth mentioning that with the help of three frequency terminals we can create eight distinct binary combination which would further produce eight distinct step frequencies [7]. The different set of conditions producing different output frequencies are given in table 2. Through our discussions we conclude that in order to make a viable practical prototype, performing application to application communication, we will certainly need a PLC, PLC programming software, OPC Server, LabVIEW, VFD, and a three phase AC motor. The basic block diagram illustrating our prototype is mentioned in Fig 4. As we are aware of the fact that PLC is a microcontroller based device which functions according to stored data logic. MicroWIN is one such basic ladder diagram building tool, provided by Siemens. It consist all the fundamental tools and functions essentially required to program Siemens Simatic S7-200. In this experiment, LabVIEW is used as a Human Machine Interface through which the input and output variables will be controlled. A vendor provided application, Siemens PC Access can act as an OPC Server and so it will contribute to all the communications between MicroWIN and LabVIEW. It is worth noticing that since all the input variables to the PLC are not physical voltage 1 and 0 but rather fed through LabVIEW, we have used memory switches such as M0.0, M0.1 etc. on PLC side which exist only virtually.

148

ISBN: 978-969-8680-32-9


Table 1. Assignment of different variables in contest to PLC terminals VFD Terminals

Function

PLC Terminals

M0

START/ STOP

M1

FORWARD/ REVERSE

M2

Reset

Q0.2

M3

Multi-Step Speed 1

Q0.3

M4

Multi-Step Speed 2

Q0.4

M5

Multi-Step Speed 3

Q0.5

Q0.0

1

1

Q0.1

Table 2. Binary inputs at Multi-Step Frequency control with regard to output frequency Output Frequency (Hz)

M3

M4

M5

0

0

0

0

7.14

0

0

1

14.3

0

1

0

21.42

0

1

1

28.5

1

0

0

35.7

1

0

1

43

1

1

0

50

1

1

1

Fig. 4 . Basic Block Diagram implementing OPC Communication

Once the ladder logic has been written, we now are required to share the desired variable through Siemens PC Access. Once the variables are shared throughout windows platform, almost any third party application can manipulate these variables including LabVIEW which comes with excellent data interfacing capability. After that we setup Boolean buttons in LabVIEW. After buttons are created, a user can open the properties of button, go to Data Binding, choose DataSocket, browse to DSTP Server, navigate to s7.200OPCServer, locate the variables and choose to tag them. This process is illustrated in fig 5.

1

Binary 0 and 1 corresponding to switch Open and Close respectively.

149

ISBN: 978-969-8680-32-9


Fig. 5. Tagging Variables from the button properties

Results The application described above was tested, and the results acquired were just as predicted. It is also observed that a slight delay persist between the input and the execution of command. The range of communication was also extended up to a noticeable length by using Ethernet Module CP243-1. As mentioned previously that a program on Step 7 MicroWIN was written in order to execute the control commands on PLC. A snap shot of a very simple program is shown in fig 6. In PC Access we are required to setup new PLC and add items (shared variables) with the addresses given in table 1 which, when done practically, is shown in fig 7. The screen shots picturing LabVIEW utilized in achieving objective is shown in fig 8. Finally a picture showing all the mentioned components working together is given in fig 9, in which we can find Motor, VFD, PLC Kit, and a laptop running LabVIEW.

Fig. 6. Ladder Logic constituting Memory Switches.

150

ISBN: 978-969-8680-32-9


Fig. 7. Tagging Variables in PC Access

Fig. 8. LabVIEW Design

Fig. 9. Overall system employing given method

Conclusion With the advancement in Instrumentation and Control, it is of high demand to create a system that allow an operator to remotely access the PLC data. Lack of device compatibility put certain limits for such operation. Proprietary based solution are quite expensive and what’s more it increases the complexity. In that case OPC Server play a vital role in interconnection of field devices. Due to the work being done in software system of

151

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 process control, it is wieldy to find OPC Server application. NI OPC Server is one fine example of it. The OPC Server itself can be setup in minutes. Availability of different OPC Server Configuration such as OPC DA, OPC AE, OPC CPX, OPC HAD etc. provides control operator with more commands and functions over field devices, thus improving the overall efficiency of an industry. The results obtained through this setup is step speed control i.e. frequency varies with the resolution of 7 Hz. This obstacle can be overcome if RS-485 communication takes place. Then for PLC it is necessary to have two RS-485 port (one for the PC and other for VFD) or an Ethernet Module CP 243-1 which would establish an industrial Ethernet communication between PLC and VFD.

References 1. Xuejun, Xiang, Xia Ping, Yang Sheng, and Liu Ping.: Real-time digital simulation of control system with LabVIEW simulation interface toolkit. In: Control Conference, 2007. CCC 2007. Chinese, pp. 318-322. IEEE, 2007. 2. Sumangala, B. V., and K. Bhargava Ram.: Advantages of LabVIEW over Embedded System in Home Automations. In: International Conference on Advancement in Engineering Studies & Technology. 2012. 3. Zheng, Li, and Hiroyuki Nakagawa.: OPC (OLE for process control) specification and its developments. In: SICE 2002. Proceedings of the 41st SICE Annual Conference, vol. 2, pp. 917-920. IEEE, 2002. 4. Dongjiang, Li, and Sun Ruiqi.: Implement of communication between configuration software and OPC server based on Modbus/TCP. In: Electronic Measurement & Instruments (ICEMI), 2011 10th International Conference on, vol. 1, pp. 218-221. IEEE, 2011. 5. Velagic, Jasmin, Admir Kaknjo, Nedim Osmic, and Tarik Dzananovic.: Networked based control and supervision of induction motor using OPC server and PLC. In: ELMAR, 2011 Proceedings, pp. 251-255. IEEE, 2011. 6. Sowmiya, D.: Monitoring and control of a PLC based VFD fed three phase induction motor for powder compacting press machine. In: Intelligent Systems and Control (ISCO), 2013 7th International Conference on, pp. 90-92. IEEE, 2013. 7. Barsoum, Nader N., and Pin Rui Chin.: Ethernet Control AC Motor via PLC Using LabVIEW. In: Intelligent Control and Automation 2 (2011): 330.

152

ISBN: 978-969-8680-32-9


A Survey Based Study on Communication and Coordination Challenges in Offshore Software Development Outsourcing Relationships from Vendors’ Perspective Rafiq Ahmad Khan and Siffat Ullah Khan Department of Computer Science & Information Technology, University of Malakand, Khyber Pakhtunkhwa, Pakistan. [email protected], [email protected]

Abstract Offshore Software Development Outsourcing (OSDO) is a business phenomenon of Global Software Engineering (GSE). From the last two decades OSDO has gained competitive advantages due to the development of high quality software at low cost, access to high skilled persons, access to modern technology and round the clock development. However, OSDO is not out of risks and software development organizations face various challenges like geographical dispersion, cultural differences, language differences, communication and coordination challenges and lack of ICTs etc. Communication and coordination challenges are the main stumbling block between offshore outsourcing clients and vendors. A research survey was conducted in OSDO industry to explore different challenges that can negatively affect communication and coordination processes in outsourcing relationships. We have performed questionnaire surveys with 42 experts from different software companies. In the survey we asked from the participants to grade each challenge on a seven point Likert scale to determine the perceived consequence of each challenge. Our findings reveal that cultural differences, geographical dispersion, language differences and lack of ICT/technological cohesion are the most frequently cited challenges faced by OSDO vendors in communication and coordination process with their client organizations in outsourcing relationships. Keywords: Communication and Coordination, Software Outsourcing, Challenges, Empirical Study

Introduction Software development companies across the globe are trying competitively to enhance their business profits by improving the time-to-market of their products, reducing costs by hiring people from countries with cheaper work-hours and defying the "clock" by running the projects during 24 hours. As a result of this, large number of software development projects are managed globally at distributed sites and normally located in different countries. This distributed setting of managing a software project is termed as Global Software Development (GSD) or Global Software Engineering (GSE) [1]. Offshore Software Development Outsourcing (OSDO) is a business phenomenon of Global Software Engineering (GSE) [2]. OSDO can be defined as a contractual relationship between vendor and client organizations in which one or more vendors can get contracts of all or part of the clients' software development activities, and the vendors provide agreed services in return for payment [2, 3]. Due to the rapid growth of this sector, several companies in the developed countries have started to use OSDO paradigm by outsourcing their software development work to vendor organizations located in India, China, Pakistan and other countries with a lower labor cost [4, 5]. Several new Web 2.0 platforms and sites are now developed offshore while the entrepreneurs and management is located in the Western countries such as US, UK and EU [4, 5]. The advantages most revolve around better cost-control over the process, which means that there is lower cash-outflow [4, 5]. Furthermore, the time difference when working with India and China for the Western world allows work to be done round the clock adding a competitive advantage [4, 5]. Though OSDO provides various benefits but it faces several challenges like diversity of communication and coordination, geographical dispersion, cultural differences, lack of trust, language differences, time zone differences and lack of ICT/technological cohesion [2, 3, 6-8]. Khan and Khan [2] conducted systematic literature review and found that the most critical challenge in OSDO is the geographical dispersion, because in geographical dispersion some of the projects activities like coding and testing are carried out at one geographical location while other activities like requirement analysis, implantation and testing are done at another location. So the supervision and controlling of these activities at the same time is difficult [9]. Verner et al [7] found that vendor's poor infrastructure, vendor instability, lack of intellectual property rights, vendor incompatibility with clients and vender behaves opportunistically can also makes hurdles in OSDO activities. Software development process risks like asymmetry in processes, policies and standards, collaboration difficulties, limited tools and resources, poor communication

153

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 bandwidth and large team size effect GSD [7, 10]. Niazi et al [11] have performed SLR and identified that cultural understanding in team, lack of communication and coordination, time zone problem, lack of knowledge management and transfer among teams as key project management challenges in GSD. Initially we conducted a systematic literature review to identify the communication and coordination challenged in OSDO relationships form vendors perspective. The findings of the SLR have been published [2, 3]. In this paper we are presenting findings of the empirical study conducted in the outsourcing industry. The aim of this study was to validate the findings of our SLR study, previously conducted and published, through industry practitioners and to find any other challenges/risk factors apart from the identified ones. OSDO has very significance in the GSD paradigm but little empirical research has been carried out in this area. Particularly there is no empirical research has been carried out in the identification of communication and coordination challenges in OSDO relationships from vendor's perspectives. So by picking up this area and we aim to try to address the following research questions: RQ1. What are the communication and coordination challenges, as identified in the real-world, faced by vendors in offshore software development outsourcing relationships? RQ2. Do the identified challenges vary from company size? The structure of the paper is organized as: Section 2 explains the background. Section 3 explains the research methodology. The findings and its analysis are presented in section 4. Summary is presented in section 5. Limitation part of the paper is presented in section 6. Conclusion and future work is presented in section 7.

Background OSDO is the practices of hiring an external organization to perform some business functions in a country other than the one where the products or services are actually developed or measured [12]. Over the last decade outsourcing functions gain competitive advantages due to many reasons, such as the drastic growth in the ICTs, increased demand of clients and Information System professional in the US and Western Europe [4]. In addition, China and India has made the OSDO a reality because the presence of high qualified educated persons, the availability of resources and skills and also providing better business and economic environment [4, 5]. In OSDO relationship, Khan et.al [13] identified various critical barriers faced by vendor organization. In these identified list of challenges, communication and coordination is reported as the critical challenge to vendors in OSDO relationship. In order to further elaborate communication and coordination challenges in OSDO relationship is the key motivators for this research. Culture bias may lead to erroneous decision and insecurity about other participants' qualification and it can have a devastating impact on communication, coordination and collaboration efforts [6, 14, 15]. Cultural bias occurs when project participants consider their norms and values as universal and neglect to reflect onto what extent values, norms and biases are founded in their own cultural background [14, 15]. Time zone and geographical dispersion can make hurdles in face-to-face communication, increases complexity of planning and coordination activities, causes unproductive waits, delays feedback, makes multisite virtual meetings hard to plan and complicates simple things [2, 15]. The lack of media richness in ICT communications on outsourced software projects can lead to miss-communications and team members may have more difficulty in establishing trusting relationships [16]. Problems such as lack of a common frame of reference, time delays, language differences and language understanding make frequent and uninterrupted communication among offshore software development teams difficult [17]. Nonverbal communication, which is an important component of team communication, is usually missing in OSDO teams because our current technology is able to convey only a limited set of perceptual cues [17]. The four fundamental coordination challenges such as increased coordination cost, reduced informal contact, inconsistent work practices and reduced cooperation arising from misunderstanding creating problems in coordination activities in OSDO relationships [18]. Technological factors, such as different software and hardware architecture, operating systems and interfaces and database system will cause the problems of communication in GSD process [19]. Lack of trust and confidential problems affect the relationship of communication and coordination sides and would bring about many potential problems in GSD process [19]. The lack of informal communication results in lower awareness and poor coordination [20]. As teams grow larger and more distributed, other complications for coordinating work emerge such as establishing common ground, time zone and cultural differences, handling tightly coupled work and so on [21]. The literature reveals that most of the outsourcing projects have been failed due to the poor communication and coordination between vendor and client organizations [2, 3, 7, 22]. It is argued that proper

154

ISBN: 978-969-8680-32-9


communication and coordination between vendor and client organizations are the backbone and two major pillars of the outsourcing relationship [22-24]. In this paper we have conducted an empirical research study to identify communication and coordination challenges in OSDO relationships. Previously, no empirical study has been performed on this topic. Research in this area is expected to provide useful information for outsourcing vendor organizations.

Research Methodology In order to investigate communication and coordination challenges we have performed a survey. A survey is a form of empirical study for providing a quantitative or numeric description on some fraction of the population or the sample through the data collection process by asking questions to gain tacit knowledge on a particular problem [25]. A similar approach has been used by other researchers [26-30]. We conducted the survey through online questionnaire, using the Google Docs free online tool. The main aim for conducting the questionnaire survey is to validate our SLR findings through industry practitioners and to find any new challenge apart from the existing ones. The detailed process for the performing questionnaire survey is described in the following sections:

Questionnaire Design We have used both open and closed ended questions in the questionnaire with checkboxes to test the validation of our findings. We designed 18 questions with the help of the SLR findings. In these questions we have asked the participants to give their response on a 7-point Likert scale (Extremely Agree, Moderately Agree, Slightly Agree, Not Sure, Extremely Disagree, Moderately Disagree and Slightly Disagree). We have also included some open ended questions in the questionnaire e.g. we have asked the participants to mention some other challenges apart from the listed ones. The questionnaire was piloted through five members of the Software Engineering Research Group (SERG_UOM) at the university. There are many ways to develop the survey either through self development or commercial tools. We have used Google docs, commercially available tool, for entering the questionnaire and it generates the necessary web page and scripts.

Designing an Online Questionnaire Survey Our designed survey consists of two steps sampling and questionnaire design. Sampling is a process of selecting participants to answering the desired questionnaire [25]. Questionnaire design consists of set of questions for the samples (participants) to answer the questions. Both are explained in detail in the following subsections.

Sampling To obtain sampling there are two ways; one way is systematic approach and the other is non-systematic approach [25]. In systematic approach the list of entire population is available and samples are drawn from that list based on the statistic, and non-systematic approach is used for small-scale survey [25]. We have used non-systematic approach for conducting our survey because this survey is on small scale and it is also difficult to get contacts form may companies and listing all the members and selecting people from that list. A similar approach is used by other researchers [30]. We used personalized approach in sampling method for performing non-systematic sampling. The questionnaire distribution process was performed by writing an invitation letter containing short summary of the research and was posted to the following websites. 1. LinkedIn Groups (www.linkedin.com), Details are given in Table 1. 2. Software Companies at Pakistan Details are given in Table 2. 3. We also invited for participations the authors of the industry papers selected through the SLR; emails were available in the published papers.

155

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 4. From this invitation participation a total of 110 participants showed their willingness and the questionnaire link were sent upon receiving their willingness. Finally we received 48 completed questionnaires. After applying the quality criteria six questionnaires were dropped. So our total sample became 42, among these responses 36 participants are in the vicinity and 6 participants from overseas. Table 1. Summary of Online Outsourcing Professionals Groups

S/ No

Group's Name

1 2 3 4 5

Software Development Outsourcing to Pakistan India Outsourcing Software Outsourcing in Ukraine Outsourcing and Offshoring BP0-Business Process Outsourcing IT/Software Development Outsourcing & Offshore Outsourcing 2 India Global Outsourcing & Offshoring (IT)Professionals /Buyers /Providers / Freelancers/ Entrepreneurs Remote Project Outsourcing Direct Management for Web Development, Software & Business Project Outsourcing Institute Providers / Influencers Forum

6 7 8 9 10

Number of Members (at request time) 167 30, 937 380 29,631 24, 789 17,676

Date (request posted) 8th November, 2014 8th November, 2014 8th November, 2014 9th November, 2014 9th November, 2014 9th November, 2014

7,891 8,115

9th November, 2014 10th November, 2014

5,504

10th November, 2014

987

10th November, 2014

Table 2. Software Development Companies

S/No

Name of IT Board Companies

1

IT Park Peshawar Companies http://www.kpitb.gov.pk/?page_id=208

2

IT Park Abbotabbad Companies www.kpitb.gov.pk/?page_id=210

3

Punjab IT Board www.pitb.gov.pk

Names of Registered Software Houses Maddar Technologies, Famz Solution, Comtel, iFast, Gravity IT Solutions, iFahja, Deans Techno, Smart Bakhtar Solutions, IT Netmericles, Precise Artificer, Tech, Web Solutions, Next Bridge, Viper Telic Technologies, Step Next, Shalazaon and Vision International Xpert Technologies, Solo Soft Technologies, Nexus Web Solutions, Crystal Communications, Best Experts Providers, Techliance Pvt Ltd, iDeveloperz, Web Evolution Company, Telecomsians Techno Softwarez, Cube Medical Billing, Ion Resource Group and Air Band Communiation Arfa Software Technology Park, Microsoft Innovation Center, Public Cloud and Data Cener and Plan 9 Tech Incubator

Date of request send

8th November, 2014

8th November, 2014

8th November, 2014

Executing On-line Surveys

By using email-ID we invite all the 110 participants and asked them to answering the questionnaire survey. Figure -1 shows the number of responses per day for our survey. It can be seen from the figure that some participants answered the survey within a short time as they received the invitation. We resend an email to the participants at the end of second day to increase the response rate, which helped significantly.

156

ISBN: 978-969-8680-32-9


Fig. 1. Number of Daily Responses

Data Analysis We sent the questionnaire’s link 110 participants upon receiving their consents. Amongst these 48 participants filled the online questionnaires. Out of these 48 responses, we have dropped down 6 responses because the participants were not directly relevant. Therefore the final sample size is reduced to 42 responses. Thus we got the response rate of 38.18% in the survey. These 42 completed questionnaires were further analyzed based on different variables.

Empirical Results In order to answer RQ1, Table 3 presents the list of communication and coordination challenges identified through empirical study. In the questionnaire different participants have selected different options on the 7point Likert scale suggestion for each of the listed communication and coordination challenge. For analysis, we have categorized their responses into three categories as shown in Table 3. First category is Optimistic (Extremely Agree + Moderately Agree + Slightly Agree). Second category is Pessimistic (Extremely Disagree + Moderately Disagree + Slightly Disagree). Third category is Impartial (Neither Optimistic nor Pessimistic). Table 3 shows that all challenges have greater than 70% in optimistic list. 'Cultural differences' and 'Geographical dispersion' among the identified list, i.e. 98% are the most common challenges in communication and coordination to outsourcing vendor organizations. This confirms the findings of the literature as reported below:  Offshore software developing outsourcing companies faces many challenges like cultural differences, geographical dispersion in the communication and coordination process [6].  Communication and collaboration among OSDO developers with distinct cultural backgrounds and geographical separation are the essential elements in GSD projects [31]. Among the identified list, another challenge 'Haziness' (uncertainty) i.e. 93% is the third ranked challenge for OSDO vendor organizations. This in turn supports the findings of the literature which is reported as:  Requirement uncertainty is a peculiar challenge for coordination mechanism in GSD project implementation [32].  The critical factors affecting GSD teams include task characteristics (e.g. complexity, uncertainty, ambiguity and difficulty) [33]. Our results also indicate that 'language differences' i.e. 88%, 'lack of informal/face-to-face communication' i.e. 88%, 'incongruity in infrastructure, processes and goals i.e. 88%' and 'lack of knowledge management and transfer among teams i.e. 88%' are the fourth common challenges which affects the OSDO vendors. These empirical results complement the findings of the literature which is reported as follow:  Language differences and language understanding make frequent and uninterrupted communication among OSDO teams difficult [17].  Lack of informal/face-to-face communication negatively impacts relationship building, social integration of teams, scheduling, task assignment and cost estimation in the GSD environment [7].  Communication, coordination and trust are the most critical challenges of GSD and the root cause of these challenges are cultural and time zone, process and management issues, infrastructure, organization and product architecture [34, 35].

157

ISBN: 978-969-8680-32-9

Accepted Papers - Proceedings of IMTIC'15 Table 3. Summary of Communication and Coordination Challenges identified Through Empirical Study S/ No

Challenges

Total Expert Responses = 42 Optimistic

Pessimistic

Impartial Pessimistic%

Slightly Disagree

Moderately

Disagree

Extremely

Disagree

Optimistic%

Slightly Agree

Moderately

Agree

Extremely

Agree

Not sure

%

1

Cultural Differences

20

11

10

98

0

0

0

0

1

2

2

Geographical Dispersion

17

16

8

98

1

0

0

2

0

0

3

Haziness

4

12

23

93

0

0

1

2

2

5

4

Increased Coordination Cost

7

20

9

85

1

0

1

5

4

10

5

Incongruity in Infrastructure, Processes and Goals

3

12

22

88

0

2

0

5

3

7

6

Inappropriate Task Coupling

3

11

21

83

0

2

2

10

3

7

7

Language Differences

13

17

7

88

2

2

0

10

1

2

8

Lack of Team Cohesion

5

13

16

80

2

1

1

10

4

10

9

Lack of Knowledge Management and Transfer among Teams

6

11

20

88

0

1

0

2

4

10

10

Lack of Informal/ Face-To-Face Communication

26

7

4

88

0

1

1

5

3

7

11

Lack of Common Understanding of Requirements

7

17

12

86

0

1

0

2

5

13

12

Lack of Training in Communication and Collaboration Tools

14

13

6

79

1

1

3

12

4

10

13

Lack of Credence

4

15

11

71

1

0

1

5

10

24

14

Lack of Activities

2

9

20

73

2

2

0

10

7

17

15

Lack of Frequent Feedback

7

17

8

76

3

1

0

10

6

14

16

Legal, Political and Intellectual Property Rights Issues

8

19

4

73

3

2

1

15

5

12

17

Lack of Cohesion

11

13

9

78

0

0

1

2

8

20

18

Lack of Antagonism Management Activities

2

6

22

71

0

2

1

7

9

22

Change

Management

ICT/Technological

 Asad and Khan [36] conducted SLR and recommended that knowledge sharing management is an important factor in OSDO, but it is negatively affected due to geographical barriers, project complexity, ambiguous nature of knowledge, and lack of synchronous communication. We also found that 'increased coordination cost' i.e. 85%, and 'lack of common understanding of requirements' i.e. 85% are the fifth significant communication and coordination challenge to OSDO vendor's organization. The literature reveals these challenges as:  Increased coordination cost become more problematic in GSD environments as a result of volatile requirement diversity and lack of informal communication [7].  Lack of common understanding of requirements can also negatively impact the OSDO organizations [2]. In the pessimistic and impartial category no one of the challenges got a frequency greater than 24%. This suggests that all the participants of survey were completely sure about the role and importance of these challenges in OSDO relationships from vendor's perspectives.

158

ISBN: 978-969-8680-32-9


Communication and Coordination Challenges for OSDO Vendors based on Company Size The sample size of our empirical study is 42 responses from different software development companies as shown in Table 4. We have followed the categorization of Australian Bureau of Statistics [37] for finding the company size. The Bureau has categorized the companies into small, medium and large based on the number of employees, company which has less than or equal to 19 employees will be considered as SMALL sized company, and medium-sized company having the employees range as 19 < Medium = 8

It means for four bit data number of parity bits needed will be three and it is known as Hamming (7,4). These parity bits will be added to the data at the powers of two’s positions and the length of code will become 7. C=K+M [3] In this perspective, setting of the code will be as follow K4 K3 K2M3 K1M2M1 Calculation of Parity bits M1, M2, and M3 M1= K1XOR K3XOR K5XOR K7 M2= K2XOR K3XOR K6XOR K7 [4] M3= K4 XOR K5 XOR K6 XOR K7

Extended Hamming Code The single-bit hamming code detection and correction is discussed in the previous section whereas, in order to detect double bit errors extended Hamming code is used. Extended hamming code working principle is almost same as that of Hamming code for the detection and correction of single bit errors. However there is one difference that is additional parity bit is used which is termed as Overall parity bit which checks all the data bits as well as parity bits [2] and finally sets overall code with even number of one’s. For example extended Hamming code converts Hamming (7, 4) into Hamming (8, 4).

Detection and Correction of Errors in Hamming Code In Hamming code, detection and correction of errors is done by creating a data packet in which redundant bits are added. For instance, four bit data as discussed previously is converted into eight bit code which contains four data bits, three parity bits often known as “Syndrome”, and an overall parity. Enabling for the detection and correction of errors, receiver/decoder will calculate the parity bits in the same fashion as sender/encoder has set their values. On the receiver side following possibilities will be there: 1. Syndrome and overall parity both are zero; it shows communication has no any error. 2. Overall parity is ‘1’, and syndrome is also non-zero, it shows a single bit error has occurred. Syndrome will indicate the erroneous bit. By flipping that bit error can be corrected. 3. Overall parity is ‘0’and Syndrome is non-zero which is the identification of two bit errors, which can only be detected but not found. 4. Overall parity is ‘1’ and Syndrome is “000” it is the indication of error in overall parity bit.

System Design The system design of the Hamming code is done in VHDL and following blocks are especially designed: a. Hamming Encoder b. Process of extraction of message bits from encoded code by the encoder c. Process for generation of overall parity at the decoder d. Process for generation of parity bits at decoder side e. Process for checking either sent parity bits and received parity bits are same or not f. Process for correction of single bit errors g. Process for detection of Double Bit Errors The flow chart of hamming encoder and decoder is shown in figure 1 and figure 2. The VHDL code for hamming encoder process is given in the table -1. In this code, we have initiated two ports data_in and data_out. Data_in is for desired input data and data_out is for encoded data leaving from encoder. Output data from encoder i.e., Data_out is calculated by using X-OR gate as explained in section II. As discussed above, the encoder has to convert the actual 4-bit long message into an 8-bit long code, containing both the parity bits and the actual message. In table – 2, is given VHDL code that extracts the message from the transmitted code (i.e., encoded). This is done by knowing the positions of message bits in the code and assigning those bits to a new signal i.e. msg (3 downto 0) in this process. The code transmitted by the encoder also contains the parity bits. Which are placed at the positions . In this section bits at the positions are assigned to a signal i.e. rec_parity (2 downto 0). This task is accomplished by the VHDL Code given in the Table – 3. For the detection of double bit errors extended Hamming code is used. The extended Hamming code has one additional parity bit known as overall parity bit. Overall parity at the decoder side is generated through the VHDL code given in table – 4. Overall parity bit performs bit to bit XOR operation including parity bits also. It counts the total number of one’s in the code. If the total number of One’s is even then overall Parity bit is set to “0” if the number of 1’s is odd, then overall parity bit is set to “1”, to make even number of one’s.

212

ISBN: 978-969-8680-32-9


START CALCULATION OF PARITY BITS

START

Calculation of Parity Bits

CALCULATION OF PARITY BITS AND Insertion at Proper INSERTION OF THEIR AT PROPER POSITIONS 8 Bit Code

8- BIT CODE

STOP

STOP

Fig. 1. Flow chart of hamming encoder START

8 Bit Received Code

Get Message and Parity

Calculation of Syndrome

Check

single bit error

Identify and Correct

double bit error

Detect Error

STOP

Fig. 2. Flow chart of hamming encoder

As Hamming code is based upon the parity bits. The encoder calculates the parity bits by XOR operation on the actual message and put those bits on . Decoder will also calculate the parity bits by the same method as encoder and VHDL code is given in table - 5. At the decoder side message is extracted as described in the section B. In this section of the system design it is explained how parity bits are calculated at the decoder side. The decoder has extracted the parity bits in the section C, and it has assigned those values to the signal rec_parity. In the previous section it is explained how the parity bits are calculated at the decoder side. In this section the decoder will check either the received parity bits are same or not. As the property of xor gate is that it produces output ‘0’ at the same inputs, and produces output ‘1’, at the different inputs. If the value of signal Syndrome is “000” it will indicate that communication is error free. If the value of Syndrome is non-zero, it will indicate the error position. For example if it is “101”, it is equal to 5, means the fifth bit is erroneous. This syndrome generation is done through VHDL code as given in table – 6.

213

ISBN: 978-969-8680-32-9


Table 1. Hamming Encoder

Table 2. Message Extraction

Hamming_encode: Process (Data_in) Begin Data_out (0) Subbarayan, "Comparsion of very Short-Term Load Forecastiong," vol. 11, 1996. 43. Y.Y.Hsu, "Fuzzy Expertsystem: An Application to Short-term Load Forecasting," vol. 139, 1992. 44. Y. H. R.H Liang, "Fuzzy Linear programming: An Introduction to hydro electric generation scheduling," vol. 141, 1994. 45. M. L. D.Srinivasan, "Survey of Hybird Fuzzy Neural Approaches to Electric load Forecasting," 1999. 46. H. H. Mori, "optimal Fuzzy Inference for Short_term Load Forecasting," vol. 11, 1996.

375

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15

Pervasive learning environment with emerging technologies and learning transformation. Muhammad Yaqoob Koondhar1, Muhammad Malook Rind1, Fida Hussain Chandio1, Asadullah shah1 1 Kulliyyah of Information and Communication Technology, International Islamic University Malaysia.

[email protected] , [email protected],[email protected],[email protected]

Abstract The 21st century has seen some emergence new and innovative learning trends. Among these; P-Learning (Pervasive Learning) is becoming a dominant learning method due to handheld devices’ price reduction, technological support and smartness of smartphone technology. Learners are now no more limited by time or place – they can learn whenever and wherever they are. Thus, P-learning has the potential to take learning beyond the limits of the traditional educational setup. This paper presents the idea of P-learning which is not limited to a single geographic location or mobile or location based technologies; rather, it facilitates learning from anywhere and at any time with any handheld device means 24*7. The purpose of this paper is to propose how this digital learning paradigm helps physically disabled, geographically scattered, learners through recorded or live audio / video lectures without physically attending academic classes.

Keywords: Mobile / handheld devices, Pervasive learning, broadband, educational transformation, emerging technologies.

Introduction In 1960 the idea of distance learning was introduced to distribute learning material to help physically disabled, geographically scattered, professionals and others who can’t physically attend the class through recorded or live audio / video lectures on TV or radio. With the rapid development in information communication and technology and its utilization for learning purpose web – based learning system was introduced. This learning system used digital electronic media and tools to deliver learning material and became popular [1]. In the last 10 years the usage of mobile devices (e.g., mobile / smart phones, laptop, tablet or personal data assistants [PDAs]) is significantly increased [2]. Everyone is connected all the time through mobile devices. These mobile devices are used as a medium of learning because of their mobility and portability [3]. Technological innovations are exerting more and more influence on people’s everyday lives as well as the evolution of cultures and societies. Portable devices have given us access to instant communication and a huge repository of knowledge right at our fingertips. The ways we educate and are educated have also changed in light of these technological advancements. The mobility of individuals has increased our capacity to learn in different situations and environments. Rapid progress in mobile technologies has enabled us to learn at any given time, regardless of where we are, precisely because our mobile devices connect us to endless learning opportunities. Learning using our devices is similar to learning in a traditional classroom setting where the context of our learning is enhanced by various technological features [4]. P-learners can access tools relevant to their learning interest and situations. Continuous technological progress has enabled the growth of engagement and communication between educators and learners. Learning is no longer confined within the context of traditional education. However, this does not necessarily mean the death of traditional education. Rather, this means that the World Wide Web has extensively changed the mode of education in the form of web – based learning. This mode of learning exists because of the explosive growth of information and communications technology. Nowadays, people can complete learning programs using the WWW as an educational medium, the cost of which is not unaffordable. Also, web – based learning programs have features like forums and other messaging tools, encouraging instant communication between learners and instructors. Mark Weiser’s vision of Pervasive computing has now become the norm and has greatly changed how people respond to each other and their environment. The rise of pervasive computing technologies has paved the way for Pervasive learning, or P-learning. In P-learning, educational materials given to learners

376

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 are relevant to their context and are enhanced by context-aware applications. Devices enabled with context awareness can pinpoint where learners are. These devices then use learners’ locations to more efficiently point learners toward relevant services [5]. By just using portable devices like smart phones and laptops, learners can easily pull up online learning resources [6]. Pervasive computing and mobile technology advancements open the door to endless learning opportunities for P-learners. There may even come a time when learners won’t need to bring their devices with them as the environment itself will be the one to answer learners’ needs by guiding them towards the necessary resources or services [7]. For P-learning, the goal is to create a knowledge society – one where learning opportunities are easily accessible to everybody [8]. At present, the continued existence of the traditional educational system is under threat. This is because the traditional educational model requires expensive labor but fails to effectively utilize human resources. To illustrate this point, take a look at the American school system. The normal student-teacher ratio in some urban classrooms is now at an incredible average of 40:1 or even 60:1 [9], [10]. Unfortunately, this is not the result of financial dislocation caused by economic crisis but is an indefinite transformation that has already taken over industrialized countries’ tertiary sector. Various universities / industries have been switching to technology-based models that are more costeffective and increase professionals’ productivity. In the United States, the Department of Education presented a paper, the 2010 National Educational Technology Plan (NETP), which expounds the department’s vision for an updated educational model that will work for the 21st century. The NETP highlights modern technologies from various sectors in relation to new educational modalities which are geared towards more expansive and enduring learning for students. The paper discusses how advances in technology enable life-long P-learning anytime, anyplace and with any mobile device and also structural changes from traditional classroom, school time and teacher as an information source. The paper is structured as follows: In Section 2, technological innovations in learning is explained. Section 3, addresses the objective of this paper which explains the pervasive mobile learning followed by a pervasive (smart) classroom model in section 4 and a framework of learning for 21st century in section 5. In section 6, mobile broadband / wireless devices and pervasive learning is explained. Enabling transformation is explained in section 7. Finally we conclude in Section 8.

Technological Innovations in Learning In the emergence of technological advancements, resources are being allocated for the enhancement of the learning process, which is essentially rooted in communication. Learning centers on gaining new knowledge and information and processing different ideas to come up with new ones [11][23]. As technology grows by leaps and bounds, so, too, does the learning process, which is largely affected by ICT advancements that enable learning to be more effective and productive [12]. The learning process is currently being improved with the use of new interactive technologies (media). However, the traditional learning model with its teacher-student interaction still plays an important role in the learning process [6]. In the traditional classroom, attention is centered on the teacher who has various tools and equipment on hand (sometimes enabled with Internet access) to aid in teaching. In this model, the learning process is reinforced by the students’ interaction with each other and with their teacher [13]. On the downside, the traditional model is hampered by the unavailability of satisfactory classroom resources, difficulty in getting to school due to its location, and the limited number of classrooms. Hence, electronic learning and mobile learning (or E-learning and M-learning, respectively) have become attractive modes of learning for many schools. With E-learning and M-learning, students have access to learning materials 24/7, no matter where they are, as long as they have Internet connection [14]. Education in the 20th century was focused on developing and transmitting learning programs via television and radio [15]. This later on evolved into the use of computers as the main medium for disseminating educational programs. Then came

377

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 new technology in the form of multimedia and hypermedia, pushing the learning process to the next level with educational materials on CD ROMs, which could be used multiple times and be personalized for individual learning needs [6]. The rise of the Internet gave birth to online learning, which widened people’s learning opportunities. The mid-1990s, in particular, were a pivotal period because that was when different modes of e-learning came into play. M-learning, a form of E-learning, is focused on learning using various portable devices like cell phones [15]. E-learning and M-learning may lack direct communication between learners and instructors, but these models more than make up for that with superior learning programs and the ease with which they can be accessed by learners [16][16]. These factors are what make E-learning and M-learning more appealing than the traditional educational model [15]. Furthermore, portable devices are becoming more affordable, and most people nowadays own at least one [7]. Because of all these technological advances, learning has shifted from the traditional approach to digital forms, namely Elearning, M-learning, and P-learning. Information and knowledge move at a fast pace in our alwaysconnected world, so education gained from traditional learning institutions needs to be supplemented with lifelong learning supported by online learning opportunities.

Figure 1: Technological Innovation in learning

Pervasive Mobile Learning In the field of ICT, pervasive or ubiquitous computing is experiencing fast growth. Ubiquitous computing essentially is about the application and influence of ICT in people’s everyday lives and situations. It allows people to be connected all the time, regardless of where they are. The phrase “pervasive / ubiquitous computing” was coined in 1991 by Xerox PARC researcher Mark Weiser [17]. He envisioned that in the 21st century, “the most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it” [18]. At a time when those technological innovations were still just a dream, he correctly predicted just how intertwined technology and our everyday lives would become. His prediction is now a reality as evidenced by the existence of various gadgets that are small yet powerful. For M-learning to happen, learners need to have access to devices that are enabled with Internet or other forms of connectivity, such as Wi-Fi, Bluetooth, etc [19]. However, these devices may not be equipped with context awareness, which is a key factor in the learning process. This is where pervasive M-learning comes in. In this model, the learning process is supported by technology that uses sensors and other similar features to sense learners’ environment and context. On-the-go learners’ portable gadgets are enhanced with technological innovations that are constantly interacting with embedded devices in the environment, no matter where learners may be.

378

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Various kinds of gadgets and devices, such as embedded systems and sensors, are instrumental in making a pervasive environment possible. Microprocessor-equipped gadgets have the ability to sense motion [17]. Ubiquitous computing allows learners to access information anytime, anywhere. It can also support and enhance people’s learning process and activities while always taking into consideration learners’ environments [20]. The virtual models of various environments may either be built into dedicated servers or be created as these environments are sensed or identified by computers [21].

A model of learning for 21st Century The U.S. Department of Education’s NETP can be interpreted from different points of view. This paper aims to revise the traditional educational model. I believe that it must be transformed in such a way that students are better equipped to deal with both favorable and unfavorable circumstances inherent to a globalizing and knowledge-centric society [22]. Dede, in 2010 suggested following basic elements for this redesign. Learning  Learning should not be limited within the traditional educational system. Instead, it should be a lifelong, enduring process with countless learning opportunities that learners can access whenever they want.  Nowadays, learners can more easily access a larger and more versatile repository of knowledge, resources, and educators (which include parents, various experts, and mentors) outside of the traditional educational environment. Thanks to technology, all this has been made possible.  The learning process becomes more meaningful and productive when it is tailored to individual learners’ needs. Learning should happen at a pace suitable to each learner and should appeal to his or her interests and past experiences. Assessment  Competencies and facets of thinking deemed important or necessary in cognitive research can be more easily evaluated with the help of multimedia, interactive communication, and connectivity. These forms of technology also make it easier to evaluate problem-solving skills, perceive deliberate acts done by learners within simulated contexts, and design elaborate logical thinking tasks. All this is based on the everyday contexts people live and move in, including social issues and challenges.  Online learning presents learners with numerous opportunities for formative evaluation, of which learners can fully take advantage. The technology that enhances and is integrated in various learning activities and environments also assesses the learning process by collecting information based on emerging technologies, ubiquitous learning, and educational transformation philosophy. This learning system records learner input. Additionally, it gathers data about learners’ problem-solving skills and application of problem-solving strategies by looking at how many times learners attempted to solve a problem, how many clues they received, and how long they took to solve the problem. Productivity  The traditional educational model is rooted in time-bound measures when determining what learners have attained in their formal education. These measures served their purpose prior to the development of a knowledge-centric, on-the-go society with a better understanding of the learning process and with technological capabilities that cater to different learning styles. With the advent of online learning and technology integrated in classrooms, time-bound assessment will only hinder learners from getting learning experiences that will increase their chances of success and will encourage them to go on to postsecondary education which is now a norm.  The traditional educational model also insists on classifying learners based on their age, separating academics into individual disciplines, structuring the learning process throughout the entire school year via

379

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 classes with the same number of students who are taught the same content and are expected to learn at the same rate.  Before extending students’ learning time, we should consider the number of hours they are in school and our ability to give them meaningful and effective learning opportunities using various media, such as online learning. Outside of school, technology has enabled students to access a wide array of online resources and social networks where they can interact with and learn from people from different countries at any given time. The educational system should take advantage of students’ interest in using technology and the time they devote to learning on their own using technology outside of the traditional school setup. By taking these into consideration, we can increase students’ learning time in a way that would interest and excite them about their learning. The ideal educational system for the 21st century is one where schools that train future teachers use a complex technological system to educate, give licenses to, and offer professional support for all types of educators. In short, this educational model is integrated with technology.

A proposed pervasive (smart) classroom model Pervasive computing utilizes Smart Space, which refers to regular physical contexts that are embedded with systems that are sensitive to various sounds and sights and can identify and respond to people even when they don’t have any special equipment or physical interaction with them. Pervasive technologies lay the groundwork for smart spaces so that they can perceive human activities and react accordingly to them. Thanks to this technological innovation, Smart Space enhances the learning process of M-learners. Because pervasive gadgets are virtually imperceptible, they are easy to carry around and are very inconspicuous that they can be embedded in learning environments [6]. Handheld devices, such as smart phones, can easily detect these smart spaces and identify a particular area’s resources or services that users may need. Classrooms that are enhanced by the smart space model contribute positively to the overall quality of the learning process by integrating the traditional mode of learning with technology [6]. This gives teachers a wider array of media that they can use to aid in their teaching. A smart-space classroom utilizes context awareness, joint-work support, real-time recording of the learning process as it happens, and a natural user interface made possible by pervasive computing devices that perceive sounds and movements [23]. In the ordinary distance-learning model, lessons are pre-recorded and then broadcast to learners. By contrast, the smart space-enhanced classroom model goes beyond that – it captures all the learning events and context within the learning environment. The embedded devices can sense learners’ context, and based on this, the devices then supply learners with the necessary resources or services. The main advantage of this model is that the technologically enhanced environment can support multiple users and devices. Figure 1 shows the overview of the proposed smart-space classroom that can facilitate to both distance learners and local students who are physically present. The classroom features a smart board and a student board. Teachers use the smart board to show notes or discuss concepts, while the student board shows data about the distance learners. This environment as a whole is strengthened by learning module software which can direct classroom behavior. Students who are in the class physically use desktops or laptops to get involved in lectures, while distance learners use their mobile gadgets to tap into the learning environment. In this setup, teachers can conduct their lessons with both groups of students at the same time, using traditional teaching methods integrated with technology. The environment’s embedded sensors help distance learners feel that they are physically present in class. Teachers’ notes and audio-visual learning aids are recorded and transmitted to learners’ devices, which increases learners’ interest in learning [29].

380

ISBN: 978-969-8680-32-9


Figure 2. A proposed pervasive (smart) classroom model.

Mobile Broadband / Wireless Devices and Pervasive Learning Modern mobile broadband devices have been equipped with “six senses” [24]. These sense are: 1) Determining your location, 2) Connecting to and interacting with other networks, 3) Locating local content and services, 4) Acquiring relevant information, 5) Enhancing the immediate surroundings with acquired information and simulated examples, and 6) Taking notes of the user’s specific interests and learning preferences. The traditional classroom is quickly becoming a thing of the past and is being replaced by the ability of new technology to bring the world’s knowledge directly to the student’s fingertips. Technological advances have helped the learning process move past the traditional educational system where learning happens in isolation. Personalizing a lifelong and expansive learning process is now supported by mobile broadband devices that offer features which enable different kinds of educators to unify their endeavors into one consistent support system. Our research about emerging technologies, ubiquitous learning, and educational transformation virtual performance assessments shows that the process of determining which kinds of customized educational experiences would be the best steps to take can be aided by analytics based on abundant information about learners’ interactions in a digital setting [25]. On top of this research, we are also developing and scrutinizing effective augmented realities that blend virtual information and artificial experiences with the real-world context to help learners better comprehend how they can apply what they’ve learned in the traditional setting to real-life situations. Enabling Transformation Movement from the traditional educational system to distributed education means evolutionary, revolutionary, or disruptive change [24]. Evolutionary change focuses on the use of mobile gadgets both inside and outside the classroom setting for learners to access computing, electronic learning materials, and

381

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 enhanced teamwork. On the other hand, revolutionary change is concerned with using mobile broadband to widen support systems for learning outside the context of the traditional educational setup, which gives rise to new formal education models, like distributed learning. Both types of change tend towards intramural and intentional transformation of current educational establishments. Disruptive change, by contrast, centers on building new models for learning institutions, which is a concept introduced by Harvard Business School Professor Clay Christensen. These new learning institutions may be in the form of online or virtual schools focusing on the use of portable gadgets as a way to remove rigid traditional educational models [26]. In this technological era, different institutions will most probably utilize these three transformation strategies. There are many difficult issues for all three strategies for transformation. The most difficult would be those that can be resolved only through the collaboration of various stakeholders. An example of an issue is the reluctance of parents and educators to invest in the infrastructure required for mobile learning because they are not sure that high-quality learning experiences exist. On the other hand, content providers and technology vendors are willing to develop these experiences only if they know for certain that there is an actual demand for these services. Before the potential of mobile broadband usage in P-learning can be fulfilled, these four main areas must be resolved: Infrastructure and Devices: How should we weigh educational investments in relation to the existing wired computer infrastructure vs. the developing wireless infrastructure of mobile gadgets? Privacy and Security: How can access to the Internet and digital learner information be used in improving the learning process without encouraging different types of abuse? Digital Assets and Assessment: How can advancements in electronic learning resources and services be propelled, considering that the educational market is divided and is reluctant to embrace changes and that there is still little understanding of mobile learning gadgets’ benefits and restrictions? Human Capital: How can various stakeholders, including educators, be aided in comprehending the possibilities of on-the-go, 24/7 learning that will be made possible through evolutionary, revolutionary, and disruptive changes that surpass the traditional educational model? Complicating the challenge is that barriers in each area create difficulties for progress in the others. The challenge for P-learning is made more complex because the obstacles involved in the aforementioned areas hinder the development of other areas.

Conclusion In this short paper, I have focused an idea for Pervasive learning by using innovative technology to remove limits and boundaries for learning. The technological innovation will lead to the pervasiveness (smart) classroom which enable learners to exploit the available resources and to contact locally and internationally, limiting language and time barriers and establish learning network at anytime and anywhere. A model of 21st century for learning anytime on anyplace is supported by pervasive technology infrastructure. Universities should develop technological innovative infrastructure that exploit the technology for learning purpose at anytime and anywhere. Use social networking platforms to create communities that provide learning opportunities within and across universities. The paper focuses on the success of pervasive mobile learning environments. The integration of pervasive computing in mobile learning has changed from traditional system to P-learning system. Usage of technological innovations like television, telephone, mobile phones and computers etc for learning purpose appeared strange but eventually they are incorporated into education for learning. In future, applications will be developed so as to enable learners to start a task with their PDA and later shift the location and continue the task with his/her mobile phone and later revert it to PDA to finish it. Even the recent developments in mobile technologies foresee a day where learners don’t even need a device to learn.

382

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

23. 24. 25. 26. 27.

C. Peng, H., Su, Y., Chou, C. & Tsai, “Ubiquitous knowledge construction: Mobile learning re-defined and a conceptual framework. Innovations in Education and Teaching International,” vol. 46, no. 2, pp. 171–183, 2009. M. Wali, E. , Winters, N., & Oliver, “Maintaining, changing and crossing contexts: An activity theoretic reinterpretation of mobile learning.,” ALT-J, vol. 16, no. 1, pp. 41–57, 2008. G. S. Koszalka, T. A., & Ntloedibe-Kuswani, “Literature on the safe and disruptive learning potential of mobile technologies.,” Distance Educ., vol. 31, no. 2, pp. 139–157, 2010. P. V. Vinu, P. C. Sherimon, and R. Krishnan, “Towards pervasive mobile learning – the vision of 21st century,” Procedia - Soc. Behav. Sci., vol. 15, pp. 3067–3073, 2011. P. Chan, T., Sharples, M., Vavoula, G. and Lonsdale, “Educational Metadata for mobile learning,” in International Workshop on Wireless and Mobile Technologies in Education., 2004. Y. S. Yue Suo, “Towards Blended Learning Environment based on Pervasive Computing Technologies,” in Proceedings of the 1st International Conference on Hybrid Learning and Education, Hong Kong, China, 2008, pp. 190–201. A. A. Economides, “Adaptive context-aware pervasive and ubiquitous learning,” Int. J. Technol. Enhanc. Learn., vol. 1, no. 3, pp. 169–192, 2009. M. A. Lytras, M.D. and Sicilia, “The knowledge society: A manifesto for knowledge and learning,” Int. J. Knowl. Learn., vol. 1, no. 1/2, pp. 1–11, 2005. M. Dolan, “Detroit Schools Cuts Plan Approved. Wall Street Journal,” 2011. S. Dillon, “Tight budgets Mean Squeeze in Classrooms,” New York Times, March 6, 2011. “Learning.” [Online]. Available: http://en.wikipedia.org/wiki/Learning. D. K. Chia-hao Chang, Yubao Chen, “From Mobile Learning to Pervasive Learning,” in ACME Proceedings, 2012. W. Kim, “Towards a Definition and Methodology for Blended Learning,” in In the Proceedings of Workshop on Blended Learning, 2007, pp. 1–8. D. P. G. Uday Bhaskar Nagella, “Adaptive Approaches to Context Aware Mobile Learning Applications,” Int. Journalof Comput. Sci. Secur., vol. 2, no. 2, 2009. G. L. Charalampos Karagiannidis, Adamantios Koumpis, “m-Learning and m-Commerce in Pervasive Environments,” Electron. J. Emerg. Tools Appl., 2009. “E-learning.” [Online]. Available: http://en.wikipedia.org/wiki/E-learning. J. H. J. V. Jones, “Ubiquitous learning environment: An adaptive teaching system using ubiquitous technology.”, In R. Atkinson, C.McBeath, D. Jonas- Dwyer and R. Phillips (eds.), Beyond the comfort zone:,” in Proceedings of the 21st ASCILITE Conference, 2004, pp. 468–474. M. Weiser, “The Computer for the 21st Century, Scientific American,” 1991. P. D. Ahmet Soylu, Patrick De Causmaecker, “Context and Adaptivity in Pervasive Computing Environments: Links with Software Engineering and Ontological Engineering,” J. Softw., vol. 4, no. 9, pp. 992–1013, 2009. Y. Lyytinen, K., Yoo, “Issues and Challenges in Ubiquitous Computing 2003,” Commun. ACM, vol. 45/12, pp. 62–65, 2002. Y. Y. Hiroaki Ogata, Ryo Akamatsu, “Computer Supported Ubiquitous Learning Environment for Vocabulary Learning using RFID Tags,” in Proc. of IEEE WMTE2004, Taiwan, 2004. C. Dede, Reinventing the Role of Information and Communications Technologies in Education. In: Smolin, L., Lawless, K., Burbules, N. (eds.) Information and Communication Technologies: Considerations of Current Practice for Teachers and Teacher Educators, Vol. 106(2., vol. vol. 106, no. 2. NSSE Yearbook 2007, Blackwell Publishing, Malden, 2007, pp. 11–38. G. X. and W. xie Changhao Jiang, Yuanchun Shi, “Classroom in The Era of Ubiquitous Computing Smart classroom,” UBICOMP 2011, 2011. B. Dede, C., “Mobile Learning for the 21st Century: Insights from the 2010,” 2011. J. Clarke-Midura, J., Dede, C., Norton, “Next Generation Assessments for Measuring Complex Learning in Science. In: The Road Ahead for State Assessments,” Rennie Cent. Educ. Public Policy, Cambridge, pp. pp. 27–40, 2011. C. W. Christensen, C.M., Horn, M.B., Johnson, “Disrupting Class: How Disruptive Innovation will Change the Way the World Learns. McGraw-Hill, New York,” 2008. S. Thomas, “Pervasive, Persuasive eLearning: Modeling the Pervasive Learning Space,” Third IEEE Int. Conf. Pervasive Comput. Commun. Work., pp. 332–336, 2005.

383

ISBN: 978-969-8680-32-9


Harnessing of Electric Power through Multiple Renewable Sources Using Cascaded MLI for Power Optimization Muhammad Mansoor Malik1 , Ashraf Yahya1 , Sajjad Ahmed Soomro1 , and Shiraz Hasan1 1

EPE Department, PNEC-NUST, Karachi, Pakistan [email protected], [email protected] [email protected]

[email protected],

Abstract Multiple renewable sources are very efficient for providing power and improving the quality of life for millions of people living beyond the grid. Electrical power can be optimized by using Cascaded Multilevel Inverter (CMLI) when the source of energy is renewable and there can be two or more than two renewable sources available for feeding this special type of inverter. Solar Cell module and wind energy are preferable more than other types of renewable energy resources due to their wide range of use in developing countries. In this paper, an overview is presented about importance of renewable energy sources, multilevel inverters and comparison of different multilevel inverter topologies, proposed hybrid system employing 5-Level Symmetric CMLI and 9-Level Asymmetric CMLI fed by a wind turbine and a solar panel module and comparison between 5-Level & 9-Level simulation models. A prototype model is designed and analyzed with MATLAB employing Solar Energy and Wind Energy as feeders and combined Electric Power is harnessed using CMLI technology for driving an AC load. Keywords: Renewable Energy · Hybrid Power System · Multilevel Inverter (MLI) · Total Harmonic Distortion (THD) · Power Optimization

Introduction Dramatic advances have been seen by modern science in hybrid technology, giving birth to hybrid energy systems to generate electricity from different renewable sources such as solar panels and wind turbines. In recent years, the importance of renewable energy sources has been increased because of fast depleting conventional energy sources such as fossil fuels and large prices of nuclear material for nuclear power plants. Hybrid power system is a solution to overcome the need of power required by the consumer where power from conventional energy source is insufficient. Renewable hybrid energy systems are suitable for stand-alone or isolated systems and small grids. Renewable hybrid power systems is nowadays the focus of research in developing countries and several solar parks and wind mill farms are being established to overcome the need of electrical energy. The power obtained from renewable energy sources like solar and wind is DC power. As the distribution sector in all over the world uses AC power so that DC power has to be converted into AC by inverters. For high power applications, simple 2-level conventional inverters cannot be employed because of limited Power and voltage of the available Power semiconductor switches and therefore special type of multilevel inverters are used which converts DC to AC with an output wave similar to sine wave. A multilevel inverter produces desired AC output from several DC sources [1].The technology of multilevel inverter is an important alternative in high power-high voltage energy control area [2].Output voltage produced by a multilevel inverter is a staircase waveform which is similar to sinusoidal wave with low THD. Increasing to N number of levels decreases harmonics. For high power applications, multilevel inverters can easily be interfaced with Renewable energy sources like wind and solar panel [3,4]. Topologies of multilevel

384

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 inverters that are comm mercially usedd are Flying Capacitors C ML LI, Neutral Poiint Clamped MLI M and Casccaded H-bridgee MLI [5]. Thee later topologgy is preferredd as comparedd to two otherss. When coompared to con nventional invverter, the mulltilevel inverteer has several advantages:  The stairrcase quality produces p veryy low distortio on in output vooltage, reducees dv/dt stresses, so problem ms of electromaagnetic compaatibility are minimum. m  Input currrent is drawn with less disttortion.  Multileveel inverters caan be operatedd at fundamenntal switchingg frequency soo to lower sw witching lossess and hence higgh efficiency. One ddisadvantage is, i at high levvels so many semiconductoor power switches are requ uired and as a gate pulse circcuit is required by each swiitch so the oveerall system beecomes expen nsive and com mplex too.

Topoloogies of Multilevel Invverter Neutral P Point Clamped MLI NPC inveerter consists of series chainn of capacitorrs. It is also caalled diode claamped inverteer. The stress iin the voltage aacross every capacitor is limited l by claamping diodees to VDC. Diffferent ratings of the clam mping diodes arre required forr reverse voltaage blocking. If the no. of levels l are “m” ” then no. of capacitors c requuired in the cirrcuit are “m-1”, the no. of blocking b dioddes are “(m-1))(m-2)” and no. of switchess are “2(m-1)”. So the no. blocking b dioddes are quadrratically relateed with the no. n of levels [6]. Figure 1 shows the NPC multileveel inverter.

Fig. 1. Neutraal Point Clamp ped MLI (Souurce: Proteus Software) S The limittation of NPC C topology is circuit c becomees more compplex because of o clamping diodes d as the nno. of levels inccreases. The liink between a transmissionn line of high voltage v DC annd a transmission line of AC C at a high volttage is an appllication of thee diode-clampeed MLI [3]. Flying Capacitor C ML LI This typee of multileveel inverter uses capacitors innstead of clam mping diodes used in NPC.. Each capacittor is charged to a differentt level of volltage for this type of inverter and the DC D sources and a capacitors are w that varioous levels of output o voltagee are producedd when the trransistor switcching connected in such a way wn in figure 2. states aree changed. Flyying Capacitorr MLI is show For “m” ” no. of level, the required no. of auxiliaary capacitorss are “(m-1)(m m-2)/2” and thhe no. of swittches required are “2(m-1)”.Redundanciees in phase aree present in thhe flying capaccitor inverter which w impliess that two or m more correct switch s combinations can produce p the output o voltagee [7]. The app plication of flying fl capacitorr multilevel innverter is statiic var generattion [4]. The limitation l of this t flying cap pacitor topoloogy is the compplexity in pre-charging of capacitors to saame level. Alsso for real pow wer transmisssion efficiencyy and

385

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 switchingg utilization is poor and at high levels of voltagee packaging is i difficult beecause bulk sized capacitorrs are requiredd which also more m expensivve.

Fig. 2. Flyying Capacitorr MLI (Sourcee: Proteus Sofftware) M Cascaded H-Bridge MLI H circcuits. Each H--bridge contaiins 4 This typee of inverter has of series combination of identical H-bridge switches (MSOFETs, IGBTs etc.) and a is conneccted to a separrate DC sourcce (SDCS) [88]. Figure 3 shhows M Cascadedd H-Bridge MLI.

oftware) Fig. 3. Casccaded H-Bridgge MLI (Sourcce: Proteus So fo by swiitching Q1 annd Q4 at a tim me and -VDC1 of first H-Brridge +VDC1 off first H-Bridge block is formed block is formed by sw witching Q2 and a Q3 at a time. Similarrly+VDC2, -VDC2, +VDC3, -V VDC3 and so onn are o other switcches in differeent blocks of H-Bridge afteer a certain innterval of timee. By formed bby switching of adding thhe outputs all blocks and a staircase waveform w is obtained o whicch is similar to sine wavee and contains less harmonicc distortion as compared to the conventioonal inverter output o waveforrm. o clamped diiodes in its circuitry c and other o The advaantage of Cascaded MLI is that there is no need of componeent count is also reduced. For “m” no. off level, requireed no. of swittches are “2(m m-1)” and requuired SDCS arre “1/2(m-1)”. Cascaded MLI M has appliications in Staatic VAR Genneration. In reccent times, several renewablle sources aree also interfacced with this type t of invertter. They are used u as Main Traction Drivve in Hybrid EV E where seveeral battery soources are conn nected as sepaarate DC sourrces [9].

Propossed Model of o CMLI Hybrid H Systtem A model of Hybrid Po ower System is shown in figure 4. In this t model cascaded MLI fed f by solar panel p nd mills is prooposed to drivve a single phhase AC load. The advantag ge of the propposed modules and small win V and windmilll both compleement each othher i.e., in casse of non-avaiilability of solar system iss that solar PV

386

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 during nnight or clouddy weather, windmill willl still producce power andd overall sysstem reliabiliity is increasedd.

Figg. 4. Hybrid Power System using CMLI a Wind Tu urbine Solar Paanel Module and Solar pannel module and a wind turbbine both are acting as seeparate DC soources. In thee proposed hyybrid system, ssolar panel mo odule may conntain two or more m than two individual solar cell arrayss acting as sepparate DC sourcces to feed DC C power to CM MLI dependinng upon the leevels for whicch it is designeed. Similarly there can be tw wo or more thaan two wind tuurbines to act as separate DC D sources. Battery S Storage In day tim me, solar enerrgy from solarr arrays can bee stored in batttery to providde enhanced availability in nnight time. The energy obtaained from wind w is both reliable r and plentiful p in terrms of sustaiinability and wind e it needs to be sttored turbines are capable of harnessing tthat energy. Inn order to usee this energy effectively, u for solar modules and wind turbines has so batteriies are neededd to store thatt energy. Batttery storage used enormouus advantages for power opptimization annd new develoopments are being b carried for flexibilityy and stability of grid. New w solar panel designs have been propossed that all inn one device the power caan be usly from sunllight and also o stored as pow wer reserve by b the same soolar panel for later generatedd simultaneou use [10]. This is a new w concept of power p optimizzation to havee some energyy reserved righht in the panel and w increases quickly, the load can be ussed when it iss needed in clooudy seasons or nights. Whhen speed of wind cannot abbsorb extra wiind power prooduced in this situation so batteries b are ussed to store thhat extra amouunt of power foor power optim mization. Also when windd speed gets slow then for continuous suupply of pow wer to load it iss a better ideea to store it in batteries with battery management system that has under-vooltage protection, over-voltag ge protection, over current protection p etc. Boost Converrter DC-DC B Voltage oobtained from m solar panel module m and wind w turbine iss in the range of 18V-22V so for steppinng up that voltaage to a speciific DC level,, DC-DC Booost converter ccircuit is usedd. The typical circuitry of bboost converterr is shown in figure f 5.

Fiig. 5. DC-DC Boost Converrter Circuit (S Source: Proteuus Software)

387

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 The DC-DC boost circcuit may be oppen loop or cllosed loop. In open loop bo oost converter the output vooltage with the chang ge in input volltage and it iss not controllaable and in thee result outputt voltage of CMLI C change w also channges which caan be harmful for load. For power optimiization closed d loop boost coonverter shouuld be used whiich controls th he output to a fixed value by tracking thhe input voltaage change [11]. The voltagge of each sepaarate DC sourrce of Solar cell c array and wind turbine is stepped up p by boost connverter circuitt and output is supplied to Cascaded C H-brridge MLI. Symmetric Cascaded C H-B Bridge MLI 5-Level S The Voltage of solar cells c and windd turbine usedd as Separate DC sources followed f by boost b converteers is supplied to cascaded H-bridge H MLII. A prototypee model of 5-level cascaded H-bridge innverter is desiigned o MATLAB. 5-level symm metric multi leevel cascaded inverter has 2 DC and analyyzed through simulations on voltage ssources as inpputs which aree equal in maggnitude (V1=V V2) and 2 fulll H-bridge cirrcuits connected in series. Each H-bridge circuit contains 4 switches andd two triggerring pulses arre required for f each H-brridge circuitry. Total of 8 sw witches and 4 triggering t pulsses are requireed. It is shownn in Figure 6.

Fig. 6. 5--Level Symmeetric Cascadedd H-Bridge Ciircuit p mod del 8 triggeriing pulses arre obtained from f a singlee 8-bit port of PIC 16F8877A In the proposed microconntroller. The trriggering circuuit is shown in i Figure 7. These T pulses have h certain period, Phase delay d and pulsee width. The trriggering pulsses G1, G2, G3 and G4 are foormed by the OR O logic of Puulses generateed by the “Pulsse Generator” function in Siimulink and th heir specificattions are given n in Table 1.

F877A PIC Microcontroller M r Triggering Circuit C (Sourcee: Proteus Sofftware) Fig. 7. 16F

388

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Table 1. Specifications of 5-Level Triggering Pulses Switching Signal G1

G2

G3

G4

Timer

Period (sec)

Pulse Width

Phase Delay

ON Time

OFF Time

PG

0.02

60%

0msec

12msec

8msec

PG1

0.02

10%

18msec

2msec

18msec

PG

0.02

60%

0msec

12msec

8msec

PG1

0.02

10%

18msec

2msec

18msec

PG

0.02

10%

0msec

2msec

18msec

PG1

0.02

60%

8msec

12msec

8msec

PG

0.02

20%

0msec

4msec

16msec

PG1

0.02

30%

6msec

6msec

14msec

PG2

0.02

10%

14msec

2msec

18msec

PG3

0.02

10%

18msec

2msec

18msec

Symmetric CMLI is shown in Figure 8.

Fig. 8. Output of 5-Level of Symmetric Cascaded H-Bridge MLI 9-Level Asymmetric Cascaded H-Bridge MLI A 9-level asymmetric multi level cascaded inverter has 2 DC voltage sources as inputs which are not equal in magnitude but V1=3V2 and 2 full H-bridge circuits connected in series. Each H-bridge circuit contains 4 switches and two triggering pulses are required for each H-bridge circuitry. Total of 8 switches and 4 triggering pulses are required. The circuit is shown in Figure 9. The specifications of triggering pulses are given in Table 2.

389

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15

Fig. 9. 9-Level Asymm metric Cascadded H-Bridge MLI

Table 2. Specificationns of 5-Level Triggering Puulses Switching Signal G1

G2

G3

G4

Timer

Period (sec)

Pulse Width

Phhase Delay

ON N Time

OF FF Time

PG

0.02

55%

0m msec

111msec

9m msec

PG1 PG

0.02 0.02

5% 10%

199msec 0m msec

1m msec 2m msec

19 9msec 18msec

PG1 PG2 PG3 PG4 PG5 PG6

0.02 0.02 0.02 0.02 0.02 0.02

10% 15% 10% 5% 10% 5%

4m msec 8m msec 122msec 199msec 166msec 199msec

2m msec 3m msec 2m msec 1m msec 2m msec 1m msec

18msec 17 7msec 18msec 19 9msec 18msec 19 9msec

PG

0.02

10%

0m msec

2m msec

18msec

PG1 PG2

0.02 0.02

15% 30%

8m msec 122msec

3m msec 6m msec

17 7msec 14 4msec

PG3

0.02

5%

199msec

1m msec

19 9msec

PG

0.02

5%

0m msec

1m msec

19 9msec

PG1 PG2 PG3 PG4 PG5

0.02 0.02 0.02 0.02 0.02

5% 5% 15% 20% 10%

2m msec 7m msec 9m msec 133msec 188msec

1m msec 1m msec 3m msec 4m msec 2m msec

19 9msec 19 9msec 17 7msec 16 6msec 18msec

The outpput of 9-Level Asymmetric CMLI C is show wn in Figure 10.

390

ISBN: 978-969-8680-32-9


Fig. 10. Output of 9-Level Asymmetric Cascaded H-Bridge MLI Mathematical Interpretation and Comparison For calculations of THD, Fourier Transform of the output waveform is taken and magnitude of fundamental component of harmonics is plotted against frequency of all other harmonics. If the stepped waveform with s steps then the Fourier Transform is given by [8]:

Load

Model

THD

R(Ω)

L(mH)

5-Level Symmetric

1

20

0.43

9-Level Asymmetric

1

20

0.22

Table 3. Comparison of THD of Output Volatge of 5-Level Symmetric and 9-Level Asymmetric

∑ cos

cos

⋯.

cos

(1)

Here n represents the no. of odd harmonics like 3, 5, 7 and so. The conducting angles are chosen in such a way that lower components of harmonics 5, 7, 11 and 13 are eliminated and as the higher components of harmonics have low effect so THD is minimum. The THD is obtained directly by the THD block in MATLAB-Simpower and the % THD for both 5-level symmetric and 9-level asymmetric at inductive is given in table 3. As compared to conventional two level inverter which has THD = 0.48. In proposed model, the THD of 5level symmetric CMLI is 0.43 and the THD of 9-level asymmetric CMLI is 0.22 which is very low as compared to conventional two level and 5-level symmetric CMLI.

Conclusion The renewable energy sources like that obtained from solar and wind are suitable to harness electrical power acting as separate DC sources using cascaded type of multilevel inverter which is better than Neutral Point lamped and Flying Capacitor multilevel inverters. The low power rating of IGBT switches used in

391

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 cascaded H-bridges enables the proposed model to run at high power load. The triggering of IGBTs through PIC microcontroller makes the total circuit compact, easy and eliminates complexity of modulation. From the simulation results it can be shown that the output of 9-level asymmetric CMLI has lower % THD than conventional 2-level inverter and 5-level symmetric which enables to run the load with greater efficiency at the lower cost.

References 1.

J. Rodriguez, J. S. Lai and F. Z. Peng "Multilevel inverters: A survey of topologies, controls, and applications", IEEE Trans. Ind. Electron., vol. 49, no. 4, pp.724 -738 2002 2. S.Thamizharasan, J.Baskaran, S. Ramkumar and S. Jeevananthan, “A Modular Multilevel Inverter with Cascaded Half Bridge Cells” 2012 IEEE International Conference on Power Electronics, Drives and Energy Systems, pp. 1, December 16-19, 2012 3. J. S. Lai and F. Z. Peng, “Multilevel converters–a new breed of power converters”, IEEE Trans. Industry Applications, vol. 32, pp. 509–517, May/June 1996 4. L. M. Tolbert, F. Z. Peng, and T. Habetler, “Multilevel converters for large electric drives”, IEEE Trans. Industry Applications, vol. 35, pp. 36-44, January/February 1999 5. Damoun Ahmadi, KeZou, Cong Li, Yi Huang and Jin Wang, “A Universal Selective Harmonic Elimination Method for High-Power Inverters”, IEEE Transactions on power electronics, vol. 26, no. 10, pp. 27432752,2011 6. C. Hochgraf, R. Lasseter, D. Divan, and T. A. Lipo, “Comparison of multilevel inverters for static var compensation,” Conf. Record – IEEE Industry Applications Society 29th Annual Meeting, vol. 2, pp. 921– 928, 1994 7. G. Sinha and T. A. Lipo, “A new modulation strategy for improved DC bus utilization in hard and soft switched multilevel inverters,” IECON, vol. 2, pp. 670–675, 1997 8. Muhammad H. Rashid, “Power Electronics Handbook” Butterworth-Heinemann, 3rd Edition, pp. 456 9. L. M. Tolbert, F. Z. Peng, T. Cunnyngham, and J. N. Chiasson, “Charge balance control schemes for multilevel converter in hybrid electric vehicles,” IEEE Trans. Industrial Electronics,vol. 49, no. 5, pp. 1058– 1065, October 2002 10. Xi Zhang, Xuezhen Huang, Chensha Li and Hongrui Jiang, “Dye-Sensitized Solar Cell with Energy Storage Function through PVDF/ZnO Nanocomposite Counter Electrode”, Advanced Materials, volume 25, issue 30, pages 4093-4096, August 14, 2013 11. Vimal M. Vaniya, Jaydeep G. Gajipara Prof. Jayanti A. Jadav, “Single Phase PWM Inverter with Close Loop Dc-Dc Boost Converter for Solar Application”, International Journal of Engineering Research & Technology ISSN: 2278-0181, Vol.2 - Issue 5, pp. 289-293, May – 2013

392

ISBN: 978-969-8680-32-9


Flexible AC Transmission System Controllers: A Review Arsalan Masood1, Qadeer-ul-Hassan1, Anzar Mahmood1 1

EE, Comsats Institute of Information Technology, Islamabad.

Abstract. Development of power generation and transmission, in last ten years, has been inadequate due to limited resources while power demand has increased significantly. Consequently existing transmission lines are used near thermal stability limits under heavy loads and the system stability becomes a power transfer limiting factor. Substantial expansion of generation as well as transmission system in order to accommodate the increased demand is restricted by the environmental, political, social and regulatory constraints. In this environment, Flexible Alternating Current Transmission System (FACTS) controllers open the door towards the advanced control of power system at least for transmission lines. FACTS technology helps to explore some new possibilities for flow control and improves the operational capability of existing and new transmission lines. This paper presents an ample review of major and latest FACTS controllers and of their applications. It is hoped that in future FACTS devices adaption into transmission system will increase and provide more assistance in controlling the power flow through the transmission lines effectively.

Introduction: In recent years voltage stability has become a key matter of interest to operators, especially the power system that are heavily loaded and have shortage of reactive power. Voltage instability is a great threat to power system protection, safety and reliability [1]. The power systems are getting more advanced and complex due to diverse generation sources and transmission of power from these sources without modifying and adding additional transmission capability, in some case, forces the system to operating under extremely overstressed situations. Additionally, it has become difficult to meet the requirement of reactive power and to maintain the bus voltage within adequate limits [2]. To improve overall efficiency, power system operators are forced to move away from the conventional model of centralized generation, transmission and distribution to de-centralized and less regulated operations. This basically means that all aspects of power system engineering such as generation, transmission, distribution and utilization of electric power must now become accustomed to new rules and regulations. In this study, we will concentrate on the transmission part of the power system and issues related to it. Due to limited expansion of transmission lines and increased generation issues like heavily loaded lines, unscheduled power flow and power system stability are becoming more severe. To overcome these issues new kind of devices are introduced that can operate and control power flow in the power system quickly and efficiently. These devices are power electronics based and provide phase angle control, transmission line voltage control and impedance control. These challenges are encountered by power industry with the technology of FACTS and their use is preferred in some studies [3]. Some of these new power electronics based devices can control all three parameters simultaneously [4] [5]. Whereas the conventional devices lacked speed and controllability of multiple parameters at the same time.

393

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Power System Control Generation, Transmission, Distribution Power system consisting of generation, transmission, distribution and consumption of electrical energy can be detached into zones shown in Figure 1: 1 2 3 4

Generation Transmission Distribution Distributed Generation

Fig. 1. Block Diagram of Generation, Transmission and Distribution.

These days power electronic based equipment is common in all zones [6], the emphasis of this paper is on transmission zone, i.e. shifting the power from generation zone to consumption zone. Power System Constraints The power system constraints are many (listed below) and they put a limit over power transfer among areas or region. The typical constraints are: 1 Thermal 2 Dynamic Voltage and voltage stability 3 Power System Oscillation Damping 4 Steady-State Power Transfer 5 Short Circuit Current and Other limitations Some of the above constraints also influence the transmission system, hence a requirement for a solution to use with the transmission lines with highest possible efficiency. Power system controllability To improve the performance of a power system there are three key variables that must be controlled. The three main variables are: 1 Voltage 2 Angle 3 Impedance

394

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 AC netw work controlleers used to im mprove the performance p o a power syystem can be classified inn two of categoriees, conventionnal network coontroller and FACTS F controoller. Overvieew of these coontrollers is shhown in Fig. 2..

AC‐Network A C Controller FAC CTS controllers Thyristo or Valve e

VSC

SVC TCS C DPFC C

SSSC S STATCOM (W/O energy storage) U UPFC IPFC

Conventional Controllers Transform mer Hybrid (R,L,C) Controlleer Swittched s Serries Compe ensation STA ATCOM Switcheed Shunt (With h energy Compeensation Transformer sto orage) L LTC Fault current Synch hronous Limiter Cond densor PST

Fig.. 2. Overview of o Conventionall network controoller and FACT TS Controllers.

pment like switched shunt capacitor, c seriies capacitor, phase shiftinng transformerr etc. Conventiionally, equip were used to control thhese parameteers. Most of conventional c d devices are jusst able to conttrol one param meter at a time. With FACTS S controllers comes c ability to control onne or more parrameter at a tim me. Some FA ACTS controllers such as SS SSC, UPFC annd IPFC are capable c of conntrolling all th hree parameteers simultaneoously [7]. To contrrol voltage, conventionally switched shunt capacitor, Low taap changing transformers and synchronnous condenseer were used. For impedannce and anglee control seriees capacitor and a phase shiifting transform mers were useed respectivelly. Table 1 shows s some cconventional equipment ussed for enhanncing power syystem control. Taable. 1. Conveentional Equipm ment for Enhanccing Power Sysstem Control

Eq quipment

Impedaance control

Voltagee control 

Angle control

Sw witched-shuntt capacitor  Seeries capacitorr  Trransformer LTC  Phhase shiftinng transform mer (P PST)  Syynchronous coondenser With thee developmentt of FACTS controllers onne or more parameters cann be controlleed simultaneoously. Table 2 eexplains what parameter/s each e device caan control.

395

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Table. 2. FACTS Controllers for Enhancing Power System Control

Equipment

Impedance control

Static synchronous Compensator (STATCOM) Static Var Compensator (SVC) Thyristor Controlled Series Compensator (TCSC) Static Synchronous series Compensator (SSSC) Unified power flow controller (UPFC) Interline Power flow controller (IPFC)

Voltage control  

Angle control

  

  

   

Classification and description of FACTS Facts devices as shown in Fig. 2 are classified in the literature as first and second generation based on their functionality as well as technical feature. Their connection to the network is another classification and based on this classification FACTS devices can be differentiated in four categories i.e. series controllers, shunt controllers, series to series controllers and series to shunt controllers. These two classifications are independent, as many devices of a group of first classification may belong to the other group of second classification. In this study we are reviewing devices on their first classification. FIRST GENERATION First generation devices uses thyristor valve with devices like SCR. Some of these devices can exchange active/reactive power but are not able to generate reactive power and some can generate or absorb reactive power but can’t exchange reactive power. Static VAR Compensator (SVC): This device provides reactive power quickly to HV transmission lines thus enhancing the line performance. The word “static” indicates that it has no moving part such as circuit breakers. This SVC device was designed for impedance matching so that power system come closer to unity power factor. If the reactive load of power system is leading, the SVC will consume VARs mainly using thyristor controlled reactors, however if the load is lagging, the capacitor banks are switched in automatically offering greater control of system voltage. Thyristor-Controlled Phase Shifter (TCPS) In this control method the phase shift angle is observed as a non-linear function of rotor angle and speed. But, when we talk about electrical power system with more than one alternators, the angle computed of one alternator as compared with system angle will not be very significant [8]. Thyristor Controlled Series Capacitor (TCSC) Thyristor-Controlled Reactor (TCR) is used in shunt with capacitor bank in TCSC. The arrangement of linking TCR with capacitor bank in shunt will permit the control of capacitive reactance over a wide range.

396

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Similarly linking TCSC with transmission line in series will gives the opportunity of controlling the line impedance. TCSC is a first generation FACTS device which is an economical and effective way of solving the transient stability problem as well as problems of dynamic, steady state voltage stability in transmission networks [9] [10]. SECOND GENERATION Second generation devices can exchange active and reactive power as well as capable of absorbing or generating these automatically. Static Compensator (STATCOM) A STATCOM or static Compensator is a shunt connected device used on AC transmission systems and is a good alternative of conventional static VAR compensator. It belongs to the second generation of FACTS family and is based on power electronics voltage source converters (VSC). As it is connected in parallel it is also called shunt connected controller. The output current of STATCOM can be regulated autonomously without any regard for the system voltage, independent of the detail that it is inductive or capacitive. Usually it is used to support voltage regulation and in power networks of reduced power factor [11]. It can provides dynamic stability and active AC power when connected to source, but most commonly it is used to provide voltage stability in power system [12]. Figure 3 shows circuit diagram of static compensator (STATCOM) without energy storage.

Fig. 3. Shunt Connected Controller

STATCOM system with energy storage system is shown in the fig 4. As shown in the fig interface provides coupling of Dc side of the STATCOM and energy storage which can be photovoltaic systems or capacitor banks.

Fig. 4. STATCOM with storage

397

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Static Synchronous Series Compensator (SSSC) SSSC works similarly like static compensator. The VSC in SSSC is serially connected through transformer to a transmission network as shown in fig 5. In order to regulate active power flow, SSSC is capable of injecting voltage in quadrature with sending or receiving line end voltage. For reactive power, it does not absorb reactive power from the AC system because having a DC capacitor itself forms the reactive power requirement. This makes it capable of regulating both active and reactive power flow [13] [14] [15]. Furthermore, if we want to just balance or maintain the reactive power, quite small energy source which provides a continuous voltage could be used. If our aim is of controlling the phase angle of voltage injected, it is possible only if energy source is big enough.

Fig. 5. Circuit diagram of SSSC

Unified Power Flow Controller (UPFC) UPFC is one of the very complex and advanced FACTS controllers. It is one of the most adaptable and versatile FACTS device ever used to enhance the operation of power system [16]. The concept of UPFC was proposed by Gyugi in 1991.It can control all the parameters such as voltage, phase angle and impedance, individually and simultaneously. It is a blend of STATCOM and SSSC. Primarily, it is used to control power flow in transmission line. Secondarily voltage control, transient stability improvement, and oscillation damping can also be done individually or simultaneously by it in an adaptive fashion [17] [18]. UPFC is based on one dc link which operates two switching inverters as shown in figure 6. Inverter 1 provides or absorbs the real power accordingly to dc link which will be coupled to transmission line through parallel connected transformer; after it is converted back to ac. Inverter 2 performs the key function of UPFC, it injects AC voltage with controllable phase angle and magnitude, which is connected in parallel with transmission line[19][20]. There are two terminals due to common dc link. AC terminal, in which inverter 2 generates reactive power and DC terminal in which real power is exchanged and is converted in to dc power.

Fig. 6. Basic UPFC scheme

398

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Interline Power Flow Controller (IPFC) The Interline Power Flow Controller was initially introduced by Gyugyi in 1998 and used as a solution to the difficulty of compensating multi-transmission lines at a substation. In other words, the IPFC provides many VSCs attached at the similar DC terminal and all offer series compensation for its individual transmission line [21] [22]. In this scheme, the power optimization of the whole transmission network can be achieved in the way of suitable power wheeling via shared DC link from over-loaded power lines to under-loaded power lines. A basic IPFC contains two VSCs as demonstrated in Fig 7. Each inverter injects series voltage to compensate transmission line and shared DC link is denoted using a bi-directional link for real power transmission among these two voltage sources [23]. The alteration in IPFC is that the active power required by inverter 1 is compensated by additional series inverter 2 utilizing additional line in place of shunt inverter in UPFC [24].

Fig. 7. Interline power flow Controller with Two inverters.

FACTS Applications and Advantages To complete this review of FACTS devices, an overview of flexible AC transmission devices applications to problems of power system is discussed in this part. The basic applications of FACTS controllers are: 1. 2. 3. 4. 5. 6.

Power flow control and steady flow control. Voltage control and Power quality Improvement Increase of transmission capability Stability improvement Congestion management Reactive power compensation.

Steady-State Applications FACTS controller’s applications in steady state condition are pre and post contingency voltage control, power flow control, increase of thermal loading and reduction in short circuit level. Different devices are used for different applications. SVC and STATCOM can be used for voltage control and reactive power improvement in steady state. TCSC can be used for power flow control, to reduce congestion, available transfer capability (ATC) improvement and load flow on parallel circuits. UPFC can be used for redirecting power from overloaded lines to under-utilized lines, reactive power compensation, and power flow control and it also allows more power transaction to improve ATC.

399

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Transient State Applications FACTS controller applications in transient state are oscillation damping, transient and voltage stability improvement. In transient state SVC, TCSC and STATCOM are used for dynamic voltage control and oscillation damping. UPFC can be used for control voltage during contingency to protect power system from voltage collapse. To Ideal Power Flow In previous few years, researchers established new procedures or algorithms/process of resolving the ideal power flow problem. FACTS controllers are one of the major inventions of researchers during these years. Different FACTS controllers are used to improve the control and increase the power transfer capacity. In load flow studies, the thyristor-controlled FACTS controllers such as TCSC and SVC are modelled as impedance controlled devices in a transmission system [25] [26] [27] [28]. However, controllers based on VSC such as SSSC and IPFC, shunt controllers like STATCOM and combination of these two; shunt and VSC based controller, like UPFC are more complex and are modelled as source controllable devices [15] [25] [29] [30] [31]. The Interline Power Flow Controller (IPFC) is quite similar but cost effective than UPFC. It is a kind of the VSC based FACTS Controllers which uses multi-line Transmission System for efficient power flow management. To De Centralized Electricity Market At the present time, electricity demand is increasing rapidly. With no new major projects of enhancing or reinforcing power transmission system or networks, it is necessary to construct new power transmission systems. However, a number of factors such as cost, environment and difficulties in obtaining right of way have continuously delayed such construction. As a result, existing transmission lines are operated on overload conditions. As the electricity market is decentralizing, creating an atmosphere of competiveness in open market. FACTS controllers can be a substitute to ease the power flow in overloaded transmission lines, causing improved load ability, low system loss, improved network stability and less production cost by controlling the power flow. The advancement in supply industry of electricity is presenting new areas of power system operation associated with de centralized market. Commercial pressures of getting better results from existing transmission networks indicates a significant role for management using FACTS devices and energy storage. Advantages Of Facts The practical assistances of FACTS and versatility in resolving problems like transient and dynamic stability, load flow current and voltage control are explained in Fig. 8. The conventional answers of these problems are less expensive in comparison of FACTS devices, but they are not as versatile as FACTS.

Conculsion In this paper different FACTS devices or controllers are reviewed, compared and discussed. In previous few years, experimental installation of FACTS controllers on transmission lines are successfully done to improve voltage stability and power flow [32] [33] [34] [35]. But, the significant up front cost of FACTS controllers remains high as the main hindrance to their common use. It is hoped that in future FACTS devices adaption into transmission system will increase and provide more assistance in controlling the power flow through the transmission lines effectively.

400

ISBN: 978-969-8680-32-9


4 3 2 1 0 Load Flow

Voltage Control SVC

Transient Stability

STATCOm

TCSC

Dynamic Stability

UPFC

Fig. 8. Practical Advantages of the main FACT devices

References 1. T. Jaimol and J. Tibin B. Anna, "Analysis of Voltage Collapse in the Kerala Power Grids Using SVC, UPFC & SSSC," in International Multi-Conference on Automation, Computing, Communication, Control and Compressed Sensing (iMac4s), 22-23 March 2013. 2. A. R. Bhowmik and C. Nandi, "Implementation of Unified Power Flow Controller (UPFC) for Power Quality Improvement in IEEE 14-Bus System," International Journal of Computer Technology and Applications, vol. 2, no. 6, pp. 1889-1896, 2011. 3. T. Sýkora, J. Švec, and J. Tlustý, Z. Müller, "Dynamic Loading Control Using FACTS in Extraordinary States," in In Proceedings of the 12th International Scientific Conference Electric Power Engineering 2011, Ostrava, 2011, pp. 1-4. 4. A. Garg and S. K. Agarwal, "Modeling and Simulation of SVC Controller for Enhancement of Power System Stability," International Journal of Advances in Engineering & Technology, vol. 1, no. 3, pp. 79-84, July 2011. 5. V. Vahidinasab and A. Mosallanejad A. Kazemi, "Study of STATCOM and UPFC Controllers for Voltage Stability Evaluated by Saddle-Node Bifurcation Analysis," in First International Power and Energy onference PEC, Putrajaya, Malaysia., November 28-29, 2006, pp. 191-195. 6. R. M. Mathur and R. S. Basati, "Thyristor- Based FACTS Controllers for Electrical Transmission Systems," in IEEE Press Series in Power Engineering, 2002. 7. R. Kravec, M. Kanalik, Z. Conka and M. Kolcun M. Novak, "UPFC Influence to Transient Stability of Power System," in 10th international conference ELEKTRO, Rajecke Teplice, 18-20May 2014, pp. 343-346. 8. A. K. Mohanty and A. K. Barik, "Power System Stability Improvement Using FACTS Devices," International Journal of Modern Engineering Research (IJMER), vol. 1, no. 2, p. 666 672, 2013. 9. Manoj Nair, "EHV AC & DC TRANSMISSION," in EHV AC & DC TRANSMISSION, 2nd ed. Bhopal:, India: Balaji Learning Books, 2010, pp. 161-162. 10. S. Jamhoria and L. Srivastava, "Applications of Thyristor Controlled Series," in International Conference on Power, Signals, Controls and Computation (EPSCICON), 2014. 11. Chong Han, Alex Q. Huang, Mesut E. Baran, Subhashish Bhattacharya, Wayne Litzenberger, Loren Anderson, Anders L. Johnson and Abdel-Aty Edris, "STATCOM Impact Study on the Integration of a Large Wind Farm into a Weak Loop Power System" IEEE Transactions on Energy Conversion, Vol. 23, No. 1, March 2008. 12. Z. Muller, J. Svec and J. Tlusty J. E. Essilfie, "STATCOM Effect on Voltage Stability in Ghanaian Electrical Grid," in 15th International Scientific Conference on Electric Power Engineering (EPE), Proccedings of the 2014, Brno, 12-14 May, 2014, pp. 235 – 240. 13. S. S. Rangarajan, M. Ambili, P. Sujyothi and V. G. Nithya S. Sreejith, "Enhancing The Power Transfer Capability In A Power System Network Using Series Connected Facts Devices For Increased Renewable Penetration," in International Conference on Advances in Electrical Engineering (ICAEE), Vellore, Jan. 2014, pp. 1 – 6. 14. T. Nguyen and C.T. Vu, "Complex-Variable Newton-Raphson Load- Flow Analysis with FACTS Devices," in IEEE Transmission and Distribution Conference and Exhibition, 2006, pp. 183 – 190. 15. Y. H. Song and Y. Z. Sun Y. Xiao, "Power flow control approach to power systems with embedded FACTS devices," IEEE Trans on Power Systems, vol. 17, no. 4, November 2002.

401

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 16. M. F. Firuzabad, M. Shahidehpour, R. Feuillet A. R. Ghahnavieh, "UPFC for enhancing power system reliability," IEEE Transactions on Power Delivery, vol. 25, no. 4, pp. 2881–2890, 2010. 17. H.Samir, Z. S. Ahmed and D. Mohamed K. Y. I Djilani, "Modeling a Unified Power Flow Controller for the Study of Power System Steady state and Dynamic Characteristics," in 5th International Conference of Modeling, Simulation and Applied Optimization (ICMSAO), 2013. 18. P. Pavan Kumar and N. Poornachandrarao, "Improvement of power flow in the power system network by using UPFC device," International Journal of Engineering Research and Applications (IJERA), vol. 2, no. 3, pp. 11941199, May-June 2012. 19. T. R Rietman, A.Edris, C.D. Schauda and S. L. Willirms L. Gyuqyi, "The unified Power Flow Controller: A New Approach to Power Transmission Control," IEEE Transactions on Power Delivery, vol. 10, no. 2, April 1995. 20. S. Khanchi and V. Kumar Garg, "Unified Power Flow Controller (FACTS Device): A Review," International Journal of Engineering Research and Applications (IJERA), vol. 3, no. 4, pp. 1430-1435, Jul-Aug 2013. 21. Xuan Wei, Joe H. Chow, B. Fardanesh and Abdel-Aty Edris “A Dispatch Strategy for a Unified Power-Flow Controller to Maximize Voltage-Stability-Limited Power” Transfer IEEE Transactions On Power Delivery, Vol. 20, no. 3, July 2005. 22. S. Zelingher, B. Fardanesh, B. Shiperling, S. Dave, L. Kovalsky, C. Schauder and A. Edris “ Convertible static Compensator Project- Hardware Overview” in Proc. IEEE Winter Meeting, 2000, pp. 2511–2517. 23. P. Simon and M. P. Selvan S. Sreejith Sishaj, "Investigations on Power Flow Solutions Using Interline Power Flow Controller (IPFC)," in Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System (SEISCON 2011), 2011. 24. T. Lie and D.M. Vilathgamuwa J. ChenTjing, "Basic Control of Interline Power Flow Controller," IEEE Power Engineering Society Winter Meeting, vol. 2, pp. 521 - 525, 2002. 25. L. Gyugyi and N. G. Hingorani, "Understanding FACTS: Concepts and Technology of Flexible AC Transmission Systems," in IEEE press New York, 2000. 26. C. R. F. Esquivel, and H. A. Perez E. Acha, "Advanced SVC model for Newton-Raphson Load Flow and Newton Optimal Power Flow Studies," IEEE Transaction on power systems, vol. 15, no. 1, pp. 129–136, 2000. 27. C. R. F. Esquivel, H. A. Pérez and C. A. Camacho E. Acha, FACTS: Modeling and Simulation in Power Networks. London, U.K: Wiley, 2004. 28. D. J. Gotham and G. T. Heydt, "Power Flow Control and Power Flow Studies for Systems with FACTS Devices," IEEE Transactions on Power Systems, vol. 13, no. 1, pp. 60–65, 1998. 29. A. Ghosh, and A. Joshi G. N. Pillai, "Torsional Interaction Between an SSSC and a PSS in a Series Compensated Power System," IEE Proceedings on Generation Transmission and Distribution, vol. 149, no. 6, pp. 653–658, 2002. 30. K. K. Sen, and C. D. Schauder L. Gyugyi, "The Interline Power Flow Controller Concept: a New Approach to Power Flow Management in Transmission Systems," IEEE Transactions on Power Delivery, vol. 14, no. 3, pp. 1115–1123, 1999. 31. D. Qifeng and Z. Bomina T. S. Chung, "Optimal Active OPF with FACTS Devices by Innovative Load-Equivalent Approach," IEEE Power Engineering Review, vol. 20, no. 5, pp. 63–66, 2000. 32. M. Gernhardt, E. Stacey, T. Lemak, L. Gyugyi, T. Cease, and A. Edris C. Schauder, "Development of a 00 MVAr Static Condenser for Voltage Control of Transmission Systems," IEEE Transactions on Power Delivery, vol. 10, no. 3, pp. 1486–1496, July 1995. 33. M. Ahmed, R. Gutman, R. O’Keefe, R. Nelson, and J. Bian M. Rahman, "UPFC Application on the AEP System: Planning Considerations," IEEE Transaction on Power Systems, vol. 12, no. 4, pp. 1695–1701, November 1997. 34. A. Edris, C. Schauder, and J. Provanzana A. Mehraban, "Installation, Commissioning, and Operation of the World’s First UPFC on the AEP System," in in Proceedings of International Conference on Power System Technology.

402

ISBN: 978-969-8680-32-9


Use of Information Technology in Hospitals at Pakistan: Case Study of Shaukat Khanum Memorial Cancer Hospital and Research Center (SKMCH&RC), Lahore Engr.Imran Anwar Ujan1, Dr Mohd Adam Suhaimi1, Manzar Abbas Hydari2 1

KICT, International Islamic University, Malaysia

2

Shaukat Khanum Memorial Cancer Hospital and Research Center (SKMCH&RC), Lahore

[email protected], [email protected]

Abstract With the vast reserves in Information Technology (IT), the IT has become gradually important. Societies endure to the welfares from IT. Furthermore, the influence of technology on health sector consequences such as patient’s satisfaction and quality of hospital. The effort in identifying impacts from technology has been the separation of benefits of IT from other factors that may also subsidize to organizational presentation. Furthermore, benefits from technology in hospitals may be realized over an extended period of time. We take the data of key modules and implementation dates and Health Information Systems ((HIS) development costs in US dollars. Then we analyze the data and show the how much cost saves to implement Information Technology infrastructure at hospital. Keywords: HIS, SKMCH&RC

Introduction Health care specialists, policymakers, and customers study health information innovations, for example, electronic health records and computerized supplier request section, to be serious to converting the health care industry. Information administration is important to health care conveyance. Given the fragmented nature of health care, the large volume of transactions in the system, the need to integrate new investigative proof into practice, and other complex information management activities, the limitations of paper-based information management are instinctively apparent. While the profits of health information innovation are clear in principle, adapting new information systems to health care has demonstrated troublesome and rates of utilization have been limited. Most information innovation applications have focused on administrative and financial Transactions rather than on conveying clinical care. The utilization of data innovation is compulsory and vital regarding both clinical consideration and administration work because of the data serious nature of the social insurance. Disregarding becoming interests in data engineering, there are still numerous challenges in the execution, utilization and ease of use of health awareness data frameworks. Contrasted with different divisions, the inside IT capacities in human services associations are lacking and linger behind in the powerful advancement of data frameworks to meet the expanding requests for consideration, quality and proficiency . Besides, social insurance is a complex environment for data frameworks because of quick changes in the working connection and attributes such expert society, complex association structure and administration framework and additionally pressure in the middle of these and the data frameworks [1]. The Pakistan Government Under the head of Health Affairs and Services, a total allocation of Rs 9,863 million has been made in the budget estimates 2013-14, which is higher by 25.7% and 24.9% respectively when compared with budget and revised estimates 2012-13. The allocation for hospital services forms the major component under this classification. Shaukat Khanum Memorial Cancer Hospital and Research center (SKMCH&RC) is a modern tertiary-care cancer focus located in lahore, Pakistan with a referral base from all over the country and adjoining areas. The hospital opened in 1994 and provides full administration

403

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 cancer care to the penniless who com-prise approximately 70% of the cancer patient population. To support this endeavor, the hospital's parent trust raises philanthropic donations (approximately US $20 million a year) from inside the nation and from the Pakistani diaspora. This 180 bed hospital has 2000+ employees and sees in excess of 6000 new cancer patients each year. In addition, the hospital has a nationwide chain of phlebotomy focuses to give diagnostic laboratory administrations to the general population. In 2011, there were in excess of 142,000 outpatient visits, 7600 admissions, 7800 surgical operations, 54,600 chemotherapy visits, 44,500 radiation treatments, 144,000 imaging strategies and 3.25 million lab-oratory tests. The pathology laboratory gets specimens for testing from 90 destinations from all over Pakistan. In perspective of the size and complexity of the operation, utilization of computing for automation, quality assurance, safety and compelling financial management is a natural advancement. [2] Initially hospital installed a network with 250 nodes of 10 MBS before December 1994. This system provided email, registration and scheduling of patients, some parts of Pharmacy and radiology covered. Other clinical and non-clinical systems are not included because of non-availability of Information Technology experts in Pakistan. Later in year 1999 several clinical sections were used some software’s. In other hand non clinical sections like finance wing used locally developed Software.

Development of Hospital Information System For developing Hospital Information System SKMCH& RC used ORACLE products with the development team of IT experts in year 2000. The team was responsible to develop an integrated HIS in which clinical, administration and finance wing are included. A steering committee was formed with senior management and clinical experts to monitor weekly performance of developing team. Facility given to patients the team developed modules and prioritized based on perceived benefits to the hospital. The development phase continued with each section staff support. Development team implements a system in each section and gives training to staff working there. Table 1 shows key modules and there implements date; at SKMTH&RC. In 2012 Hospital Information System is automated and fully functional. From OPD to Laboratories, billing system all records of patient kept at access control system. “Patient human resource information, procurement, inventory and assists/materials management as well as various financial modules are part of this hospital wide system.” [2]. Until now maintenance and improvements has been carried out with hospital management support. Working of HIS at SKMCH&RC This time the hospital is paperless, the importance of uninterrupted availability of the system is more critical than ever. [2] One server at off-site place and two back-up servers at in house which stores the realtime data, same as main server store. The real time backups are reserved daily basis for emergency use can be reinstated at new server to restart operations. Avoiding of disturbance of patient care, departments test their eventuality plans manually.

Research Methodology and Results The research will use to understand the complexities involved in accepting Information Technology as well as the requirements of an optimal system. The Questionnaire will be divided in three sections, each encompassing different theme: Back ground Information: This section aimed at getting information about the organization. This include the size, time of organization has been operational, the information technology level. The information about which information they use. Information Technology in Hospitals:

404

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 This section will aim to getting information about use of information technology at organization. The information asked for included the age, investment, authorization of usage, questions about the usage and implementation of the system. Table 1. Modules with Implementation Dates Module Patient Registration and scheduling Patient Financial assessment Human Resources Physician order entry Pharmacy with inventory Radiology process and reporting Medical records Pathology Modules Outcome analysis module Financial Module Inpatient unit ( admission and discharge) Surgical module Blood bank Critical Laboratory alerts Physician’s notes Patient billing Endoscopy Material management, procurement life cycle Fixed assets management Radiation oncology Chemotherapy Cardiology

Implementation Date April 2001 June 2001 January 2002 February 2002 February 2002 April 2002 April 2002 November 2002 – February 2004 December 2002- January 2004 March 2003-Decmber 2004 May 2003 May 2003 March 2004 2005 May 2006 December 2006 December 2008 December 2009 December 2009 December 2010 December 2011 December 2012

Source: Development of an in-house hospital information system in hospital in Pakistan, International Journal of Medical Informatics 83 (2014) 180-188

Manual Information management System: This section aimed at getting information about the manual mechanism used to store, retrieve and manage the information system. The Survey was carried out for first author to visited Shaukat Khanum Memorial Cancer Hospital and Research Center (SKMCH&RC), Lahore in the early August 2014. Two days lived at hospital premises and carefully watch every ICT system there and conduct Interview with every sections heads. According to survey following results are come. The SKMCH&RC has been operational since 20 years. There are 2000+ employees in it; from them 500 have access to computers. Staffs that access the computers are unit coordinators, patient care officers, Nursing staff, Physicians, Technologists, Technicians and lab staff. All computers are connected with Local Area Network (LAN). 100% computers are connected with Gigabit Ethernet, ATM and WAN, Wi-Fi. The purposes of network usage are E-mails, Printing the documents and lab reports, File sharing, web access and other applications. The IT system architecture is file server and client server system with web based. Hospital also access internet. Internet is access through LAN and DSL. 70 % of computers access it. Internet used for access of HMIS, Browsing, Social Media and Research work. The maintenance of system is responsibility of System and Network Maintenance Departments.

405

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 The name of HIS is “In-house developed integrated Hospital Information Management Systems”, which covers all departments and functions of hospital. It also includes Customer/Patients Relationship Management System (CRMS). Individual staffs used client server and web interface to interface with system to interfacing Pathology and Radiology equipment’s. Computer network allow access to the system from multiple points. Data get entered into system using HMIS or integrated interface. Quality checks of system are carried out. The following types of check used (1) Ranges (2) Valid values, Reasonable values (3) Decision Support System and suggestions from different sectors. Staff training for use of HMIS service different techniques used, such as, In-house training, hands on training and Formal class room trainings arranged. There are 1400+ hardware items and property software’s (HMIS and PACS) are used. Application system development cost is 2 million US$ annually and maintenance cost of all components (recurring) is 1 million US$ annually. The expenses during development are shown in Table 2. Table2. HIS Development costs in US Dollars Expense Category 2001-2006 2007 2008 2009 2010 2011 Total Salaries and benefits 260,987 111,205 160,935 223,25 246,72 296,49 1,299,594 Hardware 26,567 5889 9096 10,949 24,789 37,513 114,804 Maintenance 12,907 6299 9866 10,637 97,789 10,689 60,119 Licensing and Professional 21,065 3873 3543 5361 1022 15,356 50,221 Supplies, Stationary etc. 28,531 3992 7693 5157 8063 19,742 73,177 Total 350,057 131,25 191,133 255,35 290,31 379,79 1,597,915 Source: Development of an in-house hospital information system in hospital in Pakistan, International Journal of Medical Informatics 83 (2014) 180-188 After full functional of HMIS at hospital, it saves lot of printing paper, Radiology films cost and staff. The staff was decreased in number at selected sections where system is fully working. HMIS implementation saves time and increased productivity. The table 3 shows the savings in US$ after HMIS working. Table3: Savings in US Dollars 2001-2006 2007 2008 2009 Saving Paper Printing 98,283 46,053 51,978 69,570 Radiology 142,403 147,368 18,947 189,474 Manpower 415,403 307,216 491,736 588,044 Total 655,791 500,638 722,661 847,088 Source: Development of an in-house hospital information system Journal of Medical Informatics 83 (2014) 180-188

2010 2011 87,509 127,620 273,684 284,211 756,762 878,518 1.117,955 1,290,34 in hospital in Pakistan,

Total 481,014 1,215,789 3,437,678 5,134,482 International

Until now our survey shows that the SKMCH&RC HMIS project gets excellent result in relations of investment and asset utilization It is valuable to consider the majority of the measures ( payback, NPV,IRR) on the grounds that decision makers with a different piece of relevant information. Distinctive measures give diverse data. Figuring or measurement may help; more weight may be given to one measure over other, however it would be indiscreet to disregard the data gave by different strategies. [2] The HMIS of SKMCH&RC is not implied for re-venture, in this manner NPV is a more precise measures of its esteem. This was computed by discovering the present estimation of all inflows and surges and after that marked down at the HMIS task expense of capital. The HMIS venture NPV was computed by taking the aggregate of all reduced money flows. [2] NPV estimation is abridged in Table 4.

Conclusion This study demonstrates that the introductory expense of actualizing a paperless framework can be high however an electronic HMIS is savvier than paper records. Furthermore, the inventory cost of paper records and salary of staff, file and maintain them is eliminated with electronic records. An electronic HMIS can be

406

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 startling at the outset;; however succh feelings arre normally brrief, once thee profits that a paperless reecord t be clear to all. a SKMT hosspital is paperr less now. framework offers get to Table4: NE ET Present Vaalue calculation Summarizttation.

Source: Development D off an in-house hospital h inform mation system in i hospital in Pakistan, P Interrnational Journnal of Medical Innformatics 83 (2014) ( 180-188

Referen nces 1. 2. 3. 4. 5.

6.

Tuula Kivinen,, Johanna Lamm minta, The succcess of a manaagement inform mation system in n health care- A case study from Fin nland, Internatioonal Journal of Medical M Inform matics, 82 (2013 3) 90-97. Faisal Sultan, et al, Development of in--house hospitaal information system in hoospital at Pakkistan, International Joournal of Mediccal Informatics,, 83[2014] 180--188. D. Naranjo-Gill, F. Hartmann,, How CEOs usse managementt information syystems for strattegy implementtation in hospitals, Heealth Policy 81 (1) (2007) 29––41. C.P.Lamdiigan n, et.al, Temporral trends in raates of patient harm resultingg from medical care. N.Engl.JJ,Med [2010]. mmunity Health h Sciences, Thhe Aga Khan University, U Karaachi.), Health Sector S A. Islam (Department of Com Reform in Pakiistan: Why is itt needed? (JPMA A 52:95; 2002)). Maria Lluch, Healthcare prrofessionals’ organisational o b barriers to heaalth informatioon technologiees—A w,Internationall Journal of Med dical Informaticcs (2011), pp. 849–862. 8 literature review

407

ISBN: 978-969-8680-32-9


Performance analysis of cross layer protocols for Flying Ad-hoc Networks Wajiya Zafar1, Bilal Muhammad Khan2 1

Dept. of Electronics & Power Engg., National University of Sciences and Technology (NUST), Islamabad, Pakistan

[email protected] 2

Dept. of Electronics & Power Engg., National University of Sciences and Technology (NUST), Islamabad, Pakistan

[email protected]

Abstract This paper addresses the challenges of low latency and high reliability in Flying Ad-hoc Networks (FANETs) by presenting first of its kind comprehensive analysis of MAC and routing protocols. Multiple Unmanned Aerial Vehicles (UAVs), operating autonomously as a part of FANETs, are expected to show exceptional promises in military, civilian and commercial surveillance applications. Low latency and high reliability are the key requirements of such applications. In this paper experimental performance evaluation of FANETs is presented to determine the suitability of cross layer approach in which communication is carried via IEEE 802.11 and 802.15.4 MAC layer protocols and AODV, DSDV and AOMDV are used as routing protocols. Reference Point Group Mobility (RPGM) model has been used to simulate swarm of UAVs. A comparative analysis of the network layer protocols has been done for each MAC layer protocol on the basis of packet delivery ratio, average delay and energy consumption. Comprehensive comparative analysis shows that 802.11 outperforms 802.15.4 in terms of latency and energy consumption in a FANET scenario. Keywords: Flying Ad-hoc Networks, 802.11, 802.15.4, routing, latency, reliability, UAVs

Introduction Setting up an ad-hoc network amid UAVs is a relatively new and emerging research area. When multiple UAVs are made to communicate and coordinate such that the advantages of increased range, fast and reliable response time and low cost are obtained, the resulting networking model is called a Flying Ad-Hoc Network. Such networks are expected to not only automate military and civilian applications but also boost up the performance of commercial applications such as wind estimation, traffic monitoring, remote sensing and relaying purposes [1,14]. When UAVs are used as mobile nodes in Ad-Hoc networks, several demanding challenges are encountered. A lot of research is being done to minimize these design constraints. The main challenge is communication because of very high mobility, frequent change in topology and large distances between FANET nodes. According to [1], typical speed of UAV is 30-460 km/hr which requires a highly challenging communication protocol design. The initial research on FANET is being carried out using existing MANET protocols. The IEEE 802.15.4 protocol is widely known for low power, low data rate wireless personal area networks (WPANs). This standard is highly configurable and offers flexibility to optimize its parameters as per the application needs. For bandwidth hungry FANET applications such as search and destroy operations, border surveillance and traffic monitoring, IEEE 802.11 can offer maximum advantages because it supports bandwidth upto 11Mbps which can effectively accommodate multimedia images and videos. On the other

408

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 hand wildfire management, remote sensing and relaying networks require control information to be exchanged within the network. Such applications do not necessarily require a high data rate protocol such as 802.11. UAV to UAV communication can be carried efficiently by deploying a low data rate protocol such as IEEE 802.15.4 which operates at a raw data rate of 250kbps. In this paper, we have analyzed the effect of routing protocols for UAV to UAV communication in FANET environment for IEEE 802.11 and IEEE 802.15.4. The novelty of our work lies in simulating a swarm of UAVs using RPGM model on IEEE 802.15.4 which as to this writing has not been investigated for FANETs. We have studied the effect of high speeds as required by FANET nodes for different routing protocols, i.e. Ad-hoc On Demand Distance Vector (AODV), Destination-Sequenced Distance Vector (DSDV) and Ad-hoc On Demand Multipath Distance Vector Routing (AOMDV).

Related work The research available is targeted to the optimization of commercial off the shelf hardware components to meet the challenges of FANET MAC layer. In [2] throughput and link quality measurements are made for single hop and multi-hop communications involving UAVs and ground station by deploying multi-antenna extensions to 802.11a. The authors have performed a comparative analysis of infrastructure and mesh modes of 802.11. The results show that in order to support high throughput and low jitter applications, the standard mesh protocol is insufficient and need to be improved for carrying out autonomous operation of multi UAV systems. In [3] Adaptive MAC protocol scheme for UAVs (AMUAVs) has been proposed in which both omnidirectional and directional antennas are used for improving throughput and end to end latency for multi UAV systems. In [4] a new MAC scheme called Token-MAC is introduced in which fullduplex radio and multi-packet reception (MPR) are used to obtain improved Chanel State Information (CSI) for the UAVs. The scheme showed improved results for throughput and latency. Various reactive, proactive and position-based routing protocols have been tested for FANET environments with position based routing outperforming the others [5]. In [6], Topology Broadcast based on Reverse-Path Forwarding (TBRPF) is used to minimize routing overheads in communication between UAVs. In [7,8] Dynamic Source Routing (DSR) protocol which is a reactive protocol has given promising results for high mobility and frequent changes in topology. Optimized Link State Routing Protocol (OLSR) has been improved to Directional Optimized Link State Routing Protocol (DOLSR) by the use of directional antenna which resulted in using less number of multipoint relay (MPR) nodes and thus reducing latency factor [12]. In [9], the authors have used Ad-hoc On-demand Distance Vector Routing (AODV) protocol with a time slotted implementation. The results showed an improved packet delivery ratio (PRR).

Simulation Methodology The ultimate goal of this research is to compare the performance of routing protocols for IEEE 802.11(with RTS/CTS on) and IEEE 802.15.4(beacon-enabled mode) based FANETs.The node speeds are varied using Reference Point Group Mobility model (RPGM). It is described as a group model in which nodes move in a group. The movement of the nodes is governed by a logical center known as the group leader. The coordinated speed, acceleration, location and direction of the mobile nodes are in accordance with the movement of the group leader [10]. In beacon enabled mode, a duty cycle operation is provided by IEEE 802.15.4 to achieve low power operation. Two parameters that are used to set desired duty cycle are macBeaconOrder (B0) and macSuperframeOrder (S0). FANETs do not encounter energy resource constraints; hence 100% duty cycle operation is carried in our simulations. Superframe order is related to

409

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 the Superframe Duration (SD) i.e. the active period of the node during Contention Access Period (CAP) and beacon order is used to define Beacon Interval (BI) by the following set of equations: Beacon Interval (BI) = aBaseSuperFrameDuration * 2B0

(2)

Superframe Duration (SD) = aBaseSuperFrameDuration * 2S0

(3)

Where aBaseSuperFrameDuration = Number of slots * Slot duration, for which default values of 16 and 60, respectively, are used. The duty cycle is calculated by the ratio of active period (SD) of the superframe to beacon interval (BI). Duty cycle = SD/BI*100% = 2(BO-SO) = 100% (B0=S0=5)

(4)

The simulations for beacon enabled mode are run using S0=B0=5. This means that the change of superframe order and beacon order takes place simultaneously. The inactive period has been eliminated to achieve higher packet delivery ratios. The default implementation of 802.15.4 in ns2 does not have GTS support. To address this issue we have implemented TDMA (Time Division Multiple Access) in the entire active period i.e. superframe duration. Each node in the network reserves time slots before transmission of data; as the packets are generated nodes in their respective reserved allocated time slots start transmitting In FANETs a line of sight is maintained between UAVs [1]. Hence, the radio propagation model used is free space model as it works on the assumption of a single clear line of sight between transmitter and receiver. In our simulations we assume that the channel is perfect i.e. Bit Error Rate (BER) is zero.

Fig. 1. Network Topology

The mobility scenarios have been generated using bonnmotion utility which creates, analyzes and exports mobility scenarios to different network simulators [11]. Simulation studies have been carried on ns-2.35 network simulator which incorporates all the protocols. Every simulation run generates a trace file which is processed using awk language script to obtain throughput, average delay and packet delivery ratio. To have a fair comparison, the transmission range is set to 75m for both protocols using the threshold utility provided in ns2. The maximum distance to the group center is 70m in RPGM model. The topology consists of a single cluster of 3 nodes that are deployed in an area of 100m x 100m. AODV, DSDV and AOMDV routing protocols are used for both 802.11 and 802.15.4 protocols. The data rate chosen for simulation is 50 kbps. The performance evaluation parameters used for our FANET topology are: Packet Delivery Ratio (PDR) It is the ratio of total data packets delivered to the destination node to the data packets generated by the source node. Higher value of PDR corresponds to low packet loss rate and thus reliable communication. Packet Delivery Ratio (PDR) = PacketsRx / PacketsTx

(5)

Where, PacketsRx = Total data packet delivered at the destination PacketsTx = Total data packets generated by the source

410

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Table 1. Simulation Parameters Parameter Simulation tool Simulation time Propagation model MAC Protocol Channel capacity 802.11 Channel capacity 802.15.4 Simulation Area Routing Protocol Traffic Number of nodes Transmission Range CBR rate Packet size Speed Mobility model

Value NS-2.35 100 FreeSpace 802.11, 802.15.4 11 Mbps 250 kbps 100 x 100 AODV, AOMDV, DSDV CBR 3 75m 50 kbps 100 bytes 10,20,30,40,50,60,70,80,90,100 Reference Point Group Mobility Model

Average End-to-End Delay The average time it takes for the successful transmission of a data packet from source node to destination node. This parameter includes all delays: End-to-End Delay= Tt+Rt+Bt+Prt

(6)

Where, Tt = Transmission time Rt = Retransmission time Bt = Buffering time Prt = Processing time Energy consumed (in joules) The average energy a mobile node consumes transmitting and receiving the data packets. It is given calculated by: Energy consumed = Initial energy – Remaining energy at each node

(7)

Results and Analysis Figure 2, figure 3 and figure 4 show average delay of the network for AODV, DSDV and AOMDV, respectively. It can be seen that for all three routing protocols average delay of 802.15.4 based network is more than that of 802.11 due to signaling overhead in beacon enabled mode in 802.15.4. The beaconing coordinator reacts slowly to a beacon request due to longer superframe. IEEE 802.11 outperforms 802.15.4 because it supports a high data rate of 11 mbps resulting in minimal MAC level retransmissions, a larger contention window, and does not suffer beacon overheads. The average delay for DSDV is less than the other two protocols. DSDV being a table driven protocol outperforms AODV when it is used in a small scale network as in our case. Less overhead is encountered in the routing table of DSDV since it proactively keeps routes to all destinations in its routing table whereas in

411

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 AODV and AOMDV reuse of routing information does not occur and route discovery process is initiated again and again leading to higher delays.

Mac 802.15.4 Mac 802.11

Average Delay AODV Average Delay seconds

0.035 0.03 0.025 0.02 0.015 0.01 0.005 0

10

20

30

40

50

60

70

80

90

100

Speed m/s

Fig. 2. Average Delay versus speed for AODV Mac 802.15.4 Mac 802.11

Average Delay DSDV Average Delay seconds

0.015 0.013 0.011 0.009 0.007 0.005 0.003 0

10

20

30

40

50 60 Speed m/s

70

80

90

100

Fig. 3. Average Delay versus speed for DSDV Mac 802.15.4 Mac 802.11

Average Delay AOMDV Average Delay seconds

0.017 0.015 0.013 0.011 0.009 0.007 0.005 0.003 0

10

20

30

40

50

60

70

80

90

100

Speed m/s

Fig. 4. Average Delay versus speed for AOMDV

Figure 5, figure 6 and figure 7 show both 802.11 and 802.15.4 can provide high reliability in a FANET scenario. The 100 % reliability exhibited is due to the mobility model selected. The Reference Point Group Mobility Model (RPGM) sets a specific distance between nodes such that they all are in the radio range of each other during the network operation. Since the movement of UAVs is pre-planned so the mobility model suppresses the random movement of nodes allowing them to move as a cluster.

412

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 MAC… MAC 802.11

Packet Delivery Ratio AODV 100

PDR

80 60 40 20 0 0

10

20

30

40

50 60 Speed m/s

70

80

90

100

Fig. 5. PDR versus speed for AODV MAC 802.15.4 MAC 802.11

Packet Delivery Ratio DSDV 100

PDR

80 60 40 20 0 0

10

20

30

40

50 60 Speed m/s

70

80

90

100

Fig. 6. PDR versus speed for DSDV MAC 802.15.4

Packet Delivery Ratio AOMDV 100

PDR

80 60 40 20 0 0

10

20

30

40

50 60 Speed m/s

70

80

90

100

Fig. 7. PDR versus speed for AOMDV

Figure 8, figure 9 and figure 10 show the percentage of energy consumed during the network operation. Energy used by the nodes is much higher in 802.15.4 then in 802.11. Although 802.15.4 is designed for energy saving but due to the same values of S0 and B0 the duty cycle is 100% which means that there are no sleep cycles and thus energy saving benefits are not obtained. Since 802.15.4 is designed for a specific transmission area so its performance deteriorates when the range increases as more power is required to transmit at a longer range. The topology considered is single hop network with no intermediate nodes. Thus, transmission between nodes covers an entire distance of 70m from source to destination without benefitting from energy saving properties of relaying nodes. Among the routing protocols, AODV shows winning behavior in terms of energy consumption since there are no periodic route updates. In case of DSDV the routing table updates are transmitted periodically at an average distance of 70m between the nodes resulting in high energy consumption. However, in FANETs

413

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 energy is not an issue because the communication hardware of FANETs can easily be supported by the power resource of the UAVs. [1]

Percent Energy consumed J

Mac 802.15.4 Mac 802.11

Percent Energy consumed AODV

2.9 2.7 2.5 2.3 2.1 1.9 1.7 0

10

20

30

40 Speed m/s 50 60

70

80

90

100

Fig. 8. Percent Energy consumed versus speed for AODV Mac 802.15.4 Mac 802.11

Percent Energy consumed DSDV Percent Energy consumed J

2.95 2.9 2.85 2.8 2.75 2.7 2.65 2.6 2.55 0

10

20

30

40 Speed m/s 50 60

70

80

90

100

Fig. 9. Percent Energy consumed versus speed for DSDV

Percent Energy consumed J

Mac 802.15.4 Mac 802.11

Percent Energy consumed AOMDV

3.2 3.1 3 2.9 2.8 2.7 2.6 2.5 2.4 2.3 0

10

20

30

40

50 60 Speed m/s

70

80

90

100

Fig. 10. Percent Energy consumed versus speed for AOMDV

Conclusion This paper presents detailed comparative analysis of AODV, DSDV and AOMDV for two MAC protocols: 802.15.4 and 802.11 in a typical Flying Ad-hoc Network (FANET) scenario of UAV to UAV communication. The simulation results show that in such particular scenario IEEE 802.11 out performs

414

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 IEEE 802.15.4 in terms of latency with DSDV as the most suitable choice of routing protocol alongwith Reference Point Group Mobility Model (RPGM). IEEE 802.15.4 fails to provide acclaimed benefits in FANET scenario with higher average delays and energy consumption. However, the mobility model used makes both 802.11 and 802.15.4 equally reliable. Furthermore, the results also suggest that 802.15.4 can be further improved to be used in FANETs which does not require energy saving protocols due to intense hardware capabilities. A future research is possible to evaluate the feasibility of 802.15.4 for FANETs by lowering average delays using a cross layer approach along with certain improvements.

References 1. 2. 3. 4.

5.

6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

Bekmezci, I., Sahingoz, O.K., Temel, S., Flying Ad-Hoc Networks (FANETs): A survey, Ad Hoc Networks vol. 11, no. 3, pp. 1254-1270 (2013) Yanmaz, E., Hayat, S., Scherer, J., Bettstetter, C., Experimental Performance Analysis of Two-Hop Aerial 802.11 Networks, IEEE WCNC, pp. 3118-3123 (2014) Kung, H.T., Lin, C.-K., Lin, T.-H., Tarsa, S.J. , Vlah, D., Measuring diversity on a low-altitude UAV in a groundto-air wireless 802.11 mesh network, In: IEEE Globecom Workshops, pp. 1799-1804 (2010) Cai, T., Yu, F.R., Li, J., Zhou, Y., Lamont, L., MAC performance improvement in UAV ad-hoc networks with full-duplex radios and multi-packet reception capability, In: Proc. of IEEE International Conference on Communications, pp. 523-527 Ottawa, Canada (2012) Hyland, M.T., Mullins, B.E., Baldwin, R.O., Temple, M.A., Simulation based performance evaluation of mobile ad hoc routing protocols in a swarm of unmanned aerial vehicles, In: Proceedings of the 21st International Conference on Advanced Information Networking and Applications Workshops, Vol. 02, pp. 249–256, IEEE Computer Society, Washington, DC, USA, (2007) Bellur, B., Ogier, R.G., A reliable, efficient topology broadcast protocol for dynamic networks, In: Proceedings of 18th Annual Joint Conference of the IEEE Computer and Communications Societies, vol. 1, pp. 178–186 (1999) Brown, T., Doshi, S., Jadhav, S., Henkel, D., Thekkekunnel, R., A full scale wireless ad hoc network test bed, In: Proceedings. of International Symposium on Advanced Radio Technologies, pp. 50–60, Boulder, CO, (2005) Khare, V.R., Wang, F.Z., Wu, S., Deng, Y., Thompson, C., Ad-hoc network of unmanned aerial vehicle swarms for search & destroy tasks, In: Intelligent Systems, 4th Int.l IEEE Conference , vol.1, pp. 6-65,6-72, (2008) Forsmann, J.H., Hiromoto, R.E., Svoboda, J., A time-slotted on-demand routing protocol for mobile ad hoc unmanned vehicle systems, SPIE, vol. 6561, (2007) Roy, R.R., Reference Point Group Mobility, Handbook of Mobile Ad Hoc Networks for Mobility Models, pp. 637-670, (2011) Aschenbruck, N., Ernst, R., Gerhards-Padilla, E., Schwamborn, M., BonnMotion - a Mobility Scenario Generation and Analysis Tool, In: Proceedings of the 3rd Int. ICST Conference on Simulation Tools and Techniques, (2010) Alshabtat, A.I., Dong, L., Li, J., Yang, F., Low latency routing algorithm for unmanned aerial vehicles ad-hoc networks, International Journal of Electrical and Computer Engineering vol. 6, no. 1, pp. 48–54, (2010) Sahingoz, O.K., Mobile networking with UAVs: Opportunities and challenges, International Conference on Unmanned Aircraft Systems (ICUAS), pp.933-941, (2013) Cetin, O. Zagli, I, “Continuous Airborne Communication Relay Approach Using Unmanned Aerial Vehicles”, Journal of Intelligent & Robotic Systems, vol. 65, no. 1, pp. 549-562, (2012) Lin, L., Sun, Q., Li, J., Yang, F., A novel geographic position mobility oriented routing strategy for UAVs, Journal of Computational Information Systems, vol. 8, pp. 709–716 (2012)

415

ISBN: 978-969-8680-32-9


A Simulink Based Adaptive UFLS Scheme Syed Zia-ud-deen1, Anzar Mahmood2,*, Guftaar Ahmed Sardar Sidhu1, Ahmed Bilal Awan2 1 2

EE, COMSATS Institute of Information Technology, Islamabad, Pakistan EE, College of Engineering, Majmaah University, Al-Majmaah, Kingdom of Saudi Arabia

Abstract This document is in the required format. Power system stability is affected by different variations like over/under voltage, current, and frequency. Among these, frequency is highly sensitive parameter which should be controlled in a sophisticated manner in order to ensure smooth system’s operation. This paper presents a brief review of Under Frequency Load Shedding (UFLS) schemes and a Simulink model to implement UFLS in a new way. Frequency can be used to indicate an imbalance load condition as it remains same at all voltage levels. Whenever frequency drops from its actual value (50 Hz or 60 Hz), load shedding is required to avoid imbalance between load and generation. In this paper adaptive UFLS control in islanded region is presented in which swing equation is used to calculate the imbalance and stability conditions. Depending upon the calculations, automatically, the imbalance power (extra load) is shed to keep the frequency of an islanded region constant or a generation source is added from distributed generation/storage to maintain the stability of the system. Key Words: UFLS, Power system, Stability

Introduction For stable operation of a power system, generation must be equal to the sum of all connecting loads and losses. There are different parameters of power system (over current, over/under frequency and voltage levels) which affects the power system stability and in severe cases cause complete collapse or blackouts. It is impossible to eliminate faults and failures completely in power systems, however, faults and failures can be reduced. If faults and failures are not reduced or controlled, the disturbance due to them in a system will be changed into major incidents that cause severe outages and system collapses which will affect the consumers as well as the utility in financial terms. To avoid this condition, the power system must be designed to withstand unexpected failures and continue operating with minimum interruption of supply to the consumers. In power system active and reactive power play important role to control the imbalance of the system. The frequency and the voltage are the two important parameters to run the system in stable state. Frequency depends on active power while voltage depends on the reactive power. Thus any change in active or reactive power has direct influence on the stability of the power system. Constant frequency is significant to ensure the appliances within the safety limits such as constant speed of motors (induction and synchronous). To keep the frequency of power system constant, input power (mechanical) to the prime movers of a generator must be equal to the summation of the entire connected loads and all active power losses. This equilibrium is disturbed, if input or output change. Mathematically stable condition is define as (1)

Generation = load + losses

Sudden losses of generator, transformer, and transmission line produce a non-equilibrium state in which load will become greater than the generation, causing a frequency depress. The relation between system frequency and load are shown in Table 1.

Why islanding UFLS control The use of distributed generation (DG) is increasing day by day which will cause increased islanding operation at distribution level. It is expected in future that the whole region (national grid) will be divided in to a number

416

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 of separately controlled islanded regions. Like a national grid (inter connected system), islanded regions also require effective protection schemes. For stability purpose the generation and load of an islanded region must be balanced. The most important protection scheme for islanded region is under frequency load shedding scheme. Furthermore, the reliability of islanded region can be improved by continuously providing power from DG or by storage system [4]. Table 1. Effect of load on frequency Condition of the system

ncy of the system

Generation > demand + losses

ncy increase

Generation < demand + losses

ncy decrease

Generation = demand + losses

ncy unchanged

Related work Several items must be considered while designing or improving a load-shedding scheme. The first important thing is frequency, i.e. When 50 Hz frequency is used, lower limit is 47.5 and upper is 52.5 (as ±5% frequency fluctuation is allowed to operate power system in stable state, frequency higher or lower then this effects power system stability). In certain conditions UFLS is the only technique to stop power system from collapsing [1]. Implementing recent techniques and new technologies into power system depends on the economy of a country. Automation and integration of communication technologies into the system will bring improvement. It is important in power system to know about the actual state of a system, by knowing certain important parameter (voltage, frequency, active and reactive power), automatic regulation becomes possible [9]. To control UFLS there are many different methods available in the literature. One thing is common among all, which is frequency; because frequency is the only important parameter which shows, the required power to be shed or power imbalance [7]. Majority of the existing methods used by power system's operators are of traditional types, which result in more grid failure (in Pakistan, India recently). Traditional UFLS schemes do not really encompass the disturbance magnitude in the system and often shed more or less load than required. In literatures different methods of dynamic or adaptive UFLS are suggested and discussed. Swing equation is used in majority of adaptive UFLS schemes. Initial ROCOF (rate of change of frequency) is used as basic parameter for the setting of under frequency relays [3, 8]. In [13], the author explained that existing load shedding schemes are not efficient and reliable because of not estimating the power deficit accurately. A new technique is presented in this paper, which use swing equation (equation 2) to find magnitude of disturbance. Also this new technique improve consistency (supply to consumer) and to maintain stability using communication infrastructure. (2 * H/ f) * df/dt = Pm - Pe

(2)

where, H = inertia constant of a generator in seconds, f = normal value of frequency (Hz), Pm = mechanical power pu, Pe = electrical power in pu In [2], comparison between different UFLS schemes is presented. The first of which is traditional UFLS scheme in which frequency threshold is used to shed loads, semi adaptive method used frequency and rate of change of frequency (ROCOF) and an adaptive method has also used because of its importance and efficiency compared with the other two in modern grid. The semi adaptive schemes have superior performance than traditional schemes when the magnitude of disturbance is small; performance is very similar to traditional method for large disturbance in a system. The complexity of power system is increasing due to high demand (always increasing) and different variety of loads, as a result the power system operation is less secure. Moreover, the use of electronic machine is increasing day by day, which affect the power quality and more burdens on generation. To balance generation and load a storage system is used [10]. This storage system is called super capacitor energy storage system (SCESS). SCESS provide power to the consumer when the load is greater than generation. Hence SCESS is used to overcome the load on the system and stabilize the frequency. The usage of distributed generation (DG) at distribution level has increased significantly. The advantage of integrating DG, in distribution network is continuous supply of power even when disconnected from the main grid. DG energizing distribution network in islanding mode has become an emerging research topic. In islanding

417

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 mode DG G acts like a main m grid coveering a small portion. p As prootection of griid is importan nt therefore a protection p scheme ffor an islanded network is i also required which is still an issuee to be solvedd. Further ann islanded distributiion network is i more sensiitive to frequuency, and theerefore requirres a more fast fa and reliabble UFLS scheme aas compared to o the existing power networrks [5, 6]. For islannding mode a different UFL LS strategy iss required, because UFLS scheme alread dy used in larrge power system caannot be used d here, as bothh systems hav ve different chharacteristics. In islanded mode m for UFLS S scheme, communiication betweeen componeents is not essential. e Wheen system frrequency drops it is stabbilized by automaticcally shedding g a suitable looad based on previous loadd history. Thee authors also mentioned thhat load is prioritizeed. The load of the consumeer who will paay more will shed after thosse who pay lesss. [9]. In [13], a new intelligeent UFLS schheme is presennted. The mainn objective off this scheme is i to retain staability and to enhannce the reliabiility of system m in a distrib bution networrk. Loads aree prioritized first, f when diisturbance occurs, thhe magnitude of disturbancce is estimated by using sw wing equation n. Less importtant loads are shed first and then in a sequencee of increasingg importance. t authors ex xplain that thee consideratioon of only onee parameter i.e. frequency gradient, to find fi active In [12], the power deeficit in an islanded mode is not enouggh. Other impportant parameeters such as spinning reseerve, total system innertia etc. cannnot be ignoreed when dealinng with UFLS S. A smart UF FLS is propossed in [11] in which w the whole areea (power sysstem) is divideed into five power districtss. Every power district hass generating sttation and load (connsumers). Theese power disttricts are connnected with eaach other thro ough transmisssion lines. Eacch district is equippped with smaart meter andd a power defficiency meteer (deficiency y can be calcu ulated by usiing swing equation)). Measuring device (smarrt meter) has the capabilityy of communnication with other smart meter m and knows thhe actual statee of the systeem like the consumption c o active pow of wer in each arrea and the amount a of generatioon. “Operator”” is introduceed as a gateway to which all a the measurring instrumeents are conneected. The "Operatoor" has full conntrol on the syystem and also o knows the current c state of the system (area ( wise actiive power consumpption). When deficiency appears ap in anny of power districts (duee to trip of generation g orr fault on transmisssion line) imbalance takes place p betweenn generation annd demand in that district.

Propossed UFLS Scheme S There aree many differrent techniquees and methoods available iin literature to t design and control UFL LS. In this paper, thhe effect of load (active power) on frequuency is demoonstrated usinng Simulink and a Sim poweer system. Sim Pow wer Systems work w togetheer with Simullink software to model ellectrical systeems. It operattes in the Simulinkk environmentt and providess different com mponent for simulating s eleectrical powerr systems. Furrthermore, different methods are presented to control loadd shedding in islanded regiion. In this paper p we deveeloped an Under Frrequency Conntroller (UFC)) which is cappable of calcuulating deficitt power (usinng swing equaation) and sheds thee load accordiingly. It is also capable of adding a extra generation g (orr storage systeem) into the nnetwork if availablee as discusse ed in sec 4.2. This con ntroller prevvents the sy ystem from blackouts and a also

providees a protection scheme for the pow wer system. Determ mining poweer imbalancee Generatoor rating: Generatoor of 125 kVA; Output powerr = 100 kW (00.8 power facttor) Frequency (f) = 50 Hzz; Number of Poole (P) = 2 Initial Load = 100 kW Speed = 3000 rpm; l is increassed from 100 KW to 108.5 KW. As a The geneerator is operaating at full loaad (100 KW) initially. The load result, off extra load, frrequency dropps. In order to calculate an iimbalance pow wer swing equuation is used i.e. equ # 3. (3)

The valuue of ∆P is thee power imballance. By sheddding the amoount of load equal e to ∆P, th he frequency will w come to its norrmal value (50 Hz). An im mbalance poweer calculatingg block is dev veloped, by ussing inertia coonstant H, frequencyy and its deriivative. In this demo the amount of incrreased power is 8.5 KW, power p imbalannce block calculates and show itt. After removving 8.5 KW of o load, frequuency comes to its normal value. v The freequency is falling coontinuously as a shown in figure fi 2, it is because the load l connecteed to the geneerator is moree than the

418

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 generatioon. An imbalaance power off 8500 W is calculated. Affter shedding load equal to o imbalance ppower the frequencyy comes to itts normal vallue. Frequenccy response att high load and a when load d is removed from the system iss shown in figgures 2 and 3, respectively. The T system iss also tested, when w total loadd at the system m is lower than the ggeneration. As A a result an imbalance i pow wer of -9400 W is calculateed. To stabilizze the frequenncy, a load equal to 9400 W is addded. Frequenncy response at low load iss shown in figure 4. The result r is summ marized in Table 2.

F 1. Calculating an imbaalance power of Fig. o over load syystem

Fig. 2. 2 Calculatinng an imballance powerr of over loaad system

Figg. 3. Frequenccy after load shedding

419

ISBN: 978-969-8680-32-9


Fig. 4. Frequuency at low load l Providin ng an extra po ower when looad is increassed Generatoor rating: Generatoor of 500 kVA (3 generatorss); Outpput power = 1500 kW Frequenccy (f) = 50 Hzz; Num mber of Pole (P P) = 2 Speed = 33000 rpm ; Initial Load = 10000 kW The mechanical power (Pm) increaases from 10000 KVA to 15000 KVA as shhown in figurre 5. This is because b of l of 1000 KW K is conneccted. Two geneerators are suppplying, whilee the third load incrrease at t=4 seec. Initially a load generatorr (DG or storrage system) is on standbyy mode. At t= =4 sec, circuit breaker is closed and thhe load is increasedd from 1000 KW to 1500 KW. To keeep the frequenncy constant, a controller adds a sourcce (DG or storage system) to the system. The controller subbstrate the currrent load from m the total ouutput power caapacity of p can the generrators, the erroor signal is givven to the thirrd generator as an input mecchanical poweer. The extra power also be ccalculated by using u swing equation e as discussed previously. A pow wer source (DG G or storage system) s is added too the system automaticallyy to fulfill the demand annd to keep thhe frequency of the islanded region constant. The results are a shown in figures 6 & 7. 7 In figure 7, 7 rotor speed starts to decrrease and thenn become stable; deecrease is duee to extra loadd added to thee system (systtem overloadeed). In figure 6, the frequenncy of the system iss shown, after oscillations itt becomes stabble. The results are summarrized in Table 3.

Conclu usion The pow wer quality as well as system m componentts are affectedd when system m frequency chhanges. Loadd shedding schemes based on freq quency alone (traditional load l sheddingg schemes) haave several diisadvantages including: i o power deficcit. In this offline deecision to sheed load, not coonsidering thee level of distturbance and the amount of paper, wee have implem mented a moddified adaptive under frequuency load sheedding (UFLS S) method in which we considereed the power imbalance annd shed the looad accordingg to it. Anothher advantage of the adaptiive UFLS scheme is that it shedss loads in one step as compaared to the traaditional load shedding scheeme. The timee required to stabilize the power system in adaaptive UFLS is less as com mpared to tradditional load shedding s scheemes, as it sheds loaads in one stepp. Some of thhe issues that have h not beenn explored in this t work and d can further eextend the current work w include: detection of isslanded regionn followed byy controlling and a stabilizatioon (generationn and load balancingg) and subsequ uently integraation of the en ntire islanded region r with thhe grid.

420

ISBN: 978-969-8680-32-9


Fig. 5. Automatically adding a generator



421

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Table 3. Font sizes of headings. Table captions should always be positioned above the tables. Load

Frequency

Extra Generation

1000 KW 1000 KW

Constant

Not added

1000 KW 1500 KWop and trip the system

Not added

Available Generation

1500 KW 1500 KW

constant

(1500 – 1000 = 500 give as an input signal to the extra generator in the system by the ntroller or provide power from any storage system equal to imbalance power)

References 1. Mohd Zin, H. M. (2004). Static and Dynamic Under-frequency Load Shedding A Comparison. IEEE International Conference on Power System Technology, (pp. 941-945). 2. B. Delfino, A. M. (2001). Implementation and comparison of different under frequency load-shedding schemes. IEEE Power Engineering Society Summer Meeting, (pp. 307-312). 3. Fox, J. G. (1994). Adaptive load shedding for isolated power system. 1EEE Proceeding on Generation Transmission Distribution, (pp. 491-496). 4. J. A. Laghari, H. M. (2012). An Intelligent Under Frequency Load Shedding Scheme for Islanded Distribution Network. IEEE International Conference on Power Engineering and Optimization, 40-45. 5. J. A. Laghari, H. M. (2013). A New Under-Frequency Load Shedding Scheme for Islanded Distribution Network. IEEE Innovative Smart Grid Technologies (ISGT), 1-6. 6. M. Karimi, H. M. (2012). Under-Frequency Load Shedding scheme for islanded distribution network connected with mini hydro. International Journal of Electrical Power & Energy Systems, 127-138. 7. Mihalic, U. R. (2011). Comparison of Adaptive UFLS Schemes in Modern Power Systems. IEEE Conference on Electrical Power and Energy (EPEC), (pp. 233-238). 8. Mirhydar, P. M. (1992). An adaptive method for setting under frequency load shedding relays. IEEE Transactions on Power System, 647-655. 9. P. Mahat, Z. C. (2010). Under frequency Load Shedding for an Islanded Distribution System With Distributed Generators. IEEE Transactions on Power Delivery, 911-918. 10. R. Nema and A. Trivedi. (2012). Load Frequency Control of a Small Isolated Power Station by Using Super capacitor Based Energy Storage System. International Journal of Advanced Research in Computer Engineering & Technology (IJARCET), 162-168. 11. Roman, P. a. (2012). Smart Load Shedding System. 3rd IEEE international symposium on Power Electronics for Distributed Generation Systems (PEDG) , (pp. 64-71). 12. Rudez, R. M. (2011). Analysis of Under frequency Load Shedding Using a Frequency Gradient. IEEE Transactions on Power Systems, 839-846. 13. S. Hirodontis, H. L. (2009). Load shedding in a distribution network. International Conference on Sustainable Power Generation and Supply, SUPERGEN '09, (pp. 1-6).

422

ISBN: 978-969-8680-32-9


Scrum Adaptation to Improve Software Quality in Software Engineering Semester Projects 1

2

2

1

Emmen Farooq , Adnan Iftekhar , Raheela Shahzadi , Syed Farooq Ali 1

Department of Computer Science, University of Management and Technology, Lahore, Pakistan {emmen.farooq,farooq.ali}@umt.edu.pk 2 Department of Computer Science, COMSATS Institute of Information Technology, Sahiwal , Pakistan {adnan,raheela}@ciitsahiwal.edu.pk

Abstract The motivation of this research is to make students comprehend the encouraging effects of Scrum software development model on software quality, at academic level. The first part of this research consist of half students of a Software Engineering class doing their semester projects via waterfall model and the remaining half of them doing their semester projects via Scrum. These projects were given as semester testing assignments, in the Software Quality and Testing course in the next semester. The testing students had to determine Test Case Pass to Fail Ratio, Average Percentage of High Priority, Critical De-fects, Average Percentage of High Priority, NonCritical Defects, Average Per-centage of Low Priority, Critical Defects, .and Average Percentage of Low Priority, Non-Critical Defects. It was discovered that , projects developed using Scrum were way ahead on quality scales , as compared to those that had been developed using Waterfall Keywords: Agile, Scrum, Software Quality, Software Testing, Academic, Teaching, Semester Project, Software Engineering

Introduction Software has established its importance in all domains of human life [1, 2, 3, 4, 5].With this established significance of software in human everyday life, improving and maintaining software quality becomes a vital avenue of research. To stay abreast with this principal requirement of software quality assurance, software industries and IT organizations are discovering and investing in novice techniques and methodologies [6]. The skill, commitment to efficient software engineering practices, technical expertise and commitment to software quality assurance of the industry personnel is an enormously imperative dimension of sustainability of software quality [7]. It has also been established that Scrum process model and software engineering practices, when applied together for software development, demonstrate a considerable enhancement in software quality [8]. Computer Science and Software Engineering graduates from Universities constitute a large segment of man power for the software industry. Hence, Software Engineering is a crucial course that makes its mark in nurturing future software engineers of the globe. The intent of teaching Software Engineering is to impart realworld software engineering practices that give the students a flavor of software development in the industry, and confidence to enter the software industrial arena [9]. It is imperative that students experience rigorous hands-onpractice in software engineering, so that they are not left behind in the feud in the industry [9]. To accomplish this aspect, semester projects related to software development can be a critical element of software engineering courses [10]. If software engineering practices are taught effectively and followed dedicatedly in the course project, this module can equip the students with ample knowledge to step into the challenging sphere of software development [11]. For successful grooming of students to meet the evolution of IT industry, it is necessary that software engineering teaching approaches are also upgraded and innovated in alliance with industry demands [16, 17]. This paper presents a joint research venture between Department of Computer Science, University of Management Technology, Lahore, Pakistan and Department of Computer Science, COMSATS Institute of Information Technology, Sahiwal, Pakis-tan. The synopsis of this endeavor is depicted in Fig. 1. The FALL 13, Software Engineering class at COMSATS Sahiwal was given five ideas for their semester projects. Half of the class was required to develop the as-signed projects following the Waterfall Model; the other half of the class was ad-vised to develop the projects using Scrum methodology.

423

ISBN: 978-969-8680-32-9


Fig. 1. Central Idea of the Research

The implemented projects of the above phase were given as testing projects to SPRING 14 Software Quality and Testing class at University of Management and Technology Lahore. The students of the testing class were required to test the projects on the parame-ters: Pass to Fail Ratio of Test Cases, Average Percentage of High Priority, Critical Defects, Average Percentage of High Priority, Non-Critical Defects, Average Per-centage of Low Priority, Critical Defects, .and Average Percentage of Low Priori-ty, NonCritical Defects. It was deduced that projects that had been developed under the umbrella of Scrum, made their mark when it came to competing for quality assurance. The section after introduction contains Literature Survey related to our work, after that come detail of method adopted to conduct this study. At the end of paper there are Results, Conclusions and Future Recommendations.

Related Work The vitality of improving software development procedures to incorporate maximum software quality remains clearly evident [14]. Moreover, the influence of Agile tech-niques in enhancing software product quality has also been recognized [8].Numerous studies, surveys and reviews have been conducted at industrial level, that judge the influence of Scrum on quality assurance of the resultant product [13][20, 21, 22] [24][26], while the arena of teaching software quality assurance by introducing Agile practices in university courses and semester projects of Computer Science and Soft-ware Engineering students, remains largely unexplored. A qualitative analysis exists on significance of introducing Agile techniques in Indian IT industry. This artifact remarks that, Agile adoption along with harmo-nious culture, management decisions, political influences, and economic changes can enhance Indian software potential [13]. The research on adopting Scrum in university projects to make students real-ize its impact on Software Quality is still in embryonic stage; however other dimen-sions of positive effects on adopting Scrum in undergraduate projects have been stu-died. A research study concludes that student collaboration was enhanced, and super-

424

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 visor-student communication was improved, when Scrum was adopted in Final Year Projects. It qualitatively indicates that quality was improved however there are no quantitative deductions [27]. Another research venture concludes that introducing Scrum in undergraduate software projects has encouraging consequences on improv-ing individual student performance [28]. Furthermore, several qualitative research studies, qualitatively analyze the positive impact of Scrum on improved working en-thusiasm of students [29, 30, 31]. It has been concluded by a research that when Scrum was used to teach Software Engineering, an improvement in students’ estimation and planning skills was observed [32]. Another study deduced that student moti-vation was enhanced when software engineering practices were thoroughly followed in academic undergraduate projects [33]. There is no significant research on making students realize the impact of Scrum on software quality of university level projects.

Empirical Setting This research endeavor consisted of two cycles. Development Cycle In FALL 2013, students of Software Engineering class at COMSATS Sahiwal were given five project ideas to choose for their semester project: Table 1. 1. 2. 3. 4. 5. 6. 7. 8. 9.

There were a total of 36 students in the class The students were required to work individually. They could implement the project in C++ or Java. They were required to develop detailed SRS documents and Design Documents along with implementation. 8 students opted for project A. 4 of them were asked to implement project A using Waterfall Model and the other 4 were asked to implement their projects via Scrum. 6 students chose project B as their semester project. 3 were asked to implement it through Waterfall Model and 4 were asked to implement it following the rules of Scrum model. 10 students chose project C. 5 were asked to follow Waterfall Method while the remaining 5 were asked to follow Scrum for project development. 5 students opted for project D. 2 were asked to use Waterfall Method for imple-mentation and 3 were asked to follow Scrum. 6 students chose project D. 3 were asked to implement it via Waterfall and the other 3 were asked to implement it following all the rules of Scrum.

Testing Cycle The SRS documents, Design documents and implemented solutions developed by the Software Engineering class, were given as testing assignments to FALL 2014, Soft-ware Quality and Testing class at University of Management and Technology, La-hore, Pakistan. 1. 2. 3. 4. 5. 6. 7. 8. 9.

There were 60 students in the class. Each student was given a project as an individual assignment. 6 were individually asked to test water fall versions of each project and 6 were assigned to individually test Scrum versions of each project. The testing results were determined on the average of all results. Due to semester limitations, the whole projects could not be tested, hence limited features were tested. The first assignment was to study the SRS and devise 500 test cases for each project to tests its functionality. Use Case Testing, Requirements Testing, Boundary Value Testing and Equiva-lence Partition Testing were also conducted over the course of the semester [34]. It was a compulsion to test seventy percent use cases. It was compulsory to test all critical requirements.

Test Case Pass to Fail Ratio P/F With reference to SRS documents of the projects, students were required to deter-mine Test Case Pass to Fail

425

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Ratio of each project. Defect Classification. Using the definition of defects in [36] and the classification scheme in [37], defects were classified on the basis of severity and priority as: Urgent Priority, Critical Urgent Priority, Non – Critical Low Priority, Critical Low Priority, Non – Critical

Results The testing results of the projects from Table 1 are shown in Table 2. Average Per-centage of each type of Defect was determined as given in [34]. Table 2. : Testing Results

Model

Project

Water Fall

A B C D E A B C D E

Scrum

P/F Ratio 1.24 1.79 1.45 2.78 1.19 8.26 6.16 5.94 7.68 8.87

Avg % D1 2.5 3.3 2.1 1.5 1.8 1.2 1.4 1.6 0.45 0.99

Avg % D2 20.1 28.5 37 22.9 21.3 23 20 18 20 20

Avg. % D3 40.8 23.3 21.9 35.6 28.56 33 40 30 32.55 29

Avg. % D4 36.6 45 39 40 48.34 42.8 38.8 50.4 47 50.1

First five rows, show testing statistics for Waterfall versions, while the last five rows show testing statistics for Scrum versions of the projects from Table 1. It is evident that test case pass to fail ratio increased a lot for all Scrum versions of the projects. The P/F ratio for Project A increased at least by 6 times , and for Project B the increase was more than 3 times , Project C showed approximately 4 times in-crease in P/F , while for D and E the increase in P/F ratio was nearly 2 times and more than 7 times respectively. Table 2 showed that high priority, critical defects decreased for Scrum ver-sions of Project A .Most errors were low, priority errors. For Scrum versions of Project B, there was also reduction in high priority, critical and high priority non-critical errors. Scrum versions of Project C also showed decreased high priority, criti-cal and high priority non-critical errors. A similar trend of reduced, high priority, critical and high priority non-critical errors was also evident in Scrum versions of Project D. Lastly, Scrum versions of Project E also showed decline in high priority, critical and high priority non-critical errors.

Conclusions In this research activity we have tried to establish the impact of adopting Scrum on software quality, in carrying out development projects in Software Engineering course. The quality indicators: test case P/F ratio and average percentage of defect were used to draw our results. The data for this endeavor was collected in two cycles: The development stage data was collected from COMSATS Sahiwal, Pakistan in the semester FALL 2013. The data for testing was collected from University of Management and Technology Lahore, Pakistan in the semester SPRING 2014. Testing proved that the projects, which had been developed using Scrum, had an in-creased PASS/FAIL ratio of test cases and reduction in average defect occurrence.

Future Recommendations

426

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Due to limitations of an academic semester, there were limitations to the amount of time spent on testing. Students could only devise 500 test cases due to time limitations, this scope can be increased to completely testing the projects. White Box Testing was an important asset of this activity. Due to time limitations, students were asked to test parts of projects’ code for Data Flow and Control Flow techniques. White Box Testing can be expanded to whole code and the results can be analyzed in terms of project quality. This study was carried out at two universities, development was done in one uni-versity and testing was done at the other. Both universities can separately introduce this methodology, in their curriculums, to demonstrate the effect of Scrum tech-nique on software quality.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

Northrop, L.: Let's Teach Architecting High Quality Software. In: 19th Conference on Software Engineering Education and Training, CSEET, Oahu, HI, USA (2006). Wiiliamson, K., Healy, M., Baker, R.: Industrial Applications of Software Synthesis via Category Theory - Case Studies Using Specware. In: Automated Software Engineering, Vol. 8, Issue 1, pp. 7-30. (2001) Chavez – Santiago, R., Mateska, A.,. Chomu, K., Gavrilovska, L., Balasingham, I.: Appli-cation of software- defined radio (SDR) technology in Hospital Environments. In: 35th Annual International Conference of the IEEE, pp. 1266 – 1269. Osaka , Japan (2013) Song, Y., Peng, X., Xing, Z., Zhao, W.: Automatic Adaptation of Software Applications to Database Evolution by Graph Differencing and AOP-Based Dynamic Patching, COMPSAC ’12. In: Proceedings of the 2012 IEEE 36th Annual Computer Software and Applications Conference, pp. 111-118. Washington, DC, USA (2012). Hameed, S.A.Towards Software Engineering Based on Islamic Ethical Values. In: Inter-national Conference on Computer and Communication Engineering, ICCCE 2008, pp. 379 -385. Kuala Lampur , Malaysia (2008) Agarwal, A., Garg , N.K., Jain , A. : Quality Assurance for Product Development Using Agile. In: International Conference on Optimization, Reliability and Information Tech-nology, ICROIT 2008, pp. 44 – 47. Faridabad, India (2008) Paulk, M.C.: Factors Affecting Personal Software Quality. In: Crosstalk: The Journal of Defense Software Engineering, Vol. 19, No. 3, pp. 9-13. (2006) Williams, L., Brown, G., Meltzer, A., Nagappan, N.: SCRUM + Engineering Practices: Experiences of Three Microsoft Teams. In: International Symposium on Empirical Soft-ware Engineering and Measurement, pp. 463 – 471. Banff, AB (2011) Sun, Y. The Challenge and Practice of Creating Software Engineering Curriculum. In: 24th IEEE – CS Conference on Software Engineering Education and Training, CSEE & T, 497 – 501. Hawaii, USA (2011) Ding, J.: A Framework for Global Collaboration in Teaching Software Engineering. In: International Workshop on Collaborative Teaching of Globally Distributed Software De-velopment, pp. 30 – 34. San Francisco , CA , USA, (2013) Jarzabek, S.: Teaching Advanced Software Design in Team Based Project Course. In: 26th IEEE Conference on Software Engineering Education and Training, CSSE & T, pp. 31 – 40. San Francisco , CA , USA, (2013) Shahane, D., Jamsandaker, P., Shahane, D.: Factors Influencing the Agile Methods in Practise – Literature Survey and Review. In: International Conference on Computing for Sustainable Global Development, pp. 556 – 560. New Delhi , India , (2014) Mordinyi, R., Kuhn, E., Schatten, A.: Towards an Architectural Framework for Agile Software Development. In: 17th IEEE International Conference and Workshops on Com-puter Based Systems, pp. 276 – 280. Oxford , UK, (2010) K. Schwaber and M. Beedle, “Agile software development with Scrum”,1st ed., Prentice Hall, 2001. Yelmo, J.C., Fernandez – Corugedo , J.: An Experience of Educational Innovation for the Collaborative Learning in Software Engineering .In: IEEE Global Engineering Education Conference , pp. 567 – 574. Amman, (2011) Yu, G., Xiangzhong, F.: Incorporating Innovation Oriented Education into the Software Engineering Course. In: IEEE International Conference on Systems, Man and Cybernetics, 5108 -5110. San Antonio, TX, (2009) Ping, H., Mingrui, C., Shaofan, C.: The Practice Training in the Software Engineering Education. In: The 9th International Conference for Young Computer Scientists, pp. 2636 - 2640. Hunan , (2008) Rosca , D., Tepfenhart , W., McDonald, J. : Software Engineering Education : Following a Moving Target. In: 16th Conference on Software Engineering Education and Training, pp. 129 – 139. Madrid , Spain (2003) Hassan, S.Z.: Software Industry Evolution in a Developing Country: An In depth Study. In: 33rd Hawaii Annual Conference on System Sciences , Hawaii USA(2000) Korhonen, K.: Evaluating the Effect of Agile Methods on Software Defect Data and De-fect Reporting Practices - A Case Study .In: Seventh International Conference on the Qual-ity of Information and Communications Technology, pp. 35 – 43. Porto, (2010) Woodward, E.V., Bowers, R., Thio , V.S., Johnson, K., Srihari ,M.,. Bracht, C.J.: Agile Methods for Software Practice Transformation. In: IBM Journal of Research and Devel-opment , Volume 54 ,Issue 2 , pp. 3:1 – 3:12.(2010) Sison , R., Yang , T. :Use of Agile Methods and Practices in the Philippines. In: 14th Asia – Pacific Software Engineering Conference, pp. 462 – 469. Aichi,(2007) Schindler, C.: Agile Software Development Methods and Practices in Austrian IT-Industry: Results of an Empirical Study. In: International Conference on Computational Intelligence for Modeling Control & Automation, pp. 321 – 326. Vienna (2008) Asnawi, A.L. , Gravell. ; A.M., Wills, G.B.: Emergence of Agile Methods: Perceptions from Software Practitioners in Malaysia. In: AGILE India , pp. 30 – 39 . Bengaluru, (2012) Rodriguez, P., Markkula, J. ; Oivo, M., Turula, K.: Survey on Agile and Lean Usage in Finnish Software Industry. In: ACM-IEEE International Symposium on Empirical Soft-ware Engineering and Measurement, pp. 139 -148. Lund(2012) Scharff, C.: Guiding Global Software Development Projects Using Scrum and Agile with Quality Assurance. In: 24th

427

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 IEEE-CS Conference on Software Engineering Education and Training, pp. 274 – 283. Honolulu, HI(2011) 27. Ashraf, M.A., Shamail, S., Rana, Z.A. :Agile Model Adaptation for E-Learning Students’ Final-Year Project. In: IEEE International Conference on Teaching, Assessment and Learning for Engineering, T1C 18 – T1C 21. Hong Kong (2012) 28. Gamble, R.F., Hale, M. L.: Assessing Individual Performance in Agile Undergraduate Software Engineering Teams. In: Frontiers in Education Conference, .pp. 1678 – 1684. Oklahoma City , OK (2013) 29. Donizetti, Z., de Ponte, L. , Lucredio, D. :Using Scrum to Teach Software Engineering: A Case Study. In: IEEE Frontiers in Education Conference, pp. 455 – 461.Oklahoma City , OK ( 2013) 30. Damian, D., Lassenius, C., Paasivaara, M., Borici,A. :Teaching a Globally Distributed Project Course using Scrum Practices. In: Collaborative Teaching of Globally Distributed Software Development Workshop, pp. 30 – 34. Zurich (2012) 31. Paasivaara, M., Lassenius, C., Damian, D., Raty, P. :Teaching Students Global Software Engineering Skills using Distributed Scrum .In: Software Engineering (ICSE), 35th Inter- national Conference on Software Engineering (ICSE),pp. 1128 – 1137. San Francisco, CA(2013) 32. Mahnic, V.: A Capstone Course on Agile Software Development Using Scrum. In: IEEE Transactions on Education, Vol.55 , Issue .1 ,pp. 99 – 106.(2012) 33. Kropp, M., Meier, A.: Teaching Agile Software Development at University Level: Val-ues, Management, and Craftsmanship. In: IEEE 26th Conference on Software Engineering Education and Training (CSEE&T), pp. 179 – 188. San Francisco, CA,(2013) 34. P.C. Jorgensen: Software Testing: A Craftsman’s Approach, 4th Ed., Auerbach Publica-tions,2013. 35. I. Sommerville: Software Engineering, 9th ed., Addison-Wesley, 2010. 36. R.S. Pressman: Software Engineering: A practitioner’s approach, 7th ed., McGraw-Hill, 2009. 37. Soft ware Testing and Quality Assurance : Priority and Severity in Software Testing , URL : http://softwaretestingguideline.blogspot.com/2012/09/priority-and-severity-in- software.html, Sep. 14 , 2012 [ July 10 , 2014 ].

428

ISBN: 978-969-8680-32-9


Software integration practices for GSD vendors - A systematic literature review protocol Muhammad Ilyas and Muhammad Younas Software Engineering Research Group (SERG_UOM), Deptt: of Computer Science &IT, University of Malakand, KPK, Pakistan. [email protected], [email protected]

Abstract. The software development trend has changed from local to global software development (GSD) with the advances in information and communication technologies. This trend has brought some challenges along with the benefits gained from it. One of the technical challenge that GSD vendors faces is the integration of components developed by GSD teams. In our previous two studies we have identified the challenges, which GSD vendors needs to address and success factors that play positive role at any stage of integration. The objective of the current paper is to design systematic literature review (SLR) protocol for the identification of the practices and solutions needed to address the integration challenges and to implement the integration success factors. The identification of these practices will complete the second phase of our proposed software integration model (SIM). The SIM model will assist GSD vendors in effective integration of their software products.

Keywords: Systematic Literature Review, Global software Development, Software Integration Practices and Solutions Introduction From the last two decades the software development trend has been changed from local to global software development. Associated with the benefits (like reduce development cost, improved product quality, large labor pool, availing latest technology and infrastructure etc) gained from GSD are certain challenges (like communication and coordination problem, cultural, language and time zone differences, poor knowledge and contract management etc) [1-3]. Among these challenges one of the technical challenge that GSD vendors face in the later stage is the integration of software components developed by GSD teams or purchased from the market i.e. commercial of the shelf or off the shelf (COTS/OTS) components [4, 5]. Many of the hidden problems of the early stages start appearing during the integration stage [6]. Due to these hidden problems not only the workload is increased but the quality of the final product is also decreased. In most than 50% of the project, the integration phase is one of the problematic and complicated phase [7]. Larsson [6] identified three main problems in the integration of software components:  Components needed for integration may be not ready  There may be no well defined interfaces between components  There may be no well prepared environment for integration Due to little communication and coordination of the different development teams working at different geographical locations, the above mentioned problems become more critical in GSD environment d. Each team in GSD will be responsible to work on its own component without having a complete picture of the overall system[8]. Similarly each team may be using a different process model and standards which will further complicate the integration process at the end [7]. In our previous two studies we have identified a list of integration challenges and success factors for GSD vendors as shown in Error! Reference source not found.. The objective of this study is to identify the practices needed to address the above mentioned challenges and the practices required for implementation of the success factors. This will complete the second phase of our proposed model SIM[9]. Thus we have designed the following research questions: RQ1. What are the practices, as identified in the literature that can be used to address the integration challenges in GSD environment?

429

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 RQ2. What are the practices, as identified in the literature that can be used to implement the success factors for integration in GSD environment? The rest of the paper is organized as follows. In Section 2 we present background and motivation while in the remaining sections we present the steps of the SLR protocol.

Background As compared to the importance of software integration, little research work has been done on integration, particularly on the identification of integration practices in GSD environment. Majority of the software projects are delayed due to the complexities and incompatibilities found between components developed by GSD teams [4, 10]. Integration means to assemble or combine or merge two or more components or products into one system or subsystem. In the glossary of EIA (Electronic Industries Alliance) 731.1 Standard (SYSTEMS ENGINEERING CAPABILITY MODEL) integration has been defined as: “Integration: The merger or combining two or more elements (e.g., components, parts, or configuration items) into a functioning and higher level element with the functional and physical interfaces satisfied”. Herbsleb and Grinter [11] have defined integration as “all the work necessary to assemble the product from its components”. Table 1. Software integration challenges and success factors for GSD vendors S.N

Challenges

Success factors

1

Lack of Communication

Consistency in Requirements and Architecture Design

2

Lack of Proper Documentation

Intra and inter team Communication and Coordination

3

Lack of Compatibility

Component/Unit Testing prior to integration

4

Architecture Mismatch

Advance & Uniform Development Environment and Training

5

Lack of Integration Planning and lack of Management

Efficient Incremental/Continuous integration

6

Heterogeneous Devlopment Environment and Platforms

Efficient specification for Interface Compatibility

7

Improper/No unit testing

Proper Documentation & Configuration Management

8

Wrong COTS/OTS Product selection and customization

Early Integration Planning and Centralized P3 management

9

Lack of Resources, Knowledge and Skills

Careful evaluation of the COTS/OTS Components

10

Lack of proper Component Interfaces

Use of Standard Model for Process, Data and Product’s Components

11

Unclear responsibilities

Use of modular approach

12

Configuration and Versioning complexity

Use of Efficient Metrics

13

Lack of common understanding of requirements

Use of Quality assurance

14

Timely availability of Components

Specific Integration Timing

15

Big bang integration

16

Lack of common development process

McConnel [12] describes integration as “the software development activity in which you combine separate software components into a single system”. While Larsson [6] have defined Product Integration as a process as follow: “The product integration process is a set of procedures used to combine components into larger components, subsystems or final products and systems. Product integration enables the organization to observe all important

430

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 attributes that a product will have; functionality, quality and performance. This is especially true for software systems as the integration is the first occurrence where the full result of the product development effort can be observed. Consequently, the integration activities represent a highly critical part of the product development process”. Many researchers worked on the software integration, some of the important studies are as follow. Herbsleb et al [11], in a case study on “Splitting the organization and integrating the Code”, have considered the integration phase as one of the most difficult phase of the GSD project. They have pointed out that in multisite development, due to the loss of communication and coordination, integration process is badly suffered. The components needed for integration may not be available according to the planed schedule or may have not well defined interfaces. Similarly Van Moll et al [7] have conducted an exploratory study on “Identifying Pitfalls of System Integration”. They identified that in more than 50% of the GSD projects integration phase is one of the difficult and problematic phase. They suggested that better planning, monitoring and control and explicitly assigning the responsibilities to all members may help to minimize the integration problems. They have further suggested considering Integration and Integration Testing as a separate process during software development because it is often considered as a formality only. Bosch et al [10] in a case study “From Integration to Composition: On the impact of Software Product Lines, Global Development and Ecosystem” have concluded that many of the challenges in large-scale software development exists due to the integration-centric approach. These challenges and complication can be simplified through a transition from integration-centric approach to composition-oriented approach to the development of software. But they also mentioned that composition-oriented approach is not always preferable and there are many cases where integration-centric approach is preferable. Cataldo and Herbsleb [13] have performed an empirical analysis on a development organization (responsible for producing navigation system for automobiles) to identify “Factors Leading to Integration Failures in Global Features Oriented Development”. They have analyzed that cross-feature interaction, which is a measure of architecture dependencies between two product features, is one of the big failure of product integration. The big limitation of this study is that they have examined only a single system developed by a single organization. Most of the studies have presented a problem oriented report rather than suggesting solutions and practices. None of the studies has considered integration problems in depth based on the project size and product type. To bridge the gap and to ease the integration process we have planned to conduct a systematic literature review for the identification of integration practices and solutions in GSD environment. These practices will be helpful for GSD vendors to address the integration challenges and to implement the positive factors for integration that we have already identified.

Systematic Literature Review Protocol The purpose of this paper is to design the systematic literature review (SLR) protocol for software integration practices for vendors working in GSD environment following the Kitchenham and Charters guidelines [14]. We have also studied two other protocols [15, 16] for guidelines, which are also based on SLR. There are three main phases of SLR as defined by Kitchenham: Planning the review, Conducting the review and Reporting the review. This paper is only a plan for the review i.e. here we will follow the first phase of SLR. The output of this phase will be an SLR protocol which will set all the steps and procedure in advance for conducting the review. Thus SLR is protected from researcher bias through a predefined protocol although it cannot control the publication bias [14]. The SLR process is explained in Figure 1. In the remaining section we will design the protocol for the review.

Research Questions RQ1. What are the practices, as identified in the literature that can be used to address the integration challenges in GSD environment? RQ2. What are the practices, as identified in the literature that can be used to implement the success factors for integration in GSD environment?

431

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Construction of search terms Here we define some terms which will help us in designing the search string for our research questions. It should be noted that we are designing one search string for both research questions. Population: Global/Distributed software development environment Intervention: Practices/solutions, success factors, challenges Outcome of relevance: Successful and effective software product integration Experimental design: SLRs, empirical studies, case studies, theoretical studies and expert opinions

Thus using the above terms we can write our research questions as follow: RQ1&2: [What are the practices,] INTERVENTION That can be used to address integration challenges OR that can be used to implement the success factors [GSD Environment]

POPULATION

[effective and successful software product integration]

OUTCOMES OF RELEVANCE

Search Strategy Trail Search We have conducted a trial search using the following search string in Google scholar and IEEE Xplore digital library: ((“Software Integration” OR “Product Integration”) AND ("Global software development" OR “Global software engineering” OR “Distributed software development” OR GSD) AND (Practice OR solution)) The papers retrieved through this search string will be used as guidelines for the development and validation of the major search terms. Identification of search terms We will use the following plan for the identification of search terms: First we will derive the major search terms from the research questions by identifying population, intervention and outcome. Then we will find the alternative spellings and synonyms for the major terms derived in the above step Next we will verify the key words in some relevant papers. Finally we will use the “OR” operator for concatenation of synonyms and alternative spellings and the “AND” operator for concatenation of major terms. It should be noted that some database (e.g. Google scholar) does not allow lengthy search strings; therefore we will break the string formed in the above steps into two or more small strings if required. We got the following results after execution of the steps in the above plan. Result for (a) RQ1&2: Global Software Development, practices/solutions, software product integration. Result for (b) Product Integration: (“Product integration” OR “Software integration” OR "Component integration" OR "Software system integration") Global Software Development: (“Global software development” OR GSD OR “Global software engineering” OR GSE OR “Distributed software development” OR DSD OR “Distributed software engineering” OR DSE) Practice: Practice OR solution OR exercise OR advice OR “implementation initiative” OR “lessons learned” OR “process improvement” OR “process enhancement” OR “process innovation” Result for (c)

432

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 Softwaree Integration, product p Integrration, global software deveelopment, globbal software engineering, e d distributed software developmentt, practice, sollution, exercisse, advice, im mplementation n initiative, lesssons learned,, software mprovement process im

Ideentification n of a need for th he review

Ideentification n of Research questions

Seearch strateg gy

Perform P triall search

((Strings) and sources

Deefine inclusiion/ excclusion criteeria

D Design Dataa ex xtraction forrm

Syn nthesizing

Peerform pilott Daata extractio on

datta from papeers

Figuree.1 Systema atic Literatture Review w process Result foor (d) String (ffor identificatiion of practicees) (("Softwaare integratio on" OR "Prooduct integraation" OR "C Component integration" i O "softwarre system OR integratioon") AND (""Global softw ware development" OR "Global softw ware engineerring" OR "D Distributed software developmentt" OR "Distribbuted softwarre engineeringg") AND (Praactice OR sollution OR exeercise OR advice O OR "implemeentation initiaative" OR "lessons learnned" OR "prrocess improvvement" OR "process enhancem ment" OR "proocess innovatiion"))

433

ISBN: 978-969-8680-32-9

Short P Papers-Procceedings of IMTIC’15 The abovve search string will be addjusted, formaatted and breaak according to the format requirementts of each library thhat we will seaarch. Resourcees to be searcched We will ttry to search thhe following ddigital librariees/databases. IEEE Xpplore

(http://ieeeexplore.ieee..org)

ACM Poortal

(http://dl.acm.org)

Science D Direct

(ww ww.sciencedireect.com)

Springer Link

(http://ww ww.springerlink.com/)

Google Scholar S (http p://scholar.gooogle.com) Search constraints c an nd validation As mentiioned in the baackground secction, the amoount of work done d is very liittle for the identification off software integratioon challengess and successs factors in general and for the idenntification off practices/sollutions in particularr. We are thereefore intendedd to put no datte boundaries on the search h that we will made m in these libraries.

Searrch papers from m

Primarry Selection doone

releva ant journals an nd

on n the basis of

conference

reviewing the title,

prroceedings

keywo ord and abstraact

Fina al Selection bassed

Check k quality criterria

Ch hecked againstt

on reading full teext

beforee including pap per

Incllusion/Exclusioon

for final f selection..

criteeria, reading fu ull text

F Figure.2 Pub blication seelection pro ocess

Publicattion selection criteria r to thee research The publlication selecttion criterion is used to sellect only thosse publications which are relevant question.. We will conssider only thoose papers whiich are relevaant to softwaree integration in GSD. Paperrs relating to hardware integrationn will be ignoored. The publlication selectiion criterion is explained in n Figure 2. Primaryy selection Initial seelection of thee primary souurces will be performed byy reviewing th he title, keyw words and abstract. The purpose is to ignore only o those reesults which have h no relevvance to the problem p or reesearch questtions. The primary sources chossen in the initial i selectioon process will w be checked against the below mentioned m inclusionn/exclusion crriteria by reviiewing thorouughly full text of the studiies. The sourcces will be seent to the secondarry reviewer, fo or review in caase of any unccertainty regarrding the incluusion or excluusion decision.

434

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Inclusion criteria The inclusion criteria is used to determine which piece of literature (papers, technical reports, or grey literature etc.) found by the search term will be used for the data extraction. We will consider only those papers which are related to global/distributed software development/engineering written in English language. The criteria are listed below:  IC1. Studies that describe practices that have a positive impact on any stage of the integration process.  IC2. Studies that describe software/ product integration practices in GSD environment based on any project size or product type. Exclusion criteria This section describes the exclusion criteria in order to decide which piece of literature found by the search term will be excluded / ignored. The criteria are listed below:  EC1. Studies that is not relevant to the research question.  EC2. Studies other than software integration i.e. hardware integration like telecommunication or electronic component integration will be not considered.  EC3. Studies that don’t describe integration practices in global software development.  EC4. Studies that describe the practices that have no role in any stage of the integration process. Publication quality assessment The quality assessment will be performed after final selection of publications. The publications quality will be evaluated in parallel with data extraction. The quality will be gauged on the basis of the following questions: Is it clear how the integration practices for critical success factors (CSFs) were identified in GSD environment? Is it clear how the integration practices for critical barriers (CBs) were identified in GSD environment? Each of the above factors will be marked as “YES” or “NO” or “Partial” or “N.A”. My supervisor, as a secondary reviewer, will randomly score a small subset for validation.

Data Extraction Strategy Primary Study Data The focus of the study is to collect that data, from publications, which satisfy the research question for the review. We will extract the following data from each publication: Data about the publication i.e. Title, Authors, Journal/Conference title, etc Data that will address the research questions The following data will be extracted to address the research question: RQ1&2: Background information, practices which will have a positive impact on any stage of the integration process in GSD environment and practices which will address the integration challenges in GSD environment. The following data will be captured during data extraction: Data Extraction Process The review will be carried out by a single researcher, who will be responsible for the data extraction. In case of an issue the secondary reviewer will be approached for guidance concerning the data extraction. After completing data extraction process by primary reviewer, the inter-rater reliability test will be performed by secondary reviewer. The secondary reviewer will select few publications randomly from the list of publications already chosen by the primary reviewer. The secondary reviewer will independently extract data from the randomly selected publications. The results will then be compared with the results produced by the primary reviewer.

435

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 Primary Reviewer:

Muhammad Ilyas

Secondary Reviewer:

Dr Siffat Ullah Khan

Data to be extracted               

Date of review Title Authors Reference Database Methodology (interview, case study, report, survey, etc) Sample Population Publication Quality Description Organization Type (Software house, University, Research institute etc) Company size (small, medium, large) Project size (i.e. small, medium, large) Product type (i.e. bespoke, off the shelf, customize of the shelf) Integration stage (i.e. before, during or after the integration process) CSF or CB that will be addressed by the corresponding practice Practices identified in literature

Data Storage The summarize data for each publication will be kept as SPSS/Microsoft Word document and will be stored on local drive at University of Malakand. Data Synthesis

The data will be synthesized by creating one summary table for each research question having the columns S.No, success factors/challenge, practices, frequency/percentages and paper-id/title etc. Validation of the Review Protocol The protocol was reviewed by two members of software engineering research group at university of Malakand (SERG_UOM) for suggestions. The protocol after updating will be further review by the secondary reviewer and will be presented to SERG_UOM for their comments. Finally it will be published as a technical report.

436

ISBN: 978-969-8680-32-9


Review Timetable Task

Date

Start of Protocol

04-Nov-2014

Submission of the protocol for review

30-Nov-2014

Completion of the protocol

10-Dec-2014

Completion of Search

10-Feb-2015

Completion of Primary study selection

28-Feb-2015

Completion of data extraction

31-Mar-2015

Completion of data synthesis

15-Apr-2015

Completion of review reports

30-May-2015

Divergences In case of any divergence from the protocol, which may occur during the study, we will record any change in a new Appendix to this document. References 1. M. Romero, A. Vizcaíno, and M. Piattini, "Toward a Definition of the Competences for Global Requirements Elicitation," presented at 13th annual conference on Innovation and technology in computer science education, 2008. 2. B. H. C. Cheng and J. M. Atlee, "Research Directions in Requirements Engineering.," in FOSE, 2007. 3. S. U. Khan, M. Niazi, and R. Ahmad, "Barriers in the selection of offshore software development outsourcing vendors: An exploratory study using a systematic literature review," Information and Software Technology 53, pp. 693–706, 2011. 4. S. Schneider, R. Torkar, and T. Gorschek, "Solutions in global software engineering: A systematic literature review," International Journal of Information Management, vol. 33, pp. 119-132, 2013. 5. B. Tekumalla, "Status of Empirical Research in Component Based Software Engineering-A Systematic Literature Review of empirical studies." Sweden: University of Gothenburg, 2012, pp. 52. 6. S. Larsson, "Key Elements of the Product Integration Process," Malardalen University Sweden, 2007, pp. 78. 7. J. H. Van Moll and R. W. M. Ammerlaan, ""Identifying Pitfalls of System Integration -- An Exploratory Study"," presented at IEEE International Conference on Software Testing Verification and Validation Workshop, 2008. 8. A. Zafar, S. Ali, and R. K. Shahzad, "Investigating integration challenges and solutions in global software development," presented at Frontiers of Information Technology (FIT), Islamabad, 2011. 9. M. Ilyas and S. U. Khan, "Software Integration Model for Global Software Development," presented at 15th International Multitopic Conference (INMIC), Islamabad, 2012. 10. J. Bosch and Petra Bosch-Sijtsema, "From integration to composition: On the impact of software product lines, global development and ecosystems," Journal of Systems and Software, vol. 83, pp. 67-76, 2010. 11. J. D. Herbsleb and R. E. Grinter, "Splitting the Organization and Integrating the Code: Conway’s Law Revisited," presented at Proceedings of the 21st International Conference on Software Engineering (ICSE), NY USA, 1999. 12. S. McConnel, Code Complete 2, 2nd ed: USA: Microsoft Press Redmond, Wa, 2004. 13. M. Cataldo and J. D. Herbsleb, "Factors leading to integration failures in global feature-oriented development: An empirical analysis," presented at 33rd International Conference on Software Engineering (ICSE), 2011.

437

ISBN: 978-969-8680-32-9

Short Papers-Proceedings of IMTIC’15 14. B.Kitchenham and S.Charters, "Guidelines for performing Systematic Literature Reviews in Software Engineering," Keele University, UK 2007. 15. M. Ilyas and S. U. Khan, "Software Integration Challenges in Global Software Development Environment: A Systematic Literature Review Protocol," IOSR Journal of Computer Engineering (IOSRJCE), vol. 1, pp. 29-38, 2012. 16. J. M. Verner, O. P. Brereton, B. A. Kitchenham, M. Turner, and M. Niazi, "Risks and risk mitigation in global software development: A tertiary study," Information and Software Technology, vol. 56, pp. 54-78, 2014.

438

ISBN: 978-969-8680-32-9

Implementation of PLC Based VFD Fed Three Phase ...

Implementation of PLC Based VFD Fed Three Phase ...

Suggest Documents

RPWM Based Four Switch Three Phase Inverter Fed ...

The Asymmetrical Three-Phase Induction Motor Fed by Single Phase ...

Implementation Of Three-Phase Power Factor ...

Design and Implementation of Single-Phase, Three

Fuzzy Predictive Control of Three-Phase Four Switch Inverter-Fed ...

Control of Four Switch Three Phase Inverter Fed Induction Motor ...

Vector Control of Four Switch Three-Phase Inverter Fed Synchronous ...

performance evaluation of three- phase induction motor drive fed ...

performance evaluation of three- phase induction motor drive fed from ...

Three-phase Steady-State Model of Doubly Fed Induction Generator ...

EMI Study of Three-Phase Inverter-Fed Motor Drives

Control of Four Switch Three Phase Inverter Fed Induction ... - CiteSeerX

Transient Analysis of Z-Source Inverter Fed Three-Phase Induction ...

IRJET- Control of Induction Motor using PLC and VFD

VFD

A ZVS-PWM THREE-PHASE CURRENT-FED PUSH

direct torque control for matrix converter-fed three phase induction ...

Implementation of a Speed Field Orientated Control of Three Phase

Design and Implementation of PLC-Based Monitoring Control ...

PLC-based Implementation of Supervisory Control for ... - CiteSeerX

Design and Implementation of PLC based Automatic ...

Design and Implementation of a PLC based Electricity ... - IOSR Journals

Design and Implementation of a PLC based Electricity ... - IOSR journals

Model of Three-Limb Three-Phase Transformer Based ... - Telfor Journal