Deblurring Images Using the Wiener Filter

CONTENTS Page No. Fingerprint Recognition system for Blind People A. Vanilavanya, Shilpeechattopadhyay, I. Ramalakshmi.........................................................1 Morphological Parsing of Tamil Verb Using Finite State Automata R.Padmamala .........................................................................................................................7 Brain controlled artificial legs P.M. Uma, G. Indumathi ...................................................................................................... 10 A Comparative Analysis on Audio Visual based e- learning N. Muthu Meenakshi, N. Mahadevan, R.Karthiya Banu, Saravanan ...................................... 17 Study and Analysis of Visual e-learning – Empowering Visually Challenged students R.Sangeetha, S. Praveena, R.Karthiya Banu Saravanan ........................................................ 23 A Comparative Analysis of Predicting the Net Asset Values of Indian Mutual Funds using Artificial Neural Network (ANN) Models and Auto Regressive Integrated Moving Average (ARIMA) E. Priyadarshini, A. Chandra Babu ....................................................................................... 27 A Common Data mining Framework for Different ERP Models Prof. M.K. Gandhi, Dr. K. Sarukesi ...................................................................................... 32 A Framework for event matching in temporal database N. Duraimutharasan, Dr. K. Sarukesi.................................................................................... 37 Comparative Study of Various Classification Algorithms for Heart Data A.Anushya, A. Pethalakshmi................................................................................................ 42 Role of Data Mining in Network Intrusion Detection K.Chokkanathan, D.Saravanan ............................................................................................. 46 A Lossy Medical Image Compression Scheme Using Inner Elliptic ROI and Biorthogonal Spline Wavelet R Loganathan, Dr. Y S Kumaraswamy ................................................................................ 51 Medical Image Retrieval System Using Gain based Feature Extraction Sasi Kumar M, Dr.Y.S.Kumarswamy .................................................................................. 55 Security in Medical Image Transmission A.Umamageswari, M.Ferni Ukrit, Dr.G.R.Suresh................................................................. 58 Symmetry Based Cluster Approach for Automatic Recognition of the Epileptic Focus in Brain Using PET Scan Image – An Analysis A. Meena1, K. Raja.............................................................................................................. 61

Page No. Distortion Correction of an Image Intensifier S. Suganya, M.Pappa, S.Murugaanand........................................................................64 Image Reconstruction Based on Geometry Driven Approach Using Particle Filtering Technique M.Rohit Reddy, B.Narsimha Chary...................................................................................... 69 A Novel Approach for Image Compression Using 2D Dual Tree Complex Wavelet Transform (2D DT-CWT) Avvari Devendra kumar, S.Varadarajan................................................................................ 74 Deblurring Images Using the Wiener Filter Mrs. Sharada P.N., Mr. Sree Harsha H.N,............................................................................. 79 High Dynamic Images using Differently Exposed Images D. Beulah David, R.K.Saranya M. A. DoraiRangaswamy ..................................................... 83 ROI Segmentation and Specular reflection removal for Colposcopy Cervical Images P.S. RamaPraba, Dr. H. Ranganathan ................................................................................... 88 Fused MRI,CT Image Compression by Choosing ROI while Preserving the Uniformity of Image Ravi Chandra G.,Prasad A.M ............................................................................................... 91 Study of Image Segmentation Techniques- Issues and Trends Prof. Deepika Shukla............................................................................................................ 95 Traditional Color Video Enhancement Based on Adaptive Filter B.Kalyan kumar, I.santiprabha ....................................................................................... 100 Fisherface Method for Face Recognition Rohini Suhag, Pradeep Mishra ........................................................................................... 103 Texture Based Analysis and Classification of Mammogram Images S.Deepa, Dr.V.Subbiah Bharathi ........................................................................................ 106 Image Information Processing K.Vikram, Dr. Neeraj Upadyaya, Dr.A.Govardhan, Mrs.Rafath Samrin, S.IshrathAhamad................................................................................................................ 110 Image Edge Detection Using Fuzzy Logic Fazlulla Khan, Divyaprabha, Dr. M Z Kurian .................................................................... 116 Sharon’s - 86th Bio Sensor Module Incorporated with Mobile Application, Embed with Electronic Stethoscope and Cardiogram meter using Christina theory T. Shantha Kumar ....................................................................................................120 MAODV Routing Protocol Implementation in NS2 for Ad hoc Networks R. Pandi Selvam and V.Palanisamy.................................................................................... 125

Page No.

QDCCM: A Congestion Detection and Control Mechanism for Wireless Sensor Networks Mr.G.Srinivasan, Dr.S.Murugappan ................................................................................... 129 Improving Intrusion Detection System (IDS) for Mobile Adhoc Network Using Fuzzy Approach M.B. Mukesh Krishnan, Prof. Dr. P. Sheik Abdul Khader .................................................. 134 Bandwidth Reservation Technique Based on Traffic Priority for Wireless Mesh Networks K. Valarmathi and Dr.N. Malmurugan................................................................................ 137 Advanced Search Algorithm In Unstructured Peer-To-Peer Networks using Search on Odd and Even Depths technique S. Anbu, C. Jayakumar ...................................................................................................... 142 Deployment of Honeypots in the Next Generation Peer-to-peer Botnet V.Adithya Pothan Raj, M.P.Srinivasan, J.Sathiyamoorthy. ................................................. 147 Effective Intrusion Detection Mechanism in the Context of Corporate Business Model Kalyani Kundeti, Dr. M.V. Vijaya Saradhi, G. Bala Krishna............................................... 152 Epigrammatic Study of IaaS: Cloud Computing Muthuvel L, Thamarai Chelvi............................................................................................ 157 Performance Evaluation for Virtual Server Management in Cloud Computing J.Kalidas, K.Suresh ............................................................................................................ 161 Trends in Data Caching Techniques for Mobile Ad hoc Networks Preetha Theresa Joy, K. Poulose Jacob ............................................................................... 167 Performance Modeling of Efficient Monitoring for Intrusion Detection in Mobile Ad Hoc Networks Pandiyan Durairajan, Dr.T. Sasikala, .................................................................................. 172 Dynamic Decision Neural Network in War Field (Neurocop) V. Kishore Kumar, M. Shivaram ........................................................................................ 178 A Novel Approach to Prevent and Detect SQLIA N. Veeranjaneyulu, M. Ramakrishna, B. Jyostna Devi........................................................ 181 Word Sense Disambiguation using Cascade Feed-forward Back propagation Network Tamilselvi P, S.K.Srivatsa................................................................................................. 186

Page No. Application of 3G Network for Motor Vehicle Security Dhiraj Kumar, Bimlendu kumar Jha, D. Sarvanan,.............................................................. 191 Differentiated services network with Dynamic admission control algorithm Hemlata Pal (PG student) ................................................................................................... 194 Data Security in Wireless Sensor Networks Using A star topology Approach Sajana .............................................................................................................................. 198 Consistency Management Strategies for Data Replication in MANET Mr.Y.Chitti Babu, SK.Hameeda ....................................................................................... 202 A Novel Cloud Computing Infrastructure for Healthcare Application S. Nirmala Sugirtha Rajini, Dr. T. Bhuvaneswari ,.............................................................. 206 A Mobile Sensor Computing System to Scrutinize and Assess Road Surface Conditions Using RG2 S.Shalinidevi, N.Vidhya ..................................................................................................... 209 Intrusion Detection System for Manets A.Sowmya Narayanan ....................................................................................................... 213 Neural Network Model for Stock Price Prediction K. V. Sujatha, S. Meenakshi Sundaram .............................................................................. 219 Intrusion detection system using hybrid classification model J. Arokia Renjit1, Dr.K.L.Shunmuganathan....................................................................... 225 Intensified Message Encryption using Enigmatic text Kishore Bitra, V S R K A Anil Kumar, Navin J., Ramana M .............................................. 232 Palmprint Authentication System using Contourlet Transform Leo Vijilious. M.A., Subbiah Bharathi.V,........................................................................... 236 Adaptive LSB Based Data Hiding: Integrating Cryptography and Steganography P. Mohan Kumar, Dr. K.L.Shunmuganathan....................................................................... 239 Network Intrusion Detection Using Layered Approach and Conditional Random Fields Punitha Devi S., Deepthi E and Hari Pranes C V.,............................................................. 244 Watermarking an Image for Copyrights Using Trigonometry and Pythagoras Punitha Devi S ................................................................................................................. 249 Security Based Mobile Ad-Hoc Networks Ramya.R, Lavanya.S.......................................................................................................... 253

Page No. A Zero-Overhead Encryption for Multimedia Systems using Discrete Wavelet transforms Aparna Teki, Mallikarjuna P Avala ................................................................................... 258 Image Encryption using Visual Cryptograms Firoj Hussain Shaik, K Pradeep Vinaik, Srinivas Karedla ................................................... 262 Secured Double Data Compression Based On Huffman With Sparse Storage Prof. Biku Abraham, Prof. Dr. Varghese Paul..................................................................... 268 Intrusion Detection Based on NetFlow K.Kiran Kumar, K.Divya Krishna, J.Santhi ........................................................................ 271 Password Authentication using Cued Click Points R. Pitchandi, T. Sujatha, S. Asokan .................................................................................... 275 Secure Multicast Key Distribution For Multicast Communication R. Varalakshmi, Dr. V. Rhymend Uthariaraj....................................................................... 280 Grid Security for Digital Signature V. Anitha Moses1, M. Helda Mercy, S. Dhilip Kumar, N. Mohan ...................................... 286 Security in Cloud Computing Framework Based On Risk Assessments Balakrishnan.G, Ramesh.S ................................................................................................. 290 Strategy to Handle End User Session in Web Environment Divya Rishi Sahu, Deepak Singh Tomar............................................................................. 293 A Hybrid System for Clustering of Uncertain Heart Disease Dataset using Fuzzy C Means and Fuzzy Particle Swarm Optimization Anupriya S, Kavitha S........................................................................................................ 300 Ensuring Data Storage Security in Cloud Computing Syed.Azhad , Mr. Srinivas Rao .......................................................................................... 305 A FrameWork For Web Services Trust Using Security Token L.Fathima Mubashira, P.Kavipriya ..................................................................................... 309 Delegation and Revocation Using Role Based Access Control Nirmalrani V, Sakthivel P .................................................................................................. 313 Groundwater Level Prediction Using Fuzzy and ANFIS Method M.Kavitha Mayilvaganan, K.B.Naidu ................................................................................ 318 Integrating Web Services and Agent Technology with DIGIGEO Agent Lalitha M, Nisha S. .................................................................................................... 323

Page No. ARM based Advanced Microcontroller Based Architecture Mrs.Kavitha.V, Mrs.P. Meena priya Dharshini ................................................................... 327 Zigbee Based Wearable Personal Healthcare and Emergency Aid System Venkata Ramesh Mamilla, N.Venkata Ramana,Dr.G.Lakshmi Narayana Rao..................... 332 A Sensor Based Real-Time Monitoring System for Heartbeat and Respiration Rate Sadiq Ali, Satheesh ............................................................................................................ 336 Real-Time Application for Data-Acquisition Embedded System Based on Web Server G Anil Kumar, B.Prasanna Jyothi....................................................................................... 341 Efficient Calculation of Energy and Medium : A Cross-Layer Way V.Narendar, T.Swapna ....................................................................................................... 347 Search Engines-An Insight V.Vinu Chakravarthi & Dr.G.Veeramani............................................................................ 353 Improvement of Stability of a Generator Excitation Control System using Stabilizer with Time Delays G. Naveen kumar, V. Usha Reddy...................................................................................... 356 OFDM Channel estimation with two training symbols and Polynomial Equations Sandeep Parachakapu1, Ramesh.G ..................................................................................... 362 Green Computing C.Deepthi, N.Devi Bai ....................................................................................................... 367 A Novel Software Design Of Linear Decorrelator Detector For CDMA Multiple Access Interference B.Ayyappa Swamy, B.Chandran Mahesh, G.Nirmla kumari, ............................................. 371 Denoising of Spectral Data using Complex Wavelets C. Madhu, T. Sreenivasulu Reddy ...................................................................................... 377 Monitoring Chronic Diseases Using Secured WSN Integrated With Socially Interactive CDS on Cloud for u-Life Care S.Giridharan....................................................................................................................... 383 Artificial Neural Networks M.Saidharani reddy, M.Shasimohan reddy ......................................................................... 389 Simulation of Earthquakes and Tsunami through GSM Network C.Vignesh Kumar............................................................................................................... 396 Switch Automation of Smart Devices between Test beds using Distributed Control System Rajee Vuta, Y. Jaganmohan Reddy, A Mallikarjuna Prasad ................................................ 400

Page No. An Automated System for Feature Extraction from Phonocardiogram to Aid Disease Diagnosis S. Rajeswari ...................................................................................................................... 405 Intrusion Detection System in MANET at Cross layer Syed Azhad, Mr. Ramesh ................................................................................................... 408 An Efficient Packet Scheduler in Diffserv Networks D. Rosy Salomi Victoria, S. Senthil Kumar.......................................................................... 414 Virtualized Network Infrastructure for Secure Cloud Computing in VLANs P.Sivaraman, G.Barath, C. Priyadharsini , G.Appasami ...................................................... 420 Simulation of C4ISR System Architecture for Lower Echelon of Armed Forces Shambhuling R. Doddamani, Manish Singh ....................................................................... 426 Standardization of Field Firmware Upgrade for Profibus Devices B. Arunagiri, P.I. Basarkod, S.S. Manvi ............................................................................. 432 Modeling Massive RFID Data Sets: A Gateway-Based Movement Graph Approach Mr. N.Krishnaiah & G.Manikyalarao ................................................................................ 435 Reliability Growth Models In Software Productivity Chandrakanth G Pujari, Dr.Seetharam K ........................................................................... 439 Study On Software Fault and Failure Data in Software System J.Prabhu, Dr.N.Malmurugan............................................................................................... 444 A Hybride Framework For Automatic Carving And Replaying Differential Unit Test Cases Using System Test Cases B.Ramana, C. Sreedhar ...................................................................................................... 448 Is Cloned Code older than Non-Cloned Code? Syed.Azhad, T. Suneetha ................................................................................................... 451 A study on Agile Processes to Improve Software Quality B. Jyostna Devi1, S. Anusha, N. Veeranjaneyulu................................................................ 457 A review on taxonomy of software maintenance activities. D.Kavitha, Dr.Ananthi Sheshasaayee ................................................................................. 463

Budding Cloud Environment using Windows Azure for Application Fabrics Muthuvel.L, Rakesh Sharma, Suganthi K ........................................................................ 468

Proceedings of National Conference on Computing Concepts in Current Trends, NC4T’11 11th & 12th Aug. 2011, Chennai, India

Fingerprint Recognition system for Blind People A. Vanilavanya1, Shilpeechattopadhyay2, I. Ramalakshmi3 1&2

3

PG Student, Hindustan University, Chennai, Tamilnadu, India UG Student, Jayaraj Annapackiam CSI College of Engineering, Nazareth, Tamilnadu, India [email protected] stifle the use of biometrics on identity documents, in banking, and in benefits administration. The lack of common and clearly articulated industry positions on issues such as safety, privacy, and standards further increase the odds that governments might react rashly to unfounded accusations about the functions and uses of biometric technology. Biometric technology serves as the gatekeeper of confidential personal information. Biometric technology is used to erect a barrier between personal data and unauthorized access. Technically speaking, the devices create electronic digital templates that are stored and compared to “live” images when there is a need to verify the identity of an individual. The templates use proprietary and carefully guarded algorithms to secure the record and protect it from disclosure. Standing alone, these templates are of no use; they cannot be reconstructed, decrypted, or otherwise manipulated to reveal a person’s identity to someone else. Used this way, biometrics can be thought of as a very secure key, but one that cannot be passed on to someone else. Unless this biometric gate is unlocked by the proper key bearer, no one can gain access to that person’s information. Compared to other methods of proving identity—producing a driver’s license, showing a birth certificate, or revealing one’s family history—biometrics are the only currently known tools that can enhance personal privacy and still deliver effective solutions in situations that require confirmation of identity. Biometric technology is a major defense against identity theft. Identity theft—using stolen credit cards, phony checks, benefits fraud, network hacking, and other impostor scams to defraud businesses, government agencies, and consumers—costs billions of dollars per year. It is a problem that drives up prices of goods, increases taxes, complicates routine transactions, and strains law enforcement resources. Until recently, the only way to attack the problem has been to add expensive screening and administration procedures; however, steps such as hiring security guards, maintaining accurate databases, reviewing identity documents, administering password systems, and asking personal questions have proven to be costly, stopgap measures that can be defeated by enterprising crooks. Biometric technologies offer effective, low-cost solutions that streamline these traditional, labor-intensive processes. Biometric devices are an effective substitute because they create highly accurate digital records of a person’s physiological features. These records can be safely stored for later comparison against a live image that is captured from the user at the time the service or

Abstract— The objective of this paper is to provide

the fingerprint recognition system for blind people. In the industrial design field of human-machine interaction, the user interface is (a place) where interaction between humans and machines occurs. The goal of interaction between a human and a machine at the user interface is effective operation and control of the machine, and feedback from the machine which aids the operator in making operational decisions. Examples of this broad concept of user interfaces include the interactive aspects of computer operating systems, hand tools, heavy machinery operator controls and process controls. The design considerations applicable when creating user interfaces are related to or involve such disciplines as ergonomics and psychology. A user interface is the system by which people (users) interact with a machine. The user interface includes hardware (physical) and software (logical) components. User interfaces exist for various systems, and provide a means of: input, allowing the users to manipulate a system, and/or output allowing the system to indicate the effects of the users' manipulation. Keywords— Biometrics, Fingerprint.

I. INTRODUCTION The main objective of this project is to provide the ATM access for the Blind people. Blind people can’t use the ATM. They can’t use the button systems. It is very tough to access the ATM for them. It is very tough to withdraw the money from ATM for them. They can’t access the touch screens present in the ATM system. So to avoid all these problems here we introduced a new voice recognition system to access the ATM for the blind people. Through voice/speech the blind people can easily access the ATM. And also it is very peaceful for them. The public, opinion leaders, regulators, and legislators need the facts about biometric technology. During the past decade, the science of biometrics has matured into an industry that offers real-world solutions to serious problems faced by businesses, schools, and government agencies. Hardware and software produced by biometric manufacturers offer a safe and reliable means to ensure privacy, protect assets, confirm identity, and guard against unauthorized access. Clearly, the marketplace has begun to accept biometrics as a better alternative to less-secure screening and identity verification processes. This success has not yet led to broad public awareness about what biometrics do and how they work. At best, this means that consumers might resist using the technologies in place of more antiquated, but familiar, processes. At worst, regulators and legislators will make ill-informed decisions that will Sathyabama University

1


biometric projects. Unlike fingerprints, the human hand isn’t unique. It is also known that one could change the geometry of their hands by taking a hammer and smashing it. One drawback for this type of identification is that individual hand features are not descriptive enough for identification. Hand geometry is the granddaddy of the modern biometrics by virtue of a 20-year history of live applications. There have been six different hand-scanning products developed over this span, including some of the most commercially successful biometrics to date. Hand geometry biometric is by far less accurate than other biometric methods. As an extension to hand geometry analysis, a recent creation by LiveGrip™ analyzes the veins, arteries and fatty tissues of the hand. Sixteen scans are taken and a template of the individual’s hand is stored. This method of identification could be costly in terms of storage of templates because sixteen scans are taken, but at the same time, this method does analysis of distinct characteristics of an individual that cannot be changed (i.e. vein geometry, arteries, and fatty tissues of the hand). San Francisco International Airport, the USA’s fifth largest airport, has been using hand geometry based systems to authenticate airport employees for almost 10 years. The U.S. Federal Bureau of Prisons uses hand geometry to track movements of its prisoners, staff and visitors within prisons. Once a person enters the system, they must have their hand scanned.

benefit is demanded. All of the devices are nonintrusive, user-friendly units that recognize features such as an iris, a voice, a signature, a fingerprint, a hand, or a face. This gives end users such as banks, merchants, government agencies, and employers extraordinary control over transactions without inconveniencing or embarrassing the customer. II. LITERATURE REVIEW Biometrics has been around for many years. The Bertillion System, Bertillonage, or anthropometry was not based on fingerprinting but rather relied on a systematic combination of physical measurements. These measurements included measurements of the skull width, foot length, and the length of the left middle finger combined with hair color, eye color, as well as face and profile pictures. By grouping the data any single person could be placed into one of 243 distinct categories. For the next thirty years, Bertillonage was the primary method of biometric identification. Another example of biometrics in practice was a form of finger printing being used in China in the 14th century, as reported by explorer Joao de Barros. He wrote that the Chinese merchants were stamping children’s palm prints and footprints on paper with ink to distinguish the young children from one another. Fingerprints are unique to each individual and each individual has their own pattern in their fingerprints. This type of identification has been successfully used by the police to capture criminals and to find missing children. A fingerprint records the patterns found on a fingertip. There are a variety of approaches to fingerprint verification. The traditional method, which is used by police, matches minutiae (details of the fingerprint). Some other approaches are pattern matching, and moiré fringe patterns. There are some verification approaches that can detect if a live finger is presented, but not all of these approaches can provide this type of information. If fingerprintscanning techniques were to be incorporated into the flight deck to provide continuous authentication, liveness detection or testing would be a requirement for the system. Fingerprints serve to reveal an individual’s true identity and the practice of using fingerprints as a means of identification has been a helpful aid to those who chose to use this type of identification. Fingerprints are unique in the sense that there has not been any type of pattern duplication by two different people. Not even a single instance has been identified or discovered at this time. This uniqueness also applies to identical twins, as well as triplets, quadruplets, and quintuplets. One good thing about fingerprints is that any type of burn (superficial), abrasions, or cuts do not affect the ridge structure, thus the fingerprint pattern is unaffected. Hand geometry involves analyzing and measuring the shape of the hand. This type of biometric offers a good balance of performance characteristics and is relatively easy to use. The ease of integration into other systems and processes, coupled with ease of use, makes hand geometry an obvious first step for many Sathyabama University

III. FINGERPRINT RECOGNITION Every person possesses unique fingerprints from any other individual. As with other biometric methods, fingerprint identification is based on two basic premises: Invariance: The basic characteristics of the fingerprint do not change with time. However, there are instances where a fingerprint reader may not accept a legitimate user because of a cut on the finger or dry skin. Singularity: The fingerprint is unique to each individual and no two people have the same pattern of fingerprints. Fingerprint-based identification has been used for a long time and is routinely used in forensic laboratories and identification units all around the world. Fingerprint evidence has also been accepted in courts of law for nearly a century. A) How Fingerprint Recognition Works A fingerprint-scanning device is pretty easy to use. The user must place his or her finger on the device and certain characteristics of the fingerprint image are extracted into templates known as minutiae. The characteristics of each finger are different from each other. As with other biometric methods, general fingerprint matching process involves three phases: The acquisition phase or enrollment is where the fingerprint is scanned using a fingerprint sensor. Many sensors are available that capture a fingerprint based on the optical, capacitive, pressure, thermal, or ultrasound domain. The capturing of the image is 2


identification. Some sophisticated methods currently available are: Optical sensors with CCD or CMOS cameras. The finger is placed or pushed on a plate and is illuminated by a LED light source. Through a prism and a system of lenses, the image is projected on a camera. Frame grabber techniques are used and the image is stored and ready for analysis. Ultrasonic sensors. By using ultrasonic sensors, a scan of the fingerprint with a resolution of about 500 dots per inch is possible. This technique may be able to offer templates, which are full of useful detail of fingerprint information. Electronic field sensors. This technique creates an electric field with which an array of pixels can measure variations in the electric field that are caused by the ridges and valleys in the fingerprint. Capacitive sensors. This technique is similar to electronic field sensors except that when the finger is placed on the sensor, an array of pixels measures the variation in capacity between the valleys and the ridges of the fingerprint. Temperature sensors. This technique makes a distinction between the temperature of the ridges and the temperature of the valleys on the fingerprint. A temperature scan can be taken by simply swiping the finger over the sensor. Although these techniques seem very advanced and accurate, it is still possible that a desperate attacker may attempt to spoof a legitimate user by creating fake fingers. Fake fingers can be made both by the cooperation of the legitimate user (i.e. for testing methods) or without the cooperation of the legitimate user by lifting a fingerprint off of a keyboard or coffee mug. Those traces of fingerprints are known as latent fingerprints.

made easier because the sensors only require a simple touch of a finger. The live presentation phase is when the user shows his/her biometric information to the biometric device. During the matching phase, the features of the scanned fingerprint (live template) are compared to the stored template in the database Since traditional methods of fingerprinting (i.e. fingerprint capturing using ink and paper) are not used that often in fingerprint recognition technology, we are able to capture more details of that fingerprint. In addition, the newer methods of fingerprint recognition are more hygienic and less intrusive. In order for the system to offer accurate results the user has to be willing to use it correctly and they have to be willing to fully understand how the system works. For example, the user will have to know how long they would have to press their finger on the reader in order to obtain accurate results. B) Fingerprint Recognition: User Influences Fingerprint recognition methods contain influences that may affect the outcome of the authentication process of the device. Some influences are: Fingernail growth may have an effect on how firmly the user is able to place his/her finger on the scanning device. This may result in inaccurate results from the device or the user may be rejected altogether by the system. This influence also extends itself to the use of artificial nails that the user may apply to real fingernails. Fingerprint fineness may also have an effect on how the device is able to pick up details of the fingerprint. This depends on how well the depth and the spacing of ridges are on the users fingers. This influence is not controllable by the user so proper enrollment from the beginning needs to be done as well as proper placement of the finger on the scanning device at the time of authentication. There may be fingerprint-scanning devices that alleviate this influence by offering a sensitive “touching area” for the user. The condition of the fingerprint may have an effect on the outcome of the device because the user may have dry, cracked, or damp fingers. If the user has dry, cracked, or damp fingers at the time of enrollment or at the time of authentication the scanning device may not be sensitive enough to compensate for these characteristics. Another influence that falls into this category is scars and/or scratches on the fingertips of the user. Scars and scratches, depending on their location, may cover up some important characteristics of the fingerprint that the scanning device is looking for to extract. On the other hand, it may be possible for the scanning device to simply use the scar on the fingertip as a part of the characteristic extracted.

IV. FINGERPRINT PROCESS Fingerprint matching is the process used to determine whether two sets of fingerprint ridge detail come from the same finger. There exist multiple algorithms that do fingerprint matching in many different ways. Some methods involve matching minutiae points between the two images, while others look for similarities in the bigger structure of the fingerprint. In this project a method for fingerprint matching based on minutiae matching is used. However, unlike conventional minutiae matching algorithms our algorithm also takes into account region and line structures that exist between minutiae pairs. This allows for more structural information of the fingerprint to be accounted for thus resulting in stronger certainty of matching minutiae. Also, since most of the region analysis is preprocessed it does not make the algorithm slower. The algorithm for matching was not created; however, the process in which the regional data is obtained is explained in this paper. Evidence from the testing of the preprocessed images gives stronger assurance that using such data could lead to faster and stronger matches. Extensive research has been done on fingerprints in humans. Two of the fundamentally important conclusions that have risen from research

C) Fingerprint Recognition: Techniques The techniques used to gather fingerprint information has changed greatly over the years. Some sophisticated fingerprint scanning methods have emerged since the beginnings of this method of

Sathyabama University

3


are: (1) a person's fingerprint will not naturally change structure after about one year after birth and (2) the fingerprints of individuals are unique. Even the fingerprints in twins are not the same. In practice two humans with the same fingerprint have never been found.

As shown in the above figure, the users enroll their fingerprint data which is converted into a secure digital hash and stored in an encrypted format within the secure database. The stored data is then used to verify the user while authenticating with the system.

A) The Basics A fingerprint is comprised of ridges and valleys. The ridges are the dark area of the fingerprint and the valleys are the white area that exists between the ridges. Many classifications are given to patterns that can arise in the ridges and some examples are given in the figure to the right. These points are also known as the minutiae of the fingerprint. The most commonly used minutiae in current fingerprint recognition technologies are ridge endings and bifurcations because they can be easily detected by only looking at points that surround them.

Fig. 3 Verification The fingerprint data is captured from the user and converted into a digital hash and passed on to the server which performs a secured encrypted verification with the biometric data earlier stored. The server can then pass the success or the failure of the verification process to the application which requested it. V. EXISTING AND PROPOSED SYSTEM In the Existing Systems, only the normal person can use the technology such as touch screen and other facilities that are being provided by the ATM’s. Here there is no provision for the usability of Blind people. No authentication can be given for the password of Blind users. Withdraw and other ATM operations are impossible with Blind people, due to the absence of Interactive Voice Response System. But in the proposed system even blind people can use the ATM facilities. We will implement our proposed thing with the help of fingerprint and the leading language such as visual studio. The Brain Pulse of the user is collected and stored during his account opening time. When the user use the ATM machine at the time the current brain pulse print of the user will match with the existing stored copy with the help of Array Matching Algorithm.

Fig. 1 Basic Fingerprint B) The Process The system used in this project provides a secure, reliable and extremely effective way to authenticate users. It empowers the IT administrators establish the identity of the users beyond doubt effectively every time. C) The system The system reads the fingerprint data from the user and converts into a secure digital data which then used to authenticate the user using a secure verification process within the server. The users are enrolled with the system using their fingerprint data which is stored in encrypted format within the secure database.

VI. RESULTS The results of this work project is shown in the following figures.

Fig. 2 Conversion of Fingerprint into Digital Data

Fig. 4 Entering into the ATM Application


4


Fig. 9 Comparing the Input with Stored Data Fig. 5 Recognition Main Menu Input match with the stored data

Fig. 6 Registration Form for the User

Fig. 10 Input matched with Stored Data VII. CONCLUSIONS In this project, an ATM system for Blind people was implemented by using Biometrics. Biometrics is a technology that can simplify the process of authentication. Biometrics can be best used in situations where specific identity or exception identity is desired. While biometric security systems can offer a high degree of security, they are far from perfect solutions. Sound principles of system engineering are still required to ensure a high level of security rather than the assurance of security coming simply from the inclusion of biometrics in some form. The risks of compromise of distributed database of biometrics used in security applications are high, particularly where the privacy of individuals and, hence, non-repudiation and irrevocability are concerned. It is possible to remove the need for such distributed databases through the careful application of biometric infrastructure without compromising security. The influence of biometric technology on society and the potential risks to privacy and threats to identity will require mediation through legislation. For much of the short history of biometrics, the technological developments have been in advance of the ethical or legal ones. Careful consideration of the importance of biometric data and how they should be legally protected is now required on a wider scale.

Fig. 7 Creating the Digital Data for the Fingerprint Image

Fig. 8 Fingerprint Matching

REFERENCES [1] [2]


5

An MIT Enterprise Technology Review, Paper, November 2001. Anderson, E.A., “A Demonstration of the Subversion Threat: Facing a Critical Responsibility in the Defense of Cyberspace”. Master’s Thesis, Department of Computer Science, 2002, Naval Postgraduate School, Monterey, CA.

Proceedings of National Conference on Computing Concepts in Current Trends, NC4T’11 11th & 12th Aug. 2011, Chennai, India [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

Bruderlin R. “What is biometrics?” Paper, 1999-2001. Bonsor K. “How Facial Recognition Systems Work”, 2001. Chua J. Biometrics, “The future of security”, CBC News Online, September 2001. Daugman J. “Combining Multiple Biometrics”, the Computer Laboratory at Cambridge University, 2000. Daugman J. How iris recognition works, 2000. Degami A., Wiener E.L. Procedures in Complex Systems: The Airline Cockpit, NASA contractor report 177642, Moffett Field, CA: NASA Ames Research Center, 1997. Degami A, Weiner E.L. Cockpit Checklists: Concepts, Design, and Use, NASA contractor report 177549, Moffett Field, CA: NASA Ames Research Center, 1993. Degami A. On the Design of Flight Deck Procedures, NASA contractor report 177642, Moffett Field, CA: NASA Ames Research Center, 1994. Esser M. “Biometric Authentication”, Essay, October 2000. General Aviation Manufactures Assoication, Recommended Practices and Guidelines for Part 23 Cockpit/Flight Deck Design GAMA Publication No. 10 September 2000. Go Team 9- Biometrics, U.S Department of Transportation; Transportation Security Administration Technical Report. Govindarajan S. Are These Prying Eyes, article, available http://www.krify.com/articles/pryingeyes.htm (last accessed: November 2002) Hong L. and Jain A.K. Integrating Faces and Fingerprints, IEEE Trans. Pattern Anal. Machine Intell., Vol. 20, No. 12, pp. 1295-1307, December 1998. Info Security Magazine, “Biometrics Technology: Making Moves in the Security Game", pp. 28-34 Volume 12 #3 March 2002. International Biometrics Group, Tech Reports “Facial Scan Technology: How it works”, 2002. International Biometrics Group, Tech Reports “Iris-Scan: How it works”, 2002. Jain A.K and Arun R. Learning user-specific parameters in a multibiometric system, Department of Computer Science and Engineering-Michigan State University, no date given. Kolettis H. Stepping up Security, 2002. Liu S. and Silverman M. A practical guide to biometric security technology, January 2000.


6


Morphological Parsing of Tamil Verb Using Finite State Automata R.Padmamala Research Scholar, University of Madras Abstract Verb + Aspect + Voice + Modal + Tense + PNG + This paper discusses the efficiency of using Clitics1 + Clitics2 + Clitics3 + Clitics4

Finite State Automata for morphologically parsing a given Tamil verb. Different orientations of Algorithms are taken into consideration and it is suggested that left-to-right algorithm incorporating the FSA concept is more suitable for morphological parsing in Tamil. Also two different ways of storing the valid paths are suggested in this paper. These ways are discussed in detail. Keywords: Morphology, Morphotactic rules, Word classes, FSA, Transition table, Valid paths.

1.

Before finalising on an algorithm to do the parsing, let us discuss the various parameters of an algorithm and the qualities an algorithm need to possess. 2.3 Parsing algorithm Parsing algorithm can be differentiated on the basis of the following parameters. Orientation: Top-Down, Bottom-Up and Mixed Direction : Left-to-Right, Right-to-Left and Mixed Search : Breadth-First and Depth-First

INTRODUCTION

Morphology is the study of the way words are built up from smaller meaning bearing units called Morphemes. Morpheme is the minimal meaning bearing unit in a language. Morphemes can be classified into Stems and Affixes. Affixes are further divided into Prefixes, Suffixes, Infixes and Circumfixes. In Tamil language, affixes are mostly suffixes only.

Now let us see our present algorithm and discuss its category and qualities. As we have seen earlier the structure of a verb is as follows. Verb + Aspect + Voice + Modal + Tense + PNG + Clitics1 + Clitics2 + Clitics3 + Clitics4

Morphological parsing divides a word into its stem and suffixes. Morphological parsing is done based on the morphotactic rules. Morphotactic rules govern the ordering of morpheme within a word. For example, in Tamil, Plural morpheme can only follow the noun and cannot precede it, i.e. a plural morpheme can only be a suffix and not a prefix. Such morphotactic rules are to be followed to parse a given word. Also orthographic rules are to be taken into consideration. These rules are to govern the changes that occur in a word when two morphemes combine. 2.

Following the stem/root word, slots are allotted for various categories of suffixes. Certain slots may be vacant or all the slots may be filled based on the given word. Therefore in morphological parsing of a Tamil word in a Right-to-Left fashion, root check is necessary after the removal of every suffix. But whether to go for root check is not predictable till all the slots are checked out. The following is the Right-to-Left algorithm.

MORPHOLOGICAL PARSING IN TAMIL

2.4 Algorithm

Before getting into how morphological parsing is done in Tamil, let us first list out the grammatical categories. The grammatical categories can be divided into Word classes and Particle classes.

MorphologicalParsing() { BooleanCanBParsed= Check_Suff(GivenWrd) if CanBParsed = True Display the parsed output else Display error message } Check_Suff(Word) { While not Word.empty Open Suffix file While not End of File() Read Suffix file if word.endswith (Current Record Value) if positioning of suffix in correct slot Word = Substring (Word – Current Record Value)

2.1 Word classes and Particle classes The following are the word classes in Tamil: Noun, Verb, Adjective, Adverb, Intensifier, Conjunction, Interjection, Introductory, Summoners and Responsives. The particle classes in Tamil are Case, number, postposition, verbal particles, clitics and fillers. For computational convenience, a word can be stored in a database along with its word class or particle class. 2.2 Structure of a Tamil Verb A Tamil Verb has the following structure:

Rootcheck(Word) Sathyabama University

7


possible can be represented in a Table 1 transition table. The following is the transition table for the FSA.

if found in Root Check conjugation rule if conjugation rule obeyed Return True else Return False else Check_suff(word)

Based on the above seen transition table, only 191 combinations of stem and suffixes are possible for a Tamil verb. The database can be organized either as a transition table or a text file containing all possible valid paths, here in this case where it sums up to 191.

}

The following will be the valid paths in case database is organized as a text file.

Now let us check the qualities of this algorithm. 1.

Root check is done many times after the suffix category clitics are stripped off. Any slot can be empty. But if the slots are not going to be empty, root check is merely a waste of time. Thus unwanted root checks increase the time taken. Thus efficiency is brought down as number lines of code and iterations are very high. Modifying or introducing any changes in this algorithm is tough thereby decreasing the flexibility of the algorithm. In case of wrong input where conjugation rules are not obeyed, conjugation check will be done only after stripping of the suffixes, applying the orthographic rules to extract the morpheme. This is very inefficient as so many operations which are not necessary are done. To overcome the above said problems, an alternate algorithm which uses the FSA concept is being proposed in this paper.

2.

3. 4.

3.

Now let us see the algorithm. Morphological parsing() { Boolean CanBParsed = Check_Suff(GivenWrd) if CanBParsed = True Display the parsed output else Display error message } Check_Suff(Word) { While not word.empty { Rootcheck(Word) If Root found Word = Word – Root Open Suffix file While not EOF() { Read Suffix file If word. Begins with (Current Record) Outputstr = Outputstr currentstate Word = Word – Suffix Break } Check Transition table If outputstr can be arrived at Return True Else Return false } }

ALGORITHM FOR MORPHOLOGICAL PARSING USING FSA

Finite State Technology is mathematically well understood and applications to natural language are efficient, robust and elegant. Finite State Machine is a machine consisting of states including one start state and one or more final states. Transitions between states are possible only if the input is recognised. Path is a sequence of transitions over arcs to a particular state. 3.1

FSA for Tamil (Verb) Verbal Inflection Vb Asp

Vce

Md

T

P

Pst C1 2

3

4 _

Q0 Q1

_

_

_

_

_

_

_

_

_

Q1 _

Q2

_

_

_

_

_

_

_

_

_

Q2 _

_

Q3

_

_

_

_

_

_

_

_

Q3 _

_

_

Q4

_

_

_

_

_

_

_

Q4 _

_

_

_

Q5 _

_

_

_

_

_

Q5 _

_

_

_

_

Q6 _

_

_

_

_

Q6 _

_

_

_

_

_

Q8

Q9 Q10 Q11

Q7 _

_

_

_

_

_

_

Q8

Q9 Q10 Q11

Q8 _

_

_

_

_

_

_

_

Q9 Q10 Q11

Q9 _

_

_

_

_

_

_

_

_

Q10 Q11

Q10 _

_

_

_

_

_

_

_

_

_

Q11

Q11 _

_

_

_

_

_

_

_

_

_

_

Q7

Now let us see how the quality of the algorithm has improved in following this method. 1.

2.

Fig.1 (in the last page) shows the FSA for Tamil(Verb) Inflection. The different transitions Sathyabama University

+

8

Efficiency: Number of iterations has been brought down drastically as root check is done only once. Also number lines of code will come down drastically. Thus the algorithm becomes more efficient. Completeness: This algorithm deals with Thogai (Compound words) also thereby avoiding that exception. Moreover the exception stated in the discussion of the previous algorithm is also dealt with.


3.

Flexibility: Any new changes or introduction of new suffixes can be easily done in this algorithm. 4. CONCLUSION

5.

REFERENCES

1. Koskenniemi.K. Finite-State Parsing and Disambiguation, University of Helsinki, Department of General Linguistics.

First of all, I would like to thank my Guide Dr.N.Deivasundaram for his guidance, invaluable suggestions and constant technical support.

2. Jonathan Calder, Paradigmatic Morphology, University of Edinburgh, Scotland, Centre for Cognitive Science.

The proposed algorithm incorporating FSA eliminates the problems of Right-to-Left algorithm. But still many ambiguities in morphological parsing cannot be solved by this algorithm for the purpose of tagging. To resolve those ambiguities, Syntactic Parsing is essential.

3. Daniel Jurafsky & James H.Martin, Speech and Language Processing, 2003, Pearson Education Inc.

Cl Cl1 Aspe q

Reg.

q Plural

Cl1

Voi Filler q

q

Mo dal

q

PN G

q Tense

q

q

Cl 4

q11

Cl

Postpo stn Cl 4

Cl

Cl4

Cl2

Cl

q10

Cl3

Cl Cl3

Fig 1. Finite State Automaton for Tamil (Verb) Verbal Inflection


9


Brain controlled artificial legs P.M. Uma1, G. Indumathi2 III yr. B.Tech-IT, Sri Sairam Engineering College Tambaram, Chennai -44 [email protected], [email protected]

Abstract: This paper describes a brain controlled

tional device, from simple circuits to the complex

robotic leg which is designed to perform the normal

microprocessors and microcontrollers.

operations of a human leg. After implanting this leg in a human, the leg can be controlled with the help of user’s brain signals alone. This leg behaves similar to a normal human leg and it can perform operation like walking, running, climbing stairs etc. The entire system is controlled with the help of advanced microcontrollers and digital signal processors. The signals are taken out from the human brain with the help of electroencephalography technique. The person can perform operations like walking, running etc just by their thought. This system will be very much suitable for those who lost their legs in accidents and the proposed system is hundred percent feasible in the real time environment with the currently available technology.

An interesting question for the development The Brain Controlled Artificial Legs are very

of a BCI is how to handle two learning systems: The

much cost effective when compared to the normal

machine should learn to discriminate between differ-

Artificial legs which is available in the market. The

ent patterns of brain activity as accurate as possible

reduction in cost of the proposed system is found to be

and the user of the BCI should learn to perform dif-

above 80% when compared to the existing system.

ferent mental tasks in order to produce distinct brain

Moreover, the user can have full control over the arti-

signals. BCI research makes high demands on the

ficial legs which is not possible in the existing system.

system and software used. Parameter extraction, pattern recognition and classification are the main tasks

INTRODUCTION

to be performed in a brain signals. In this paper it is

A brain-computer interface (BCI), sometimes

assumed that the user of this system has one leg which

called a direct neural interface or a brain-machine

is functioning fully and the system is designed accord-

interface, is a direct communication pathway between

ingly. This system can be extended for both the legs

a human or animal brain and an external device. In

and it is not limited to the basic operation of human

this definition, the word brain means the brain or

legs such as walking, running, climbing stairs etc. It

nervous system of an organic life form rather than the

can also perform operations like cycling, hopping etc

mind. Computer means any processing or computa-


10


BRAIN WAVES

The next brainwave category in order of frequency is Alpha. Where beta represented arousal,

Electrical activity emanating from the brain

alpha represents non-arousal. Alpha brainwaves are

is displayed in the form of brainwaves. There are four

slower and higher in amplitude. Their frequency

categories of these brainwaves ranging from the most

ranges from 9 to 14 cycles per second. The next state,

activity to the least activity. When the brain is aroused

theta brainwaves, is typically of even greater ampli-

and actively engaged in mental activities, it generates

tude and slower frequency. This frequency range is

beta waves. These beta waves are of relatively low

normally between 5 and 8 cycles a second. A person

amplitude, and are the fastest of the four different

who has taken time off from a task and begins to day-

brainwaves. The frequency of beta waves ranges from

dream is often in a theta brainwave state. The final

15 to 40 cycles a second.

brainwave state is delta. Here the brainwaves are of the greatest amplitude and slowest frequency. They typically center around a range of 1.5 to 4 cycles per second. They never go down to zero because that would mean that you were brain dead. But, deep dreamless sleep would take you down to the lowest frequency. Typically, 2 to 3 cycles a second.

Fig 1: Different Types of Brain Waves


11


In the proposed system alpha waves and beta

form of electrical signals with the help of electrode

waves are used from the brain for signal processing. It

caps. The following figure shows the different types

is assumed that the person is in alpha state and beta

of waves and also the mental state of the person.

state (which is the case normally) and these waves are

Those waves usually vary from a frequency of 1Hz to

taken out from the human brain and converted in the

40 HZ.

GENERAL BLOCK DIAGRAM OF THE SYSTEM:

Fig2 : General Block Diagram of the Proposed System Fig 2 shows the general block diagram of the

brain. High-density arrays (typically via cap or net)

proposed system. Electrode cap is placed in the scalp

can contain up to 256 electrodes more-or-less evenly

of the person. The signals taken out from the human

spaced around the scalp. The main function of the

brain will be in the range of mV and µV. Hence they

electrode cap is to take the brain signals in the form of

are fed to an amplifier. Then it is sent to a Analog to

electrical signals. The signals taken out from the Elec-

Digital Converter to convert the analog brain signals

trode cap are fed to an amplifier.

in to digital form. Then it is sent to a signal processor where parameter extraction, pattern classification and pattern identification are done. These digital signals are fed as input to microcontroller unit. The last four units (Amplifier, Signal Processor, Analog to Digital Converter and Microcontroller Unit) are placed inside the artificial leg. The output of the microcontroller unit is fed to the driving circuit. Let us see about these blocks in detail. Electrode Cap: Fig 3 shows a person wearing an electrode cap. These electrode caps contains electrodes which are placed on the skull in an arrangement called 10-20 system, a placement scheme devised by the internaFig 3: A person wearing electrode cap

tional federation of societies of EEG. In most applications 19 electrodes are placed in the scalp. Additional electrodes can be added to the standard set-up when a clinical or research application demands increased spatial resolution for a particular area of the Sathyabama University

12


Signal Processor: Using the output signal from the A/D converter, parameter extraction, pattern classification and pattern identification are done. Then the signals are fed to a Fast Fourier Transform Unit. This is done to simplify the calculations. An FFT algorithm computes the result in O (N log N) operations instead of O (N2) operations. The output signals from the signal processor are fed to a Microcontroller unit. Microcontroller unit: The output signals from the signal processor are fed to a microcontroller unit. This microcontroller Fig 4: Placement of electrodes in 10-20 system

unit performs the robotic operation with the help of a stepper motor. It will control the operations such as

Amplifier:

walking, running, etc depending upon the input signal. The output signal from the electrode cap will

For different patterns of input signals it will be pre-

be in the range of mV and µV. So, these signals will

programmed to do a specific operation. The reference

not be suitable for signal processing. Hence these sig-

signal will be already stored in the microcontroller

nals are fed to an amplifier. Each electrode is con-

memory in digital form. Usually an 8 bit or a 16 bit

nected to one input of a differential amplifier (one

microcontroller is preferred depending upon the num-

amplifier per pair of electrodes); a common system

ber of operations to be performed. The complexity of

reference electrode is connected to the other input of

the microcontroller programming increases with the

each differential amplifier. These amplifiers amplify

number of operations which has to be performed.

the voltage between the active electrode and the referWorking of the Proposed System:

ence (typically 1,000–100,000 times, or 60–100 dB of voltage gain).

For every human activity the brain waves changes its pattern. For example, if a person moves

Analog to Digital Converter:

his/her hands then a specific pattern of brain wave is The output signals from the amplifier are

obtained and if the same person moves his/her legs

analog in nature. They also contain some unwanted

then a different pattern of brain wave is obtained.

signals. Hence the output signals are filtered using

Even if a person thinks of moving his/her legs a brain

high pass and low pass filters. The high-pass filter

wave of specific pattern is produced and it is sent to

typically filters out slow artifact whereas the low-pass

the legs and then the operation of moving the legs is

filter filters out high-frequency artifacts. After the

performed. The same brain waves are produced even

signal is filtered they cannot be directly fed to a digi-

for a person who is not having his/her legs. But the

tal signal processors and microcontroller unit as they

operation of moving the legs will not be performed

are in analog form. Hence these signals are sent to an

due to the absence of legs. So, just by thinking of

Analog to Digital converter to convert the incoming

moving the legs, a brain wave which is capable of

analog signals in to digital signals.

performing a specific operation is generated in the


13


Figure 5, 6, 7, 8, 9 and 10 shows some of the

brain. Due to the lack of the appropriate system, the

pictures of artificial legs.

activity will not be performed successfully. In the proposed system, the brain waves are pre-recorded for each operation to be performed and these waves are used as reference signals. These signals are stored in the microcontroller memory. For each reference signal in the microcontroller memory, the robotic leg is pre-programmed to do a specific operation. When the reference signal matches with the actual signal from the user’s brain, the robotic leg will do the pre-programmed operation with the help of the microcontroller.

Fig 5: Proposed Model of the artificial legs

For example, let us say that the user is thinking of walking. So a brain wave will be produced. These waves are processed and then it is converted in to digital signals. These signals are compared with the pre-recorded reference signals and a match in the signal pattern will be found in the microcontroller. The operation for this particular pre-recorded signal will be pre-programmed in the microcontroller circuit i.e. walking and thus the microcontroller will send the control signal to the artificial robotic leg and the robotic leg will perform the required operation.

Fig 6: Internal appearance of the artificial leg.

Usually a stepper motor controlled robotic leg is used for this purpose. Similarly to walking, other operations can also be performed using the artificial leg. This system is very user friendly and the system can be designed according to the user’s requirements i.e. the number of operations required for the user can be fixed by him and the system can be designed accordingly. So the number of operations that has to be performed by the leg can be increased or decreased and the complexity of the design varies accordingly. This idea can be extended for both the legs and both the legs can be made to do operations like

Fig 7: External Appearance of the artificial leg

walk, run etc simultaneously. Thus the system is versatile. This system is hundred percent feasible in the real time environment and it can be implanted to any human irrespective of their age. Sathyabama University

14


Power Supply: These artificial legs are powered by a small lithium-ion battery which has to be charged once in 2 day.

Lithium-ion batteries have very high charge

density (i.e. a light battery will store a lot of energy). They are of ultra-slim design and hence they occupy very less space. Moreover their life time will be longer when compared to other batteries. Hence they are preferred when compared to other batteries. Moreover they have longer life time when compared to other batteries. Fig 8: Walking down the stairs using artificial legs

Normal Artificial legs: Normal Artificial Legs, available in the market, is very costly. They use a group of sensors and a complex algorithm for their operation which makes the existing system very costly. This disadvantage has been overcome in the Brain Controlled Artificial Legs as they don’t use any sensors for their operation. Moreover the normal artificial legs are 100% dynamic in operation. Hence the chance of occurrence of an error is more in those systems. External appearance and output of both the legs are same. But the method of operation is different. Hence the Brain Controlled Artificial Legs are cost effective. Difference between the Brain Controlled Artificial

Fig 9: Artificial legs fitted to both the limbs

legs and the Normal Artificial Legs:

Fig 10: Walking with artificial legs.


15

Brain Controlled Artificial Legs

Normal Artificial Legs

1.

Ease of Construction

1. Complex in construction

2.

Cost is not more than Rs.5,00,000

2. Cost is about $80,000$90,000(Rs.35,00,000 to Rs.40,00,000)

3.

User can have full 3. User cannot have full control control over the ar- over the artificial leg. tificial leg.

4.

Semi-Automatic

4. Fully Automatic

5.

Sensors are absent.

5.Sensors are present

6.

Requires simple control unit.

6. Requires complex control unit.


10. http://www.engadget.com

CONCLUSION:

11. http://www.ibva.com/

Forty years ago, the technology was so basic. Newton said. “Leg sockets were made out of wood,

12. http://www.gtec.at/

offering the equivalent of a door hinge at the knee”.

13. http://www.sciencedaily.com

But with the recent advancement in the technology,

14. http://www.bciresearch.org

Brain Controlled Artificial leg can be made as a real15. http://www.eurescom.de

ity. The performance of the proposed system will be better than the existing artificial legs as the user has full control over the Brain Controlled Artificial Legs. Hence it behaves like a normal human leg. The builtin battery lasts anywhere from 25 to 40 hours so it can support a full day’s activity. The recharge can be performed overnight or while traveling in a car via a cigarette lighter adapter. The cost of the proposed system is found to be very less when compared to the existing ones. So, even the middle class people who cannot purchase the existing artificial legs can make use of this proposed system. With this system life can be made easier for the handicapped persons and they can also do their day-to-day activities normally without any difficulties. REFERENCES: 1.

“Digital Signal Processing Principles, Algorithms and Applications” by J.G.Proakis and D.G.Manolakis.

2.

“Digital Signal Processing: Principles, Devices and Applications” by Norman Barrie jones and J.D.Mack Watson.

3.

“A guide to Methods in the Bio-Medical Sciences” by Ronald B.Corley.

4.

“Handbook of Bio-Medical Instrumentation” by R.S.Khandpur.

5.

“Integrated Electronics: Analog and Digital Circuits and System” by Jacob Millman and Christos C.Halkias.

6.

http://www.wikipedia.org

7.

http://www.ece.ubc.ca

8.

http://www.electrodesales.com

9.

http://www.moberg.com


16


A Comparative Analysis on Audio Visual based e- learning N. Muthu Meenakshi1 N. Mahadevan2 R.Karthiya Banu3 .Saravanan4 1. .II MCA Loyola Institute Of Technology, Chennai,e-mail: [email protected] 2. II MCA Loyola Institute Of Technology,e-mail: [email protected] 3. , Asst Prof & Head, Department of Computer Applications, Loyola Institute of Technology, E-mail:- [email protected] 4. Research Scholar, Sathyabama University. E-mail : [email protected]

In order to increase the efficiency of learning and reduce the difference between the e-learning and traditional learning, e-learning system should provide much more friend interface, convenient interaction, abundant and interesting courses.

ABSTRACT: This paper introduces a Audio visual based elearning system which has the ability to evaluate, apply, or create conceptual visual representations and makes the E-learning efficient and effective. A very important part of eLearning online is how well courses are perceived by the target audience because perception determines how well they will understand the content. In order to perceive, people need sensory stimulation. In this type of learning, visual stimulation is critically important. Online learning is heavily dependent on visual elements such as text, graphics, pictures, videos and so on. It is based on client and server relationship So that it makes more interactive, with the help of the system, an instructor can receive instant voice and video feedback’s from the student and check student’s performance at any time. This can be done by several tools. In this paper we are using net meeting from Microsoft windows and Bingo’s chat for e-learning and made comparative analysis.

To accomplish visual based E-learning effectively, we are using Net meeting and Bingo’s chat. 2.

E-learning can be defined as instructional content or learning experiences delivered or enabled by electronic technology .E-learning is a kind of selflearning flexible teaching activity through the Internet, and its integrated connotation is: using modern information technology means and through the effective integration of information technology and the curriculum to realize a kind of ideal learning environment and a new learning mode which can fully reflect the main role of the students, thus the traditional teaching structure and the nature will be completely reformed and the aim of cultivating a large number of talents (that is creative talents)with 21st century capacity will be achieved.

Keywords: e-learning, visual based e-learning, net meeting, bingo’s chat

1.

E- LEARNING

E-learning resources components: (1) School resources downloaded to the server. (2) Hyperlink Internet resources. (3) Courseware or integrated ware.

INTRODUCTION

With the vast development of various technologies, learning today is no long confined in classrooms with lecture as the only method conveying knowledge. E-learning, which facilitates education using a network, has made learning possible from anywhere at any time by using the Internet, wide area networks, or local area networks In visual based e-learning system, learners can take part in the discussion group and can discuss with other learners or instructor face-to-face (problem analysis, or creative problem solving). This kind of question-and-response learning activity could force learner’s independent thinking and active participation, just like the traditional teaching process Fig: 1 e-learning environment


17


E-Learning includes 1. Knowledge about human learning and information processing in general as well as one’s own learning processes. 2. Knowledge about the learning task at hand and its corresponding processing demand

E-Learning Environments (ELE) offer increasingly better technological infrastructures necessary for synchronous and asynchronous learning and teaching. These systems however, are often created with focus on authoring and course delivery.

3.

VISUAL BASED E-LEARNING (1) Imagery: People tend to focus on the images and graphics that appear on the screen first, and then they move on to other elements such as the text. However, a certain level of control can be gained over this tendency for it depends on the design and treatment of a page or the whole course itself.

Learning is geared to the needs and interests of the individual learner and is integrated into virtually all aspects of the individual’s work and life. Technology that supports e-learning makes it possible to customize and personalize content and delivery to match individuals’ learning styles, experience and skills. New means of assessing and certifying learning results replace traditional, clock-hour measures, providing secure and reliable systems for recording and capturing what an individual knows and is able to do. We need to

(2) Position: This is one of the most common ways to represent the progression. A good example of this would be the standard newspaper positioning wherein the hottest stories are located at the top of the paper because this position attracts more attention than those at the bottom.

1. Create the highest-quality e-learning experiences possible. 2. Implement new measures and methods for assessing and certifying what individuals know and are able to do. 3. Ensure broad and equitable access to elearning opportunities.

(3) Color: In design, bright colors used with contrast attract more attention, while darker ones don't. To establish hierarchy, it is important to make high ranking elements more vivid and less ranking ones darker.

There are 3 main steps: 1. Action Mapping 2. Organize your Materials

(4) Size: Size is probably the most compelling way to depict hierarchy on a screen. There are different ways by which you can utilize this effectively. For example, making the title of a page larger, or making a relevant video encompass most of the screen.

3. Write and Create the Content There are different ways to implement instructional design for an eLearning online course. Usually, designers opt to combine different kinds of approaches for maximum impact. Here are some different ways a person can establish visual hierarchy:


18


Fig:3 e-learning technology

4.

NET MEETING A net meeting is a way for people to get together and discuss issues without having to attend to many of the expensive and inconvenient arrangements that are required by personal meetings. Videoconferencing technology is becoming a helpful tool to businesses and individuals. With an Internet-connected computer, a microphone and a camera, people can hear and see each other as if they were face to face. Net Meeting offered as a free program for both home and business.

NetMeeting is conferencing software provided by Microsoft for users of Windows versions 95 through XP. While we can participate in an instant message conference without additional components, if we have a microphone and headset or speakers we can participate in an audio conference. Add a webcam and you can hold a video conference. These features make NetMeeting a beneficial application with many uses for the office, classroom, and home office.


19


4.1 FEATURES

4 MB of free hard disk space (an additional 10 MB is needed during installation only to accommodate the initial setup files). Sound card with microphone and speakers (required for audio support).

Video and Audio Conferencing Live Chat Internet Director File Transfer Program Sharing Remote Desktop Sharing Security Net Meeting Advanced Calling (Email & built in address book) Whiteboard

To use the data, audio, and video features of NetMeeting, your computer must meet the following hardware requirements: For Windows 95, Windows 98, or Windows Me, a Pentium 90 processor with 16 MB of RAM (a Pentium 133 processor or better with at least 16 MB of RAM is recommended). For Windows NT, a Pentium 90 processor with 24 MB of RAM (a Pentium 133 processor or better with at least 32 MB of RAM is recommended). 4 MB of free hard disk space (an additional 10 MB is needed during installation only to accommodate the initial setup files). 56,000 bps or faster modem, ISDN, or LAN connection. Sound card with microphone and speakers (sound card required for both audio and video support). Video capture card or camera that provides a Video for Windows capture driver (required for video support).

System Requirements listed by Microsoft The following are the minimum system requirements to install and run Microsoft NetMeeting. 90 megahertz (MHz) Pentium processor 16 megabytes (MB) of RAM for Microsoft Windows 95, Windows 98, Windows Me 24 megabytes (MB) of RAM for Microsoft Windows NT version 4.0 (Microsoft Windows NT 4.0 Service Pack 3 or later is required to enable sharing programs on Windows NT.) Microsoft Internet Explorer version 4.01 or later 28,800 bps or faster modem, integrated services digital network (ISDN), or local area network (LAN) connection (a fast Internet connection works best).


20


5. BINGO CHAT With this program you will be able to communicate in the local network and the Internet, organize a voice conference with other users in the local network. The program has a convenient interface and beautiful design, which can change depending on your taste.

Bingo's Chat - free voice & text chat, IRC client for Local Area Networks. It works without server. It is an ideal solution for small networks. Has support for animated smilies, avatars, files and folders transferring, insertion images into chat window (even from clipboard), formatting text, hot phrases, notice-board, encryption of messages.

5.1 FEATURES

Notice-Board. Showing active LAN servers of Quake III Arena, Counter Strike 1.6 / HL in chat (hot keys: F5, F6). Hot phrases. Support bots and other add-ons. Export / Import of all settings Strong encryption of messages.

Audio compression to MP3 format - to save traffic. The level of the beginning of sound transmission, voice amplifier, recording of conferences. Support for avatars, including animated (JPEG, PNG, GIF). Insertion images into the chat window (even from the clipboard), autoscaling. Visual style, choice of skin. Personal HTML pages with information about the user. Formatting text (colored letters with the background color, bold, underline). 900 animated smilies Support for multiple languages. Ignoring users & messages. Transfering files and folders of any size to user or into channel by dragging and dropping files or folder to users in the list with resume support.


6.

COMPARATIVE STUDY

NET MEETING: Net Meeting is a product of Windows XP. So it can be used only in Microsoft Windows XP. It does not provide security. BINGO’S CHAT: Bingo’s Chat works without server. It is an ideal solution for small networks. It provides security to user by encrypting messages

21


7.

RESULTS

Graphical representation for Audio Visual E-Learning

70

6 5 4

60 50

DATA

3 2 1 0

40

VOICE

30

VIDEO

20

Net Meeting Bingos Chat

10 0

CLIENT1 CLIENT2 CLIENT3 CLIENT4

Data Voice Video

Net Meeting Vs Bingo’s Chat

8.

Compare to Microsoft Net meeting. Bingo’s Chat Provides security to users and will be a ideal solution for small networks.

CONCLUSIONS

E-learning allows learning to become a continuous process of inquiry and improvement that keeps pace with the speed of change in business and society. With elearning, the learner has convenient, just-intime access to needed knowledge and information, with small content objects assembled and delivered according to the learner’s specific needs.

9. 1.

2.

E-learning is driven by market forces, including individual decision-making and consumer choice, rather than by institutional interests. Using the Bingo Chat program we will be able to communicate in the local network and the Internet, organize a voice conference with other users in the local network. The program has a convenient interface and beautiful design, which can change depending on our taste.

3.

4. 5. 6.

If we are an educator, we can use NetMeeting to include students who are not in the classroom. The software is an ideal tool for children who are ill or too contagious to be present at school, but well enough to do the work from home. NetMeeting is also an excellent program to use as part of an online course for adult education. Students can use the application to work together on projects from separate locations or as a group study too.


7. 8.

22

REFERENCES

Tavangarian D., Leypold M., Nölting K., Röser M.,(2004). Is e-learning the Solution for Individual Learning? Journal of e-learning, 2004. EC (2000). Communication from the Commission: E-Learning – Designing "Tejas at Niit" tomorrow’s education. Brussels: European Commission http://joshuamaher.com/2007/02/21/netme eting-on-vista What Are the Benefits of NetMeeting?eHow.com http://www.ehow.com/list_6165502_bene fitsnetmeeting_.html#ixzz1DyoqzPZD Karrer, T (2006) What is eLearning 2.0? Elearningtech.blogspot.com Crane, Beverley E. "Using Web 2.0 Tools in the k-12 Classroom" Neal-Shuman Publishers Inc., 2009, p.3 http://EzineArticles.com/?expert=Don_Ro berteLearning Papers www.elearningpapers.eu


Study and Analysis of Visual e-learning – Empowering Visually Challenged students R.Sangeetha1 S. Praveena2 R.Karthiya Banu3 Saravanan4 1. II MCA, Loyola Institute of Technology, Chennai, Tamilnadu [email protected] 2. II MCA, Loyola Institute of Technology, Chennai, Tamilnadu [email protected] 3.,Asst Prof & Head, Department of Computer Applications, Loyola Institute of Technology, Chennai, Tamil Nadu [email protected] m 4. Research Scholar, Satyabama university, Chennai.

Abstract

environments, video, and animation. e-Learning, in

Existing methods of learning and training do not make provision for certain category of e people such as who are visually impaired. Many studies have demonstrated the value of several learning platforms, including mobile learning (m-Learning) but with the problems of access barriers and streamlined participation of most learners. The objective of this paper is to provide the Voice based e-learning relevance for the visually and mobility impaired learners. Voice-based learning system is a computer system which uses voice input and voice output in an educational environment. Such systems promise two-way communication with the computer using natural language. In addition to reading and writing skills, the computer can now teach listening and speaking skills Speech recognition is the process of converting an acoustic signal, captured by a microphone or a telephone, to a set of words. The recognized words can be the final results, as for applications such as commands & control, data entry, and document preparation. They can also serve as the input to further linguistic processing . Our proposed model gives clear idea about that the students can listen to the classes at home or anywhere through the mobile. All the class room teaching are stored in the server which can be retrieved anywhere at any time.

some ways, is even better than classroom learning

Keywords : e-learning, VUI, IVR, Speech

the student in oral language. While a computer does

Recognition

not possess the intelligence to interact fully as a

methods as it is a oneon-one learning method, it is self-paced and it has an experiential-learning format. e-Learning environment

puts

the

where

user objects

in

an

in

interactive

the

learning

environment can be readily adjusted, modified or manipulated to accord with the user’s preference. This

enhances

the

learning

process.

Visual

Impairments can mean a number of things.. A person who is totally blind can not see light or anything else. Some people use different things to help with their visual impairments by using adaptions such as glasses, Braille, seeing eye dogs, canes, and adaptive computer technology. There are many devices such as screen readers, computers and many other inventions that were made and still being made to help people that are visually impaired. There are many inventions for computers that make other technological devices then usable to the Visually Impaired - and can change their lives forever. This paper discussed about Voice based elearning which includes a computer, a voice input terminal and a voice output device. It interacts with

human would, it can communicate.

1.INTRODUCTION 2.E- LEARNING E-learning involves the use of a computer or

e-Learning can be done using an internet connection, a network, an intranet, or a storage disk. It uses a variety

of

media


like

audio,

text,

electronic device (e.g. a mobile phone) in some way

virtual

to provide training, educational or learning material.

23


(Derek Stockley 2003) E-learning can involve a

VUI makes human interaction with computers

greater variety of equipment than online training or

possible through a voice/speech platform in order to

education, for as the name implies, "online" involves

initiate an automated service or process. The VUI is

using the Internet or an Intranet. CD-ROM and DVD

the interface to any speech application. Controlling a

can be used to provide learning materials. Distance

machine by simply talking to it was science fiction

education

e-learning's

only a short time ago. Until recently, this area was

development. E-learning can be "on demand". It

considered to be artificial intelligence. However,

overcomes timing, attendance and travel difficulties.

with advances in technology, VUIs have become

provided

the

base

for

more commonplace, and people are taking advantage of the value that these hands-free, eyes-free

MOBILE LEARNING

interfaces provide in many situations. Even with advanced

development

tools,

constructing

an

effective VUI requires an in-depth understanding of both the tasks to be performed, as well as the target audience that will use the final system. The closer the VUI matches the user's mental model of the task, the easier it will be to use with little or no training, resulting in both higher efficiency and higher user satisfaction. The right process starts with getting all the information up front. Then it's analyzed, twisted, turned, ripped open, analyzed again, turned upside down, mashed up and analyzed yet again. This leads to a sound design for an automated call solution

3. E-Learning Program: Success Factors

when UniversalGuidelines, Project Guidelines, and

Ultimately, companies must determine the best

Interaction Guidelines are incorporated into the

delivery modality to facilitate knowledge transfer

overall project implementation process.

quickly, completely and affordably. The experience, locations and availability of learners represent key variables to determine whether e-learning or handson learning is a better fit for an organization.

Students matched to appropriate training modality

5.SPEECH RECOGNITION The most basic of speech recognition technologies is automatic speech recognition (ASR) which is the capability to automatically recognize human speech based on a word-by-word [Cox et al., 2000, pp. 1319]. The speech recognition component is aimed at

4.VOICE USER INTERFACE (VUI) Sathyabama University

24


capturing user’s spoken input and transforming it

read websites, documents, and email messages to you.

into the most probable word sequence using the given acoustic (feature) and language (structural)

7. SPEECH ENGINE

models. The following sections explain speech The LumenVox Speech Engine is a real-time, speaker-independent technology. Real-time means it can perform recognition on an audio stream immediately, and speaker independent means it does not require any special training to recognize a user's voice. Most users of the Speech Engine use it in telephony applications, such as an IVR, but it can be used for any computerized command and control application, such as a voice dialer, order taker, or even to control simple actions on a PC. The Speech Engine is not dictation software, and it does not transcribe large audio files. It is used for applications where users speak short, discrete phrases. LumenVox is also working on support for hot word spotting and audio mining, the ability to find specific words in large collections of audio.

recognition processes, the properties of speech recognition systems, and the performance of speech recognition systems respectively A comprehensive Voice and Speech Recognition program to use your Voice for command & control of your computer and dictation. Reduce or eliminate mouse clicks or keyboard input. Open Web sites, documents, or programs using your Voice. Perform navigation and editing functions simply by speaking. Dictate letters, memos, and email messages. Begin talking to your PC now using your Voice.

6. EXISTING SYSTEM

8. DISTRIBUTED ARCHITECTURE These benefits and to explain specifically how this architecture works in a real world environment with the LumenVox speech recognition software, let's discuss a hosted IP-PBX solution from Ontelnet. Ontelnet, a full service provider of communication services for residential and business customers, decided to add speech recognition functionality to its offerings. Because of high call volume, they knew they would need to deploy a distributed architecture. Ontelnet's Distributed Architecture.

Ontelnet's Distributed Architecture In addition, Ontelnet not only needed multiple Speech Servers, but also wanted to have multiple servers to accept incoming SIP trunks and host the company's applications. When speech recognition resources are required, each of the Ontelnet servers could communicate with the Speech Recognition Servers to enlist speech resources.

e-Speaking Voice features a rich set of predefined Voice commands and is based on the latest Speech technologies from Microsoft. Import hundreds of commands for Microsoft Office and other Windows programs and games. Create your own commands for the software that you use. The program includes synthesized Speech prompts and audio feedback. A variety of graphic application skins and options let you customize your e-Speaking Voice experience. Contains an animated avatar so you can see the computer speaking to you.

9.PROPOSED MODEL Our proposed model discusses about how effectively voice based elearning utilized for visually impaired students as well as helpful for the student who missed out their regular class due to unavoidable circumstances

e-Speaking Voice includes integrated text-to-Speech capabilities enabling the computer to have a Voice to


25


user

functions Purpose

user

Request Telephone Dial

Output

Input Handling

Database Response

Output Error handling

The above model given clear idea about they can listen to the classes at home or anywhere through the mobile. All the class room teaching are stored in the server which can be retrieved anywhere at any time.

Lookup To look up a definition from a database. A spoken word The target system accepts speech input from the user and then checks whether it matches the speech recognizer . If valid, the word or phrase query will request the responsive definitions from the database. Textual definitions of the word or phrase query. If the speech input does not match the speech recognizer , the target system guides the users to speak the query word clearly and loudly or, otherwise, speak another query word.

11.CONCLUSION e-Learning is learning that takes place in an electronically simulated environment. e-Learning, webbased training, internet-based training and computerbased training are the next-generation instruction methods being developed today. With eLearning, users can immerse themselves in a threedimensional environment to further enhance their learning experience. Moreover, e-Learning can be done anywhere and anytime as long as the user has the proper hardware.

10.WORKING METHODOLOGY The working of our proposed model is as follows: Process of information retrieval 1 System: Welcome to the dial lookup system. 2 Say a topic you want to look up. 3 User: operating system 4 System: You look up the definitions of operating system. Is it correct?

12.REFERENCES

5 User: Yes 6 System: The definitions are as follows.

1.

Operating system

2.

1. Definition detail 2. Definition detail 3. Definition detail

3.

Please select one of: 1. repeat the definitions 2. new lookup

4.

3. exit

5.

7 User: Three

6..

8 System: Ok then, good bye.

7.


26

The value of e-learning, white paper, David Leaser, IBM Lotus Education J. Markowitz, ‘Voice Authentication in the RealWorld”, Keynote Speech, National Centre for Biometric Studies Conference on Voice Authentication for Identity Management, 2006. Retrieved 14th September 2009 from http://www.voicexml.org/insights/JMarkowitz Consulting/jmc_voice_auth_2006.pdf P. Gallivan, Q. Hong, L. Jordan, E. Li,G. Mathew,Y. Mulyani, P. Visokey and C. Tappert,Voice XML Absentee System, Proceedings of MASPLAS, 2002. The Mid-Atlantic Student Workshop on Programming Languages and Systems Pace University. http://www.rocketdownload.com/program/ espeaking-voiceand-speech-recognition-20339.html http://www.lumenvox.com/resources/ whitePapers/distributedArchitecture.aspx http://www.developer.com/voice/article. php/1567051/Top10-Best-Practices-for-Voice-User-Interface-Design.htm http://elba.szs.unisarlsruhe. de/index.php?site=elba/index&language=EN


A Common Data mining Framework for Different ERP Models Prof. M.K. Gandhi M.Tech [IT] 1, Dr. K. Sarukesi2 1

Department of Fashion & Textiles, National Institute of Fashion Technology, NIFT Campus Rajiv Gandhi Road Taramani Chennai 600 113, India Email: [email protected] 2 Vice Chancellor, Hindustan University, Chennai, India Email: [email protected] mining approach, called "Integration of different Abstract: As Enterprise Resource Planning (ERP) ERP models”, that offers the capability of uniform implementation has become more popular and suitable for manipulation of data and meta-data in relational every business organization, it has become an essential factor for the success of a business. In this paper we have multi-database systems. We develop a precise syntax presented a new framework for the different ERP models in and semantics of queries in a manner that extends an effective way. Apart from this the Data Mining traditional SQL syntax and semantics to demonstrate approach is used for the integration of these ERP models by [5]. giving support for making the successful result. The Also it retains the flavor of SQL while proposed framework integrates data, especially from supporting querying both data and meta-data. It can heterogeneous sources, which is a hard and widely studied be used to define and create "restructuring views", problem. Here the particularly challenging issue is the views that represent data in a database in a structure integration of sources that are semantically equivalent but substantially different from original database, in schematically heterogeneous. While two such data sources may represent the same information, one may store the which data and meta-data may be interchanged [7]. information inside tuples (data/rows), while the other may Data mining approach provides a great facility for store it in attributes or relation names (schema). The Data interoperability and data/meta-data management in Mining approach which forms as an Interface to get a relational multi-database systems. We provide many solution to this problem and it is powerful enough to examples to illustrate our claims. restructure such sources into each other without any loss of It is now widely accepted that ERP systems information. Also it is able to manage any type of databases provide a viable alternative to custom application such as Microsoft Access, Oracle or even MS-SQL. Hence, development for the standard information the entire ERP database Models are in various platforms management needs and that it is often superior in namely, MS-Access, SQL and Oracle can be handled with a single Data mining query and since it is considered as the terms of quality of the implemented business process best querying language. [2]. Although current common ERP systems provide enough parameters and can be adjusted according to industry characteristics, they is still too complicated Keywords—Enterprise Resource Planning, Data to use only effective under certain conditions [6]. Mining, Querying, heterogeneous, schema ERP solutions are designed to solve the fragmentation of information in businesses, and integrate all the information flowing within a I. INTRODUCTION company is designed to solve the fragmentation of As an electronic business environment changes more information in businesses, and integrate all the rapidly under the globalization, even small and information flowing within a company [12]. medium size companies also change their business. With enterprises becoming bigger and bigger, the legacy business systems may not be flexible enough II. BACKGROUND to adapt this change and the discordance between A. Enterprise Resource Planning business and information systems in their organization may occur [4]. Therefore, recently most Enterprise resource planning integrates the companies use an ERP system for improving core informational management systems of production, competency. ERP integrates the functionality of all finance, accounting or human resources departments the business departments in an organization in a to achieve real time management. ERP also provides single system to carry out the particular needs of control and management systems for enterprises, these different departments and share their which can provide the management with immediately information very easily. Apart from this a single complied internal integral data as reference for organization that contains large number of database designed, created and maintained by number of users decision-makings. Therefore the major functions of on different location; these databases can be on control and management systems of the enterprise different operating system, different platform and cover executive information system, business different data models. There should be some planning and budgeting and profit center accounting mechanism for interoperability among databases, for [1, 3]. It can be defined as an Efficient, integrated this we provide a principled extension of a Data Sathyabama University

32


information systems are very important for companies to be competitive. Also it can help integrate a company’s operations like Acting as a company-wide computing environment. Also it includes a database that is shared by all functional areas. Apart from this it can also deliver consistent data across all business functions in real time. The various ERP models are of in various platforms namely, MS-Access, SQL and Oracle can be handled with a single Data mining query and since it is considered as the best querying language. B. Data Warehouse There are updated (structural and nonstructural) data that shows exponential growth, appearing within the enterprise daily. How to process and further analyze this updated data would be the present topic for investigation among the enterprises. The data from the data mart specializes in collecting data that meets the requirement of the process in decision-making support system (DSS) of a definite department. It is the sub set of data warehouse, which especially designed to meet the requirement of certain department. The departments that use data mart usually cover the marketing, finance, accounting, engineering and auditing sectors. Therefore, many of the data marts are under the data Warehouse system [2].

Fig.1 Proposed Framework

A. Database Selection This module is used to select two various database servers. Here we use three databases like MS-Access, Oracle and SQL Server. In this module the user select the database for access and SQL server, select the table for oracle. Each of these we have uses the provider name which is used to avoid the dsn (data source name) creation. For example the provide of access is “Provider = Microsoft.Jet.OLEDB.3.51”.

C. Data mining Data Mining is a tactical process that uses mathematical algorithms to sift through large datastores to extract data patterns/models/rules. Also the Knowledge Discovery is the process of identifying and understanding potentially useful hidden anomalies, trends and patterns. Data mining is an integral part of knowledge discovery process. Data mining, the extraction of hidden predictive information from large database, is a powerful new technology with great potential to help companies focus on the most important information in their data warehouse. The implementation of DM functionality in a DBMS, are boundless to a specific database engine and are quite flexible in a typical enterprise application landscape - heterogeneous environment [6].

B. Machine Selection This module has two forms. 1. Domain Selection In this form user choose the domain like main, workgroup. This combo box list out all domain names and user choose any one domain. 2. Machine Selection In this form the user chooses the machine name like machine1, machine2, etc. This module is automatically display the machine name with out changing the codes. C.Query Builder This module is used to build the query like create, update, delete, select, insert. Here we have used three buttons. The first two buttons are used to display the result for selected databases individually. The last button names “mixed” is used to combine the two tables from various databases then display the result. This is create a new table and stored the combined results in to this table. All the results are displayed in the MS – flexgrid. After displaying the results the new table is deleted because the users don’t need this table. We write the query on the text box. And this form is also flexible to create the query,

III. ERP AND DATA MINING SYSTEM The presented model shown in the Fig.1, described the abstract of all concerned exist in the overall framework of the proposed approach. Also it clearly described the ERP model to solve the business problems. With ERP System Meta Data Architecture Method utilize ERP that provide language pattern to accomplish the operation. Sathyabama University

33


system environment which is of different ERP models.

because the user may be forget some command used in query like sort, group by etc., So here the user select sort option, group by and so on.

3.3 Syntax The main goal is to develop a PSQL (Proposed SQL) as a principled extension of SQL. To this end, we briefly analyze the syntax of SQL, and then develop the syntax of PSQL as a natural extension. Our discussion below is itself a novel way of viewing the syntax and semantics of SQL, which, in our opinion, helps a better understanding of SQL subtleties. In an SQL query, the (tuple) variables are declared in the from clause. A variable declaration has the form . For example, (a) .select T. name from employees T where T. department =“Marketing” (b) select employees. name where employees employees.department =“Marketing” (c).select name from from employees where department =“Marketing” The expression employees T declares T as a variable that ranges over the (tuples of the) relation employees (in SQL, these variables are called aliases.) The select and where clauses refer to (the extension of) attributes, where an attribute is denoted as . , var being a (tuple) variable declared in the from clause, and attName being the name of an attribute of the relation (extension) over which var ranges. When no ambiguity arises, SQL permits certain abbreviations. The PSQL syntax extends that of SQL in several directions. The group consists of databases, with each database containing relations. The syntax allows to distinguish between (the components of) different databases. To permit meta-data queries and restructuring views, PSQL permits the declaration of other types of variables in addition to the (tuple) variables permitted in SQL.

3.1 Key Features Some of the key features required of a language for interoperability in a (relational) database are the following: (1) The language must have an expressive power that is independent the schema with which a database is structured. For instance, in most conventional relational languages, some queries (e.g., “find all department names”) expressible against the database univ-A are no longer expressible when the information is reorganized according to the schema of, say, univ-B there. This is undesirable and should be avoided. (2) To promote interoperability, the language must permit the restructuring of one database to conform to the schema of another (3) The language must be easy to use and yet sufficiently expressive. (4) The language must provide the full data manipulation and view definition capabilities, and must be downward compatible with SQL, in the sense that it must be compatible with SQL syntax and semantics. We impose this requirement in view of the importance and popularity of SQL in the database world. (5) Finally, the language must admit effective and efficient implementation. 3.2 Contribution (1) We propose a language interface in Data mining which meets the above criteria, review the syntax and semantics of SQL, and develop Schema SQL as a principled extension of SQL. As a result, for a SQL user, adapting to Schema SQL is relatively easy. (2) We illustrate via examples the following powerful features of Schema SQL: (i) Uniform manipulation of data and metadata; (ii) Creating restructured views and the ability to dynamically create output schemas; (iii) The ability to express sophisticated aggregate computations far beyond those expressible in conventional languages like SQL. (3) We propose implementation architecture for Schema SQL that is designed to build on existing RDBMS technology, and requires minimal additions to it, while greatly enhancing its power. We provide an implementation algorithm for Schema SQL, and establish its correctness. We, also discuss novel query optimization issues that arise in the context. (4) Finally, we propose an extension to SQL for systematically resolving the semantic heterogeneity problem arising in a multi database Sathyabama University

IV. RESULTS AND DISCUSSION For the practical implementation of presented model we applied data mining technique on different ERP model database data’s. PSQL permits the declaration of variables that can range over any of the following five sets: (i) Names of databases in a group; (ii) Names of the relations in a database; (iii) Names of the attributes in the scheme of a relation; (iv) Tuples in a given relation in a database; (v) Values appearing in a column corresponding to a given attribute in a relation. Variable Declarations follow the same syntax as in SQL, where var is any identifier. However, there are two major differences. (1) The only kind of range permitted in SQL is a set of tuples in some relation in the database, whereas in SCHEMA SQL any of the five kinds of ranges above can be used to declare variables. (2) More importantly, the range specification in SQL is made using a constant, i.e. an identifier 34


referring to a specific relation in a database. By contrast, the diversity of ranges possible in SCHEMA SQL permits range specifications to be nested, in the sense that it is possible to say, e.g., that X is a variable ranging over the relation names in a database D, and that T is a tuples in the relation denoted by X.

Example: List the departments in univ-A that pay a higher salary floor to their technicians compared with the same department in univ-B. select A.dept from univ-A::salInfo A, univ-B::salInfo B, univ-b:;sal(Info}->AttB where AttB category” and A. dept = AttB and A.category = “technician” and B.category = “technician” and A.salFloor > B.AttB Explanation: Variables A and B are (SQL-like) tupie variables ranging over the relations univ-A : : salInf o and univ-B : : salInf o, respectively. The variable AttB is declared as an attribute name of the relation univ-B: : salInfo. It is intended to be a dept attribute, (hence the condition AttB “category”in the where clause).

Here we illustrate via examples the many powerful features of PSQL. The following group of databases is used as our running example. Consider the group consisting of four databases, univ-A, univ-B, univC,and univ-D. Each database has (one or more) relation(s) that record(s) the salary floors for employees by their categories and their departments, as follows: univ-A has a relation salInfo (category,dept , salFloor). univ-B has a relation salInfo (category,deptl, dept2, . . . >. Note that the domains of deptl, dept2, . . . are the same as the domain of salFloor in univ-A: : salInf o. univ-C has one relation for each department with the scheme depti(category, salFloor). univ-D has a relation salInfo (dept, catl, cat2, . . . ). Note that the domains of cat 1,cat2, . . . are the same as the domain of salFloor in univ-A: : salInf o.

4.1 Aggregation with Fixed Output Schema In SQL, we are restiicted to Vertical” (or columnwise) aggregation on a predetermined set of columns, while SCHEMA SQL allows “horizontal” (or row-wise) aggregation, and also aggregation over more general “blocks” of information. We illustrate these points with examples. The query select T.category, avg(T.D) from univ-B::salInfo-> D, univ-B : : salInfo T

Salary information:Univ A Category

dept

where D “category”

salfloor

Prof

cse

25000

Asst

It

20000

Lecturer

cse

10000

group by T.category computes the average salary floor of each category of employees over all departments in univ-B. This captures horizontal aggregation. The condition D “category” enforces the variable D to range over department names. Hence knowledge of department names (and even the number of departments) is not required to express this query. From the results shown, we can know that the ERP data searching system provided can accurately locate the actual problems of enterprises and the resolution made accurately resolve the problems of the enterprises. Here we conclude that the proposed common Data mining framework for different ERP models has overcome all the existing Data mining problems immediately.

Salary information:Univ B Category

cse

maths

Prof

25000

22000

Asst

20000

15000

Lecturer

10000

9000

Salary information:Univ C Dept

prof

V. CONCLUSIONS In today’s technologies the organizations have a lot of difficulties to access the information’s in different organization’s facilities. To solve these problems a new approach has been proposed. The main goal of this research orientation is of investigation of business and ERP System with Data Mining System architecture brings Data Mining approach as integration of the Heterogeneous databases all together into one domain. Apart from this it also provide a good solution to handle schematic (structural) heterogeneity, that is, similar

Asst proff

cse

25000

15000

maths

22000

15000

Salary information:Univ D Cat:CSE Category

salfloor

Prof

25000

Asst prof

20000


35


information represented in different structures that are been integrated by a single Query.

[1]

[2]

[3]

[4]

[5]

[6] Ahmed, R., Smedt, P., Du, W., Kent, W.,Ketabchi, A., and Litwin, W. The Pegasus heterogeneous multidatabase system. IEEE.

REFERENCES First A. Abdullah S. Al-Mudimigh, Second B. Zahid Ullah, Third C. Farrukh Saleem “A Framework of an Automated Data Mining System Using ERP Model”, International Journal Of Computer And Electrical Engineering, Vol. 1, No. 5 December, 2009, Pp. 1793-8163 Wen-Hsiung Wu, Chin-Fu Ho, Hsin-Pin Fu, Tien-Hsiang Chang, “Smes Implementing An Industry Specific ERP Model Using A Case Study Approach”, Journal Of The Chinese Institute Of Industrial Engineers, Vol. 23, No. 5, 2006, Pp. 423-434. Arash Ghorbannia Delavar, Nasim Anisi , Majid Feizollahi “A New Framework For The Work Flows Of Distributed Integrated Systems By Assessment Of Effective Factors” , 978-1-61284840-2/11/$26.00 ©2011 IEEE Abdullah S. Al-Mudimigh, Farrukh Saleem, Zahid Ullah, “The Effects Of Data Mining In ERP-CRM Model –A Case Study Of MADAR” , Issn: 1109-2750 , Issue 5, Volume 8, May 2009 Ruey-Shun Chen, C. C. Chen*, C. C. Chang and M.H. Wu, “A Web-based Data Mining System for ERP Decision Making”, IEEE 2002.


[7]. Lefebvre, A., Bernus, P., and Topor, R. “Query transformation for accessing heterogeneous databases”. In Workshop on Deductive Databases in conjunction with JICSLP. [8]. Lakshmanan, L.V.S., Sadri, F., and Subramanian,I. N. “SCHEMA SQL - a language for querying and restructuring multi database systems”. Technical report, Concordia University, [9]. Krishnamurthy, R., Litwin, W., and Kent, W. “Language features for interoperability of databases with schematic iscrepancies”. [10] Gray, J., Bosworth, A., Layman, A., and Pirahesh H. “Data Cube: A relational aggregation operator generalizing group-bi, cross-tab, and sub-totals”. [11] Missier, P. and Rusinkiewicz, Marek. “Extending a multi database manipulation language to resolve schema and data conflicts”.

36


A Framework for event matching in temporal database 1

N. Duraimutharasan 1, Dr. K. Sarukesi2

Department of Computer Science & Applications, Periyar Maniammai Univeristy Vallam, Thanjavur, India Email: [email protected] 2 Vice Chancellor, Hindustan University, Chennai, India Email: [email protected] Abstract: Time is one of the most difficult aspects to handle A relational database [10], [19] is a set of in real world applications such as database systems. tables called relations. Each relation contains one or Relational database management systems proposed by Code more data categories in columns. Each row contains a offer very little built-in query language support for temporal unique instance of data for the categories defined by data management. The model itself incorporates neither the the columns. For example, a typical employee concept of time nor any theory of temporal semantics. Many temporal extensions of database would include a table that described an the relational model have been proposed and some of them employee with columns for name, address, phone are also implemented. Also a database that can store and number, and so forth. Another table would describe a retrieve temporal data, i.e. data which depends on time in department: number, name, head, location, and so some way, is termed as a Temporal Database. Moreover, we forth. A user of the database could obtain a view of can implement this to manage temporal information and the database that fits in the user's needs. For example, support the storage and querying of information that varies a human resource manger might need a report about over time. In the present scenario, writing better database employees who had availed their total leaves on a queries for databases pertaining to an organization involves certain date. A financial services manager in the same a significant amount of time and expertise. For users who feel SQL difficult to use and for novice users who would company could, from the same tables, obtain a report like to retrieve the data’s without having to learn querying on accounts that needed to be paid. Storage of data and the retrieval of stored data in mechanism such as SQL, a temporal Natural Language querying mechanism has been provided to access these the required format form the major activity in the data’s from temporal databases. database applications. Such Database information system provide user interface in a number of way Keywords— Temporal Database, SQL, Natural including query language interfaces. Moreover, users Language querying, expertise, Time vary from well-trained experts to novice users [4]. Hence, to simplify the querying, a natural language I. INTRODUCTION interface to database system is an essential component. A database that can store and retrieve temporal In the proposed system, in order to enable a novice data, that is, data which depends on time in some way, user to interact with the temporal database system and is termed as a Temporal Database [1].The classical simplify the query processing in the temporal database database is generally two dimensional, and contains system, a Temporal Natural Language interface (TNLI) only current data. The two dimensions are rows and is needed object evolution in temporal this paper columns that interact with each other at cells initially describes the background framework that containing particular value. Whereas temporal forms the basis of the proposed system. The detailed databases are three-dimensional with time interval as methodology of the proposed system, the workflow, the third dimension. Temporal Databases can also be architecture and the phases involved in the referred to as time-oriented databases, time-varying implementation are presented. The final section of this databases, or historical databases [1]. A true temporal report includes the current phase of the project that is database is a bi-temporal database that supports both being developed [6]. valid time and transaction time. Transaction time is The remainder of this paper is organized as follows: the actual time recorded in the database at which the After introducing preliminary concepts in Section 1 data is entered and the time is known as the Time- describes about the introduction of the temporal stamp. Time-stamps can include either only the date or database, Section 2 provides an overview of related both the date and clock time. Time-stamps cannot be works done Section 3 explains about the temporal changed. The other major type in temporal database is Query Processing model. In Section 4, the design of the valid time. Valid time is the actual or real world architecture to support Query processing has been time at which point the data is valid. Temporal described. Analysis of the system is presented with the databases mainly use Natural language processing to aid of Flow Chart developed during this Section 5. give the answers for users question. For that users Gives a overall results and discussion of the proposed must give the information or passage. They can work and Finally, Section 6 provides a concluding determine any type of questions related to that Query. remarks. It gives the corresponding output for the given II. RELATED WORKS questions. This system helps the users to interact with temporal databases by giving a natural language query The Query Analyzer module in NLP takes in the [2]. key words as input and separates the fields that have to be retrieved. The main intention is to get the


37


meaning of the sentence. Initially, the given input is scanned for domain specific phrases. For example, customer identification number is identified by id in the database. So, the transformation becomes important. Similarly, the temporal operators and relational operators equal to, greater than etc., are replaced by appropriate symbols. From the transformed list of words, the relational operators and other keywords such as ‘between’, ‘in’, ‘like’ etc., are located as these specifies the required condition. The conditions are separated out and have to be transformed to fit into the SQL query directly. In this system, the keywords are the ones the represent the field name in the database and also the table name it is explicitly specified. The keywords are of type noun. The words that are noun alone are taken out. This will contain the field name and the table name. A separate table is maintained where we have the list of fields and the respective table name and the kind of the data that represent whether it is a string, number or data is maintained. The enter field names point to one table. A key identifier helps to identify every table. Even if the table name is not directly specified, the table name can be obtained from the key identifier. The SQL query is formed with the field name; table name and the condition are separated out .A database that can store and retrieve temporal data, that is, data that depends on time in some way, is termed as a Temporal Database. The classical database is generally two-dimensional, and contains only current data. The two dimensions are rows and columns that with each other at cells containing particular value. Where as temporal databases are three–dimensional with time interval as the third dimension. Temporal Databases can also be referred to as time-oriented databases, time-varying database, or historical databases. A true temporal database is a bi-temporal database that support both valid time and transaction time. Transaction time is the actual time recorded in the database at which the data is entered and the time is known as Time-stamp. Time-stamp cannot be changed. The other major type in the temporal database is the valid time. Valid time is the actual world time at which point the data is valid. System have been forced to manage temporal information in an ad-hoc manner and support storing and querying of the information that various over time III. TEMPORAL & QUERY ANALYSER MODEL Time is evolutionary in nature and objects evolve in the real world with respect to change of time. A temporal element is a subset of the universe of time T. A temporal element can be expressed as a finite union of intervals. In order to query object evolution, a Temporal Event Matching Language (TEML) can be used as a standalone language or it can be interfaced with query languages for querying temporal data. The purpose of a TEML is to find the occurrence of an event in the evolution. In this language, an event is expressed as a pattern called Event expression. The language borrows the concept of cursor from SNOBOL 4 [8]. A cursor is simply an instant of time. Sathyabama University

38

As the cursor moves from one instant to another, the changes in the evolution are observed. In matching an event, one examines the evolution instant wise through the cursor. The set of instants where the matching takes place is accumulated and the accumulated set of instants are the outcome of the matching process. The time instant at which the cursor points to is called the Cursor Time. The default movement is in the increasing value of time that is one instant at a time. Setting modes can alter the nature of cursor movement. Like, Cursor Forward sets cursor to move in an increasing value of time and the mode Cursor Backward sets cursor to move in decreasing value of time. The Query Analyzer module in NLP takes in the key words as input and separates the fields that have to be retrieved. The main intention is to get the meaning of the sentence. Initially, the given input is scanned for domain specific phrases. For example, customer identification number is identified by id in the database. So, the transformation becomes important. Similarly, the temporal operators and relational operators equal to, greater than etc., are replaced by appropriate symbols. From the transformed list of words, the relational operators and other keywords such as ‘between’, ‘in’, ‘like’ etc., are located as these specifies the required condition. The conditions are separated out and have to be transformed to fit into the SQL query directly.

Fig. 1 Proposed Architecture In this proposed system, the keywords are the ones the represent the field name in the database and also the table name it is explicitly specified. The keywords are of type noun. The words that are noun alone are taken out. This will contain the field name and the table name. A separate table is maintained where we have the list of fields and the respective table name and the kind of the data that represent whether it is a string, number or data is maintained. The enter field names point to one table. A key identifier helps to identify every table. Even if the table name is not directly specified, the table name can be obtained from the key identifier. The SQL query is formed with the field name; table name and the condition are separated out .A database that can store and retrieve temporal data, that is, data that depends on time in some way, is


termed as a Temporal Database. The classical database is generally two-dimensional, and contains only current data. The two dimensions are rows and columns that with each other at cells containing particular value. Where as temporal databases are three–dimensional with time interval as the third dimension. Temporal Databases can also be referred to as time-oriented databases, time-varying database, or historical databases. A true temporal database is a bi-temporal database that support both valid time and transaction time. Transaction time is the actual time recorded in the database at which the data is entered and the time is known as Time-stamp. Time-stamp cannot be changed. The other major type in the temporal database is the valid time. Valid time is the actual world time at which point the data is valid. System have been forced to manage temporal information in an ad-hoc manner and support storing and querying of the information that various over time IV. QUERY PROCESSING The proposed system designs temporal database systems that increase the service capability of the database system to help novice users to formulate a query for database access. To help non expert users to perform query, a natural language front-end is required. For those users who feel SQL difficult to use and for novice users who would like to retrieve data without having to learn querying mechanism such as SQL, a template natural language querying mechanism has been provided to access data from temporal databases. It is designing as an event matching temporal database system comprises of major phases that comprises of message processing, query processing, event matching.

4.1 READ INPUT FILES - Read the Document which is given as the input. - Read the Full Image document. - Read the Image Tag and also the Image Alter Tag and Meta Tag to get the information. 4.2 ELIMINATE TAGS - The combination of information derived from text passages with information derived from the images. - Text surrounding an image is expected to be related to the contents of the image. - Unwanted Tags are generally eliminated from the documents. - HTML tags leads to improved retrieval of image. 4.3OUTPUT EXTRACTION In this module the Data’s read from the database, Questions which are given compared with the normal SQL Queries and the output which will be generated for the user according to their Messages. - Reading words from the database - Comparing words with question array - Make the outputs and produce to user The goal of the Natural Language Processing (NLP) group is to design and build software that will analyze, understand, and generate languages that humans use naturally, so that eventually you will be able to address your computer as though you were addressing another person. Here the goal is not easy to reach. "Understanding" language means, among other things, knowing what concepts a word or phrase stands for and knowing how to link those concepts together in a meaningful way. It's ironic that natural language, the symbol system that is easiest for humans to learn and use, is hardest for a computer to master. The process employed in getting a computer to understand sentences is mad e up of programs which together comprise the “the natural language analyzer “. The basic functions: 1. Lexical analysis 2. Syntactical analysis 3. Semantic analysis

A. Message Processing This Module is used in Message Processing the information or passage is given, in this passage one can ask any type of questions related to that Query. The given passage will be splitted into sentences and also splitted into single words and stored in the database. - Read the document - Split into sentences - Split into single words - Store into database B.Query Processing This module is used to build the query like create, update, delete, select, insert. Receiving queries (Question) from user are Splitted and they are stored in the database for comparing with normal Sql Query. - Build the query like create, update, delete, select, insert - Receiving query (Question) from user - Split the question and get subject, verb and object in that question - Make a question vector C.Image Retrievals This Module is used in the information of the images that will be retrieved based on the original image and the medical images. Sathyabama University

D.LOG MANAGER PROCESSING In the log manager processing, all the entered are updated in the log table, which in process gives out details of the login queries and statements. If the user needs to check the given queries those queries and statements are updated in the log manager processing .And if the user needs to give the same query , statement next time by the use of the this log manager table, retrieve them and enter for the process. This module displays the queries asked by the user with date and time.

A.LEXICAL ANALYSIS Dividing sentences into word and punctuation marks or pauses is called “lexical analysis”. The words themselves can be divided into roots, prefixes and suffixes. Going (word) Go (root) Going (suffix) 39


The word “understand “consists of Under (prefix) Stand (root) The lexicon defines words, but to get the computer to understand them in context is a considerably more difficult task. B.SYNTAX ANALYSIS Breaking sentences down into component parts is a vital step in getting the computer to understand language. We must be able to translate the rules of grammar and syntax into a form that computer can handle. A sentence (S) is generally composed of a noun phrase (NP) and a verb phrase (VP) which can be represented as S-NP, VP A noun phrase can usually be further broken down into determiner and a noun, represented as NP-DET, N A verb phrase can also be broken down further into a verb followed by another noun phrase, as VP-V, NP A noun phrase however can also be just a simple noun; NP-N A common way of breaking down sentences is by creating a parse tree, which is a diagrammatic representation of the syntactical structure of a sentence. “A plane flew home” The words of the sentence are identified and each word is classified by type. the word “A” is a determiner (DET) ,”plane” is a noun (N),”flew” is a verb(V),and “home” is a noun(N). C.SEMANTIC ANALYSIS The process by which the computer attempts to make meaning out of it. In an artificial intelligence system, we would apply a set of rules to establish the meaning in a way that the computer can use. Let‘s take the previous sentence as an example: DET N V N A plane flew home The semantic analyzer might have the following set of rules in its knowledge base for interpreting the sentence: RULES 1:

V. RESULTS & DISCUSSION Temporal table stores both valid and transaction time. The presented framework for generalization and specialization allows researchers as well as database and system designers to precisely characterize, compare, and thus better understand temporal relations and the application systems in which they are embedded. The framework’s comprehensiveness and its use in understanding temporal relations is demonstrated by placing previously proposed temporal data models within the framework. The practical relevance of the defined specializations and generalizations is illustrated by sample realistic applications in which they occur. The additional semantics of specialized relations are especially useful for improving the performance of query processing. Also it provides the data model to capture the time varying nature of entities and to design temporal query language to retrieve, manipulate and process the temporal data. The semantics of time is a big issue to deal with and it varies with the application and domain. Temporal relational database also holds time series data about entities and the events occurring in real time. VI. CONCLUSIONS In this paper, a Temporal Natural Language Interface to query from temporal databases has been designed and implemented. This system helps the novice users to interact with temporal databases by giving a natural language query. In order to capture events in evolutions, the temporal event matching approach in the form of a pattern recognition language has been implemented in this work. This system recognizes the evolutionary nature of time, more naturally, compared to traditional languages. Temporal databases are three-dimensional with time interval as the third dimension. Reducing the effort spent by the query users in forming the queries. The query is transformed from natural language question into SQL query.

IF a determiner is the first part of a sentence and is followed by a noun, THEN the noun is known as the subject. RULE 2: IF a verb follows a subject, THEN the verb tells us what the subject did. RULE 3: IF a noun follows a subject and verb in that order, THEN the verb is as object. RULE 4: IF a sentence has the form subject, verb, object, THEN we know what the subject did in relation to the object.


D.OVERALL PROCESS Message Processing the information or passage is given, in this passage one can ask any type of questions related to that Query. The given passage will be splitted into sentences and also splitted into single words and stored in the database. Query like create, update, delete, select, insert. Receiving queries (Question) from user are Splitted and they are stored in the database for comparing with normal Sql Query. Data’s read from the database, Questions which are given compared with the normal SQL Queries and the output which will be generated for the user according to their Messages.

40


REFERENCES [1] WiniwarterWand Ismail Khalil Ibrahim. A Multilingual Natural Language Interface for Ecommerce Applications. Ph.d thesis, University of Vienna, Austria, 2000. [2] Xiang Sean Zhou and Thanas S. Huang. Unifying keywords and visual contents in image retrieval.IEEE Transactions on Multimedia, 2(1):1–13, 2000. [3] Tsz S. Cheng and Shashi K. Gadia. The event matching language for querying temporal data. IEEE Transactions on Knowledge and Data Engineering, 14(5):1119–1125, 2002. [4] N.Mamoulis, H.Cao, G.Kollios, M.Hadjieleftheriou, Y. Tao, and D.W.L. Cheung, “Mining, Indexing, and Querying Historical Spatiotemporal Data,” Proc. 10th ACM SIGKDD, 2004. [5] H.Cao, N.Mamoulis, and D.W. Cheung, “Mining Frequent Spatio-Temporal Sequential Patterns,” Proc. Fifth IEEE Int’l Conf. Data Mining (ICDM ’05), pp. 82-89, 2005. [6] N. Mamoulis, H. Cao, G. Kollios, M. Hadjieleftheriou, Y. Tao, and D.W.L. Cheung, “Mining, Indexing, and Querying Historical Spatiotemporal Data,” Proc. 10th ACM SIGKDD, 2004. [7] H. Cao, D.W. Cheung, and N. Mamoulis, “Discovering Partial Periodic Patterns in Discrete Data Sequences,” Proc. Eighth PacificAsia Conf. Knowledge Discovery and Data Mining (PAKDD ’04), 2004. [8] H. Cao, N. Mamoulis, and D.W. Cheung, “Mining Frequent Spatio-Temporal Sequential Patterns,” Proc. Fifth IEEE Int’l Conf. Data Mining (ICDM ’05), pp. 82-89, 2005. [9] Gregersen, H. and Jensen, C. S. (2004) Conceptual Modeling of Time-varying Information. In Proceedings of International Conference on Computing, Communications and Control Technologies, pages 248–255. [10] Gregersen, H. (2005) TimeERplus: A Temporal EER Model Supporting Schema Changes. In BNCOD, volume 3567 of Lecture Notes in Computer Science, pages 41–59. Springer. [11] Hoffer, J. A. McFadden, F. R. and Prescott, M. B. (2005) Modern Database Management, 7th edition Upper Saddle River, NJ: Pearson/Prentice Hall. [12] Mkaouar M., Bouaziz R., Moalla M., “Modelling temporal databases and temporal constraints”, accepoted paper at the Third International Conference on Advances in Databases, [29] Knowledge, and Data Applications, 2011 [13] Jensen C. S., Snodgrass R. T. (Editors), Temporal Database Entries for the Springer


41

[14]

[15] [16]

[17]

[18]

[19]

[20] [21] [22] [23]

[24]

[25]

[26]

[27]

[28]

Encyclopaedia of Database Systems, Technical Report TR-90, TIMECENTER, May, 2008. Mkaouar M., Bouaziz R., “UML-TF. Un profil UML pour la représentation des faits temporels.”, Technique et Science Informatiques, vol. 26 n° 3-4, 2007, p. 305-338. Toman D., “SQL/TP a Temporal Extension of SQL.”, Constraint Databases, 2000, p. 391-399. Chomicki J, Toman D., “Temporal Relational Calculus.”, Encyclopedia of Database Systems 2009, p. 3015-3016. Lomet D. B., Barga R. S. , Mokbel M. F., Shegalov G.,Wang R., Zhu Y., “Transaction Time Support Inside a Database Engine.”, Proceedings of the International Conference on Data Engineering, 2006. Garani G., “A generalised temporal algebra.”, Data & Knowledge Engineering 57(3), June 2006. Oracle Corporation, Advanced Application Developer’s Guide: Using Oracle Flashback Technology, Oracle Documentation, 2008. Oracle Corporation, Workspace Manager Developer’s Guide, Oracle Documentation, 2009. Souveyet C., Deneckere R., Rolland C., TOOBIS Methodology. Projet TOOBIS, T23D1.1, December 1997, http://www.di.uoa.gr/~toobis/. Skjellang B., Temporal Data: Time and Object Databases, Research report n° 245, Université de Oslo, Département d’Informatique, 1997. Bertino E., Ferrari E., Guerrini G., Merlo I., “Extending the ODMG object model with time.”, Proceedings of the European Conference on Object-Oriented Programming (ECOOP), July 1998. Fauvet M. C., Dumas M., Scholl, P. C., “A representation independent temporal extension of ODMG’s Object Query Language.”, Bases de données avancées (BDA), France, October 1999. Dumas M., Fauvet M. C., Scholl, P. C., “TEMPOS: A Platform for Developing Temporal Applications on top of Object DBMS.”, IEEE Transactions on Knowledge and Data Engineering, vol. 16, n° 3, March 2004. Chen C. X., Kong J., Zaniolo C., “Design and Implementation of a Temporal Extension of SQL.”, Proceedings of the International Conference on Data Engineering, 2003, p. 689691. Mohamed Mkaouar “Querying And Manipulating Temporal Databases “,“International Journal of Database Management Systems ( IJDMS ), Vol.3, No.1, February 2011”


Comparative Study of Various Classification Algorithms for Heart Data A.Anushya1, A. Pethalakshmi2 1. Ph.D., Scholar, Department of Computer Science, Manonmaniam Sundaranar University, Tirunelveli 627 012, Tamil Nadu, India. E-mail: [email protected] 2. Prof & Head, Department of Computer Science, M.V.M Government Arts College (W), Dindigul-624 001, Tamil Nadu, India. E-mail: [email protected] Abstract: The objective of our work is to predict more accurately the presence of heart disease. Fuzzy logic and data mining are combined to build hybrid intelligent systems that contribute to improvement of the mining results. The classifiers, Decision tree and neural network are implemented to find the accuracy of hart disease. The fusion of fuzzy with decision trees enables to combine the uncertainty handling and conjugation of fuzzy and clustering could be improved to extract the attributes that could be used to obtain more precise risk factors. So, Fuzzy decision trees and Fuzzy K-Means are developed to improve the prediction performance of heart disease. The heart data sets from UCI machine learning repository is considered for these experiments.

is very important step in the management. This work proposes efficient to find the best Fuzzy logic rule based classifier, which is used as an effective tool to improve the classification accuracy. The rest of this paper is organized as follows: Section 2 discusses the related works. Section 3 presents data source. Section 4 describes classifiers. Section 5 summarizes the experimental analysis performed with datasets. Finally, Section 6 concludes the paper. II. LITERATURE SUPPORT As for other clinical diagnosis problems, classification systems have been used for heart disease diagnosis problem, too. When the studies in the literature related with this classification application are examined, it can be seen that a great variety of methods were used which reached high classification accuracies using the dataset taken from UCI machine learning repository. Our work is an attempt to predict efficiently diagnosis with the factors (i.e. attributes) that contribute more towards the cardiac disease using classification. Anbarasi et al exhibited that Decision Tree were used to predict the diagnosis of patients with the same accuracy as obtained before the reduction of number of attributes. Naïve Bayes performed consistently before and after reduction of attributes with the same model construction time. Classification via clustering performs poor compared to other two methods[1]. A model Intelligent Heart Disease Prediction System (IHDPS) built with the aid of data mining techniques like Decision Trees, Naïve Bayes and Neural Network was proposed by Sellappan Palaniappan et al. The results illustrated the peculiar strength of each of the methodologies in comprehending the objectives of the specified mining objectives. IHDPS was capable of answering queries that the conventional decision support systems were not able to. It facilitated the establishment of vital knowledge, e.g. patterns, relationships amid medical factors connected with heart disease. IHDPS subsists well being web-based, user-friendly, scalable, reliable and expandable [7]. Asha et al developed an Intelligent Heart Disease Prediction System to predict the heart disease using three classifiers Decision Tree, Naïve Bayes and Neural Networks. Naïve Bayes performed with good prediction probability of 96.6%. Also, 13 attributes were used for prediction and differed by reducing the

Keywords -- Data mining, Decision trees, K-Means, Fuzzy logic, Fuzzy decision trees, Fuzzy K-Means

I .INTRODUCTION A. Data mining Data mining is a multidisciplinary effort to extract nuggets of knowledge from data. The proliferation of large data sets within many domains poses unprecedented challenges to data mining (J.Han and M.Kamber, 2001)[6]. B. Fuzzy logic and Fuzzy sets Fuzzy logic is a superset of conventional (Boolean) logic that has been extended to handle the concept of partial truth- truth values between "completely true" and "completely false". Fuzzy logic is used in system control and analysis design, because it shortens the time for engineering development and sometimes, in the case of highly complex systems, is the only way to solve the problem. The importance of fuzzy logic derives from the fact that most modes of human reasoning and especially common sense reasoning are approximate in nature [12]. Fuzzy Set is any set that allows its members to have different grades of membership (membership function) in the interval [0, 1]. C. Heart disease Heart disease is the world's leading killer, accounting for 16.7 million or 29.2 per cent of total global deaths in 2003. The World Health Organization in 2009 estimated that almost 20 million deaths occur annually from cardiovascular disease and that by 2030 that figure could rise to almost 24 million. The World Health Organization (WHO) estimated that 60% of the world's cardiac patients are Indian. Prediction of this disease will help to prevent it in its early stage. The cost of management of Heart disease is a significant economic burden and so prevention of heart disease Sathyabama University

42


A. Decision Trees The knowledge represented in decision trees can be extracted and represented in the form of IF-THEN rules. One rule is created for each path from the root to a leaf node. Each attribute-value pair along a given path forms a conjunction in the rule antecedent (“IF” part). The leaf node holds the class prediction, forming the rule consequent (“THEN” part). The IFTHEN rules may be easier for humans to understand, especially if the given tree is very large.

number of attributes to 6 and were able to achieve the same performance [2]. Carlos implemented efficient search for diagnosis of heart disease comparing association rules with decision trees. This suggestion is promising as data modeling and analysis tools have the potential to generate a knowledge-rich environment which can help to significantly improve the quality of clinical decisions[3]. Harleen et al examined the potential use of classification data mining technique like decision tree, rule induction and artificial neural network for diagnosis of diabetic patients[4].

B. C 4.5 Algorithms: C4.5 builds decision trees from a set of training data using the concept of information entropy. The training data is a set S = s1,s2,... of already classified samples. Each sample si = x1,x2,... is a vector where x1,x2,... represent attributes or features of the sample. The training data is augmented with a vector C = c1,c2,... where c1,c2,... represent the class to which each sample belongs[13]. At each node of the tree, C4.5 chooses one attribute of the data that most effectively splits its set of samples into subsets enriched in one class or the other. Its criterion is the normalized information gain that results from choosing an attribute for splitting the data. The attribute with the highest normalized information gain is chosen to make the decision. The C4.5 algorithm then recurs on the smaller sublists.

III. DATA SOURCE Available dataset of Heart disease from UCI Machine Learning Repository has been studied and preprocessed and cleaned out to prepare it for classification process. A total of 909 records with 13 medical attributes (factors) were obtained from the Heart Disease database. Figure 1 lists the attributes. Predictable attribute 1. Diagnosis (value Heal: < 50% diameter narrowing (no heart disease); value Sick: > 50% diameter narrowing (has heart disease)) Key attribute 1. PatientID – Patient’s identification number Input attributes 1. 2. 3. 4.

5. 6.

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

C. Fuzzy Decision Tree

Sex (value 1: Male; value 0 : Female) Chest Pain Type (value 1: typical type 1 angina, value 2: typical type angina, value 3: non-angina pain; value 4:asymptomatic) Fasting Blood Sugar (value 1: > 120 mg/dl; value 0:< 120 mg/dl) Restecg – resting electrographic results (value 0: normal; Value 1: 1 having ST-T wave abnormality;value 2: showing probable or definite left ventricular hypertrophy) Exang – exercise induced angina (value 1: yes; value 0: no) Slope – the slope of the peak exercise ST segment (value 1: unsloping; value 2: flat; value 3: downsloping) CA – number of major vessels colored by floursopy (value 0 – 3) Thal (value 3: normal; value 6: fixed defect; value 7: reversible defect) Trest Blood Pressure (mm Hg on admission to the hospital) Serum Cholesterol (mg/dl) Thalach – maximum heart rate achieved Oldpeak – ST depression induced by exercise relative to rest Age in year

A fuzzy decision tree gives results within [0, 1], as the possibility degree of an object matching the class. Fuzzy decision trees provide a more robust way to avoid misclassification. In fuzzy decision tree, nodes of the tree of degree one, the leaf nodes, are labeled with what are referred to as root concepts. Nodes of degree greater than unity are labeled with composite concepts, i.e., concepts constructed from the root concepts using “AND,” “OR,” and “NOT.” Each root concept has a fuzzy membership function assigned to it. The membership functions for composite concepts are constructed from those assigned to the root concepts using fuzzy logic connectives and modifiers. Each root concept membership function has parameters that are determined by optimization[14]. D. K-Means clustering The categorization of objects into various groups or the partitioning of data set into subsets so that the data in each of the subset share a general feature, frequently the proximity with regard to some defined distance measure, is known as Clustering. K-means groups the data in accordance with their characteristic values into K distinct clusters. Data categorized into the same cluster have identical feature values. K, the positive integer denoting the number of clusters, needs to be provided in advance. The steps involved in a K-means algorithm are given subsequently:

Fig. 1. Attributes list and description IV.CLASSIFICATION

Classification is one of the forms of data analysis that can be used to extract models describing important data classes or to predict categorical labels. There are several basic techniques for data classification.


43


1) K points denoting the data to be clustered are placed into the space. These points denote the primary group centroids. 2) The data are assigned to the group that is adjacent to the centroid. 3) The positions of all the K centroids are recalculated as soon as all the data are assigned. 4) Steps 2 and 3 are reiterated until the centroids stop moving any further. This results in the segregation of data into groups from which the metric to be minimized can be deliberated. The preprocessed heart disease data warehouse is clustered using the K-means algorithm with K value as 2. One cluster consists of the data relevant to the heart disease and the other contains the remaining data [8]. E. Fuzzy K-Means clustering Fuzzy K-Means works on those objects which can be represented in n-dimensional vector space and a distance measure is defined. The algorithm is similar to k-means. Initialize k clusters Until converged o Compute the probability of a point belong to a cluster for every pair o Recompute the cluster centers using above probability membership values of points to clusters V. EXPERIMENTAL RESULTS Experiments are conducted with MATLAB. Data set of 909 records with 13 attributes are used. The classifiers such as k-means, decision trees, Fuzzy kmeans, Fuzzy decision trees, are used for diagnosis of patients with heart disease. Observations exhibit that the Fuzzy k-means technique outperforms than others. Results are shown in below: 100

DETECTION ACCURACY Data mining Techniques Decision Tree

K means

K means Fuzzy Decision Tree

80

Fuzzy Decision Tree

Fuzzy K means

1

76.2500

88.7500

68.7500

93.7500

2

82.0970

88.5480

87.0970

93.5480

3

79.7830

92.8260

91.3040

97.8260

4

81.8850

93.3610

86.8850

100.0000

5

79.2110

95.0000

82.8950

97.3680

Fuzzy K means

70 60

A ccuracy

Iteration

Decision Tree

90

50 40 30 20 10 0

6

79.6150

95.0000

83.5160

98.9010

7

78.9620

94.0570

82.0750

98.1130

8

85.9090

93.3470

84.2980

99.1740

9

90.5880

93.5290 84.5590

99.2650

10

89.7020

94.3380 86.7550

100.0000


1

2

3

4

5

6

7

8

9

10

Classifiers

Performance analysis of classifiers

VI. CONCLUSION The objective of our work was to predict more accurately the presence of heart disease. Originally, thirteen attributes were involved in predicting the 44

Proceedings of National Conference on Computing Concepts in Current Trends, NC4T’11 11th & 12th Aug. 2011, Chennai, India [9]

heart disease. In our work, the classifiers decision trees, K-means, Fuzzy decision trees and Fuzzy Kmeans were used to predict the heart disease. Also, the observations exhibited that the Fuzzy K-means technique outperformed than other classifiers.

[10] Srinivas .K : “Applications of Data Mining Techniques in Healthcare and Prediction of Heart Attacks,” (IJCSE) International Journal on Computer Science and Engineering, Vol. 02, No. 02, 2010.

REFERENCES [1]

[2]

[3]

[4]

[5]

[6] [7]

[8]

Anbarasi.M, E. Anupriya and N.CH.S.N.Iyengar: ”Enhanced Prediction of Heart Disease with Feature Subset Selection using Genetic Algorithm,” International Journal of Engineering Science and Technology Vol. 2(10), pp 53705376, 2010. Asha Rajkumar and Mrs. G.Sophia Reena: “Diagnosis Of Heart Disease Using Datamining Algorithm,” GJCST,Vol. 10 Issue 10 Ver., pp. 38-43, 2010. Carlos Ordonez: “Comparing Association Rules and Decision Trees for Disease Prediction,” ACM, HIKM’06, Arlington, Virginia, USA, 2006.. Harleen Kaur and Siri Krishan Wasan : “Empirical Study on Applications of Data Mining Techniques in Healthcare”, Journal of Computer Science 2 (2): 194-200, ISSN pp.15493636, 2006. Honggui Han, and Junfei Qiao: “A Self-Organizing Fuzzy Neural Network Based on a Growing-and-Pruning Algorithm,” IEEE transactions on fuzzy systems, vol. 18, no. 6, 2010. J. Han and M. Kamber. Data Mining: Concepts and Techniques. Morgan Kaufman, 2001. Sellappan Palaniappan, Rafiah Awang, "Intelligent Heart Disease Prediction System Using Data Mining Techniques", IJCSNS International Journal of Computer Science and Network Security, Vol.8 No.8, August 2008. Shantakumar B.Patil and Dr.Y.S.Kumaraswamy “Extraction of Significant Patterns from Heart Disease Warehouses for Heart Attack Prediction”, IJCSNSInternational Journal of Computer Science and Network 228 Security, VOL.9 No.2, February 2009.


Shantakumar B.Patil and Y.S.Kumaraswamy : “Intelligent and Effective Heart Attack Prediction System Using Data Mining and Artificial Neural Network,” European Journal of Scientific Research ISSN 1450- 216X Vol.31 No.4, pp. 642656, 2009.

[11] Sushmita Mitra, Kishori M. Konwar, and Sankar K. Pal : “Fuzzy Decision Tree, Linguistic Rules and Fuzzy Knowledge-Based Network,” Generation and Evaluation, IEEE transactions on systems, man, and cybernetics—part c: applications and reviews, vol. 32, no. 4, 2002. [12] Timm H. and R. Kruse, “ Fuzzy cluster analysis with missing values”: Proceedings of 17th International Conference of the North American Fuzzy Information Processing Society (NAFIPS98), pp. 242-246, Pensacola, FL, 1998. [13] Veronica S.and Moertini,” Towards the use of c4.5 algorithm for classifying banking dataset”, INTEGRAL, Vol. 8 No. 2, October 2003. [14] Yeung, D.S.; Tsang, E.C.C.; Xizhao Wang; “Fuzzy rule mining by fuzzy decision tree induction based on fuzzy feature subset” Systems, Man and Cybernetics, 2002 IEEE International Conference on , Volume: 4 , 6-9, Oct.2002 Pages:6 pp. vol.4 [15] Zakaria Nouir, Berna Sayrac, Benoît Fourestié, Walid Tabbara, and Françoise Brouaye, "Generalization Capabilities Enhancement of a Learning System by Fuzzy Space Clustering," Journal of Communications, Vol. 2, No.6, pp. 30-37, November 2007.

45


Role of Data Mining in Network Intrusion Detection Authors :

1. Asst.Prof.K.Chokkanathan Department Of MCA Saraswathi Velu College of Engg. Sholinghur

2. Asst.Prof.D.Saravanan Department Of MCA Sathyabama University

Abstract: Recently data mining much more used in computer network intrusion detection. In this paper we have come through how use data mining in this context and based upon our experiences in getting started on this type of project, we suggest data mining techniques to consider and types of expertise and infrastructure needed. This paper has two major categories of people like: network security professionals with little background in data mining, and data mining experts with little background in network intrusion detection. Key words: data mining, intrusion detection, computer network security

1.

to highlight that data? Etc. The analyst team got ready handle the load, and training and team coordination were the issues of the day. But the level of investigation and assail on the internet was constantly mounting; alongside with the amount of data we were gathering and putting in front of our analysts. We thought that system was inadequate for detecting the most hazardous attacks—those performed by adversaries using attacks that are new, furtive, or both. So we considered data mining with two questions in mind: What the analysts need to look at daily to minimize attacks?

INTRODUCTION TO NETWORK INTRUSION DETECTION

How to use data mining to find attacks that the sensors and analysts did not find? Pattern-based software ‘sensors’ monitor the network traffic and raise ‘alarms’ when the traffic matches a saved pattern. Intrusion detection starts with instrumentation of a computer network for data collection. Security analysts decide whether these alarms indicate an event serious enough to warrant a response. A response should be to shut down a part of the network, to phone the internet service provider associated with suspicious traffic, or to simply make note of unusual traffic for future reference. If the network is small then it is easy to analysis intrusion detection with updated signatures. But when the organizations are large network become complex, we need number of alarms they need to review. The sensors on the MITRE network, for example, currently generate over one million alarms per day. And that number is increasing. This situation arises from ever increasing attacks on the network, as well as a tendency for sensor patterns to be insufficiently selective. Commercial tools do not provide an enterprise level view of alarms generated by multiple sensor vendors. Commercial intrusion detection software packages tend to be signature-oriented with little or no state information maintained. So we need to investigate the application with the data mining concepts.

3. DATA MINING Data mining is, at its core, pattern finding. Data miners are experts at using specialized software to find regularities (and irregularities) in large data sets. Here are a few specific things that data mining might contribute to an intrusion detection project: Remove normal activity from alarm data to allow analysts to focus on real attacks Identify false alarm generators and “bad” sensor signatures Find anomalous activity that uncovers a real attack Identify long, ongoing patterns (different IP address, same activity) To accomplish these tasks, data miners use one or more of the following techniques: Data summarization with including finding outliers

statistics,

Visualization: presenting summary of the data

graphical

a

2. INTRUSION DETECTION AT THE BEGINNING 4. MAKING OUR REQUIREMENTS REALISTIC The seductive vision of automation is that it can and will solve all your problems, making human involvement unnecessary. This is a illusion in

There was lot of questions like: How would the sensors perform? How much data would we get? How would we display the data? What kind of data did we want to see, and what queries would be best Sathyabama University

46


database middleware, database administration and tuning aids), plan for acquisition of specialized tools. Also we need a person who can use the software effectively. Our team members should have working experience in data mining.

intrusion detection. Human analysts will always be needed to monitor that the automated system is performing as desired, to identify new categories of attacks, and to analyze the more refined attacks. Our motive is to release the analyst’s day to day burden. Real-time automated response is very enviable in some intrusion detection contexts. But the database must be fast enough to record alarms and produce query results at once. Real time scoring of incongruity or classification models is possible, but this should not be confused with realtime model building. Data mining is not currently capable of learning from large amounts real-time, energetically changing data. It is better suited to batch processing of a number of collected records. Therefore, we are implementing a daily processing system, rather than an hourly or minute-by-minute scheme.

7. PROCESSING THE ATTRIBUTES The individual connection records in a denial of service attack are not, by themselves, malicious, but they come in such numbers that they overwhelm your network. A single connection between an outside machine and a single port on a machine inside your network is also not malicious— unless it is part of a series of connections that attempted to map all the active ports on that machine. For this reason you will want to add additional fields containing values derived from the base fields. For example, you could distinguish traffic originating from outside your network from traffic originating inside your network. Another type of derived data, called an aggregation, is a summary count of traffic matching some particular pattern. For example, we might want to know, for a particular source IP address X, and a particular IP address Y, how many unique destination IP addresses were contacted in a specific time window Z.

5. CAPABILITY OF STAFF INVOLVING IN PROJECT STAFF We need skills in three areas: network security, data mining, and database application development. Strong knowledge in networking and intrusion detection, but they also need to be able to tackle big, abstract problems. The database developers will need good skills in efficient database design, performance tuning, and data warehousing. Sound knowledge in statistics and machine learning, but they will also need to learn detailed concepts involved in computer networking. This team will have to do a lot of cross orientation to begin working effectively. Initially, security and networking concepts must be introduced and defined.

Aggregations are generally more expensive to compute than other kinds of derived data that are based upon only a single record. A third type of derived data is a flag indicating whether a particular alarm satisfies a heuristic rule. Because data mining methods handle many attributes well, and because we don’t know for sure which one will be useful, our approach is to compute a large number of attributes (over one hundred) and store them in the database with the base alarm fields.

6. REQUIRED INFRASTRUCTURE

8. IMPLEMENTING DATA FILTERS

Significant infrastructure is required to do this sort of work. In addition to the normal processing of the data from the intrusion detection system, you will need i. Back up : A great deal of data, update this data regularly, and obtain rapid responses to complex queries, we recommend that you select a high-end production-quality database management system.

In our sensor log table, upwards of 95% of the traffic fit the profile of an IP mapping activity. That is, a single source IP was attempting a connection to hundreds or even thousands of destinations IPs. Before security specialists can start providing input to the data mining effort, this traffic must be filtered. It is a straightforward task to create a filter that can find these patterns within a data table of traffic.

ii. Storage Area: In addition to the handling of normal IDS data, we will need data and working space associated with data mining. Additional data includes calculating and saving metadata, as well as sometimes copying existing data into more convenient data types.

At MITRE, this preliminary filter is called HOMER (Heuristic for Obvious Mapping Episode Recognition). The heuristic operates on aggregations by source IP, destination port, and protocol and then check to see if a certain threshold of destination IPs were hit within a time window. If the threshold is crossed, an incident is generated and logged to the database. The reduction obtained by HOMER is significant. For example, for the period of Sep. 18 to Sep. 23, 2000, MITRE network sensors generated 4,707,323 alarms IP mapping activity does not pose

iii. Computing Ability: Data mining tools are very CPU and memory intensive. Naturally, the more memory and CPU power the better. iv. Software: In addition to what is required for the basic system (production quality database, Perl,


47


variety of algorithms. Henery [1994] classifies classification algorithms into three types: i. Extensions to linear discrimination (e.g., multilayer perception, logistic discrimination), ii. Decision tree and rule-based methods (e.g. C4.5, AQ, CART), and iii. Density estimators (Naïve Bayes, k-nearest neighbor, LVQ). In this work we have, so far, used only decision tree and rule-based methods because of their familiarity to us and because of their ability to give human understandable results. Excellent Examples. The ‘quality’ of the training data is one of the most important factors in achieving good classifier performance. Training data quality is a function of the number of examples, how representative the examples are, and the attributes used to describe them. Titled Data. Supervised classification uses labeled training examples to build a model. The labels usually come from a human expert (or experts) who manually review cases. In our application of classification to intrusion detection we obtained labeled examples by building a web-based interface that required a label to be assigned to a new incident each time it was constructed by an analyst. Using this feedback we were able to collect 12,900 labeled examples of seven different classes of incidents from August 2000 and 16,885 for September 2000. Classes. Another factor in getting good examples is to have a well-defined set of classes. It is important to maintain consistency in assigned labels over time, both for a single person and across multiple people. Label inconsistency can make classification very difficult especially if identical examples are labeled ambiguously.

much of a security threat in itself, but it can be a prelude to more serious activity. Thus, HOMER provides one other important function. HOMER handles this situation by means of an exclusion list of source IPs. A second heuristic under development, called GHOST (Gathering Heuristic for Obvious Scanning Techniques), plays a slightly dissimilar role than HOMER. Port scanning is a more embattled form of information gathering that attempts to profile the services that are run on a possible intrusion target. The GHOST heuristic uses a different set of fields, and has its own configurable time window and port threshold, which if exceeded, triggers a security incident. 9. OVERALL ARCHITECTURE REFINEMENT FOR INTRUSION DETECTION Our current architecture for intrusion detection is shown in Figure 1. Network traffic is analyzed by a variety of available sensors. This sensor data is pulled periodically to a central server for conditioning and input to a relational database. HOMER filters events from the sensor data before they are passed on to the classifier and clustering analyses.

11. PERFORM ANOMALY DETECTION Both intruder techniques and local network configurations will change. In spite of efforts to update defenses, new attacks may slip through defenses and be labeled as either normal network traffic, or else filtered as a known but benign probe. Anomaly detection techniques can help humans prioritize potentially anomalous records for review. Catching new attacks can not depend on the current set of classification rules. Much of the work in outlier detection has been approached from a statistical point of view and is primarily concerned with one or very few attributes. However, because the network data has many dimensions, we have investigated use of clustering for anomaly detection. Clustering is an unsupervised machine learning technique for finding patterns in unlabeled data with many dimensions (number of attributes). We use kmeans clustering to find natural groupings of similar alarm records. Records that are far from any of these clusters indicate unusual activity that may be part of a new attack. The network data available for intrusion detection is primarily categorical (i.e.,

Figure 1. How sensors feed into overall intrusion detection system A web server is available as a front end to the database if needed, and analysts can launch a number of predefined queries as well as free form SQL queries from this interface. The goal of this operational model is to have all alarms reviewed by human analysts. Without automated support, this task is increasingly difficult due to the volume of alarms. In one recent day at MITRE for example, sensors generated about 3.4 million alarms, of which about 48,000 are labeled priority 1. Attacks and probes can be frequent and noisy, generating thousands of alarms in a day. This can create a burden on the network security analyst, who must perform a triage on the enormous flood of alarms. 10. BUILD CLASSIFICATION RULES Examples for pre-defined categories can be assigned through Classification. Machine learning software will to this task by extracting or learning prejudice rules from examples of correctly classified data. Classification models can be built using a wide


48


aggregates reduce the downstream volume of data.

Attributes have a small number of unordered values). Clustering approaches for categorical data, such as in are not generally available commercially. Unsupervised approaches for detecting outliers in large data sets for the purposes of fraud or intrusion detection are starting to appear in the literature, but these approaches are primarily based on ordered data. Knorr and Ng [1998] recently developed a distance-based clustering approach for outlier detection in large data sets. Ramaswarny, et al. [2000] define a new outlier criterion based on the distance of a point to its kth nearest neighbor. Breunig et al. [2000] define a new local outlier factor, which is the degree to which a data point is an outlier.

While most attributes and aggregates are used to feed an automated process, don’t forget the analysts. Analysts must have efficient tools to spot check the automatically generated security incidents, and to manually comb through the raw sensor data for new or complex patterns of malicious activity. The MITRE interface is centered on a set of predefined queries of the sensor database, and a browser of the incident database. With this tool, an analyst can create new security incidents or update existing incidents with new status information. Due to the high volume and frequency of data inputs, and the variety of both automated and human data sources, there will invariably be some process failures. When a failure does occur, the condition must be caught and the security team notified. Scripts that verify the integrity of the data tables, and repair inconsistencies, are useful. If possible, the process should be halted until the error is corrected. But, in some situations, the ability to operate normally regardless of errors, and then rollback and correct statistics and attributes at the team’s convenience, may be a more practical recovery strategy.

12. MAKING EFFICIENT SYSTEM There are a number of practical considerations in building an effective intrusion detection system. Some of these derive from the use of data mining, but many of them would be present in any intrusion detection system: A central repository must be designed and enabled. The repository must allow for inputs from a potentially large number of diverse network sensors, preferably within a single data table. Any derived data, such as data mining attributes, should also be stored in this central location. It must also support the creation and tracking of security incidents.

Scheduling is an important aspect of the operational environment. Each organization must decide for itself how much of its intrusion detection system truly needs to be “realtime”. The calculation of real time statistics must be completed in a matter of seconds, and the amount of data available in this manner will always be limited. But daily batch processing of data may be adequate in many cases.

Efficient querying is essential to feed the daily operations of security analysts. A bottleneck in querying the data will affect everything else in the system. Some steps that can be taken to improve query efficiency include the inclusion of a database performance guru on the project team, statistical/ trend analysis of query performance over time, elimination of timeconsuming queries, or the retirement of old data from the database.

13. CONCLUSION We have described our experiences with integrating data mining into a network intrusion detection capability. We believe that when starting such a project you should:

Efficiency can also be improved by selecting appropriate aggregations of attributes and statistics. A manual analysis of network activity will reveal that a large volume of atomic network activity breaks down into a much smaller set of meaningful aggregates. At MITRE, two of the more useful aggregates were (source IP, destination port), used for catching some IP mapping activity, and (source IP, destination IP), used for catching port scanning activity. But, any combination of fields or attributes could also be used, resulting in a wealth of choices. Regardless of the fields used,

Choose your requirements carefully and be realistic. Assemble a team with broad, relevant capabilities. Invest in required building blocks to support data collection and data mining. Design, compute, and store appropriate attributes with your data. Filtering used to Reduce data volume.


49


13. Mounji, A. (1997). Languages and Tools for RuleBased Distributed Intrusion Detection. PhD thesis, Facult_es Universitaires Notre-Dame de la Paix Namur (Belgium). 14. Mukkamala, R., Gagnon, J., and Jajodia, S. (1999). Integrating Data Mining Techniques with Intrusion Detection Methods. In Proceedings of the 13th IFIP WG11.3 Working Conference on Database Security, pp33-46. 15. Pevzner, P. A. and Sze, S.-H. (2000). Combinatorial Approaches to Finding Subtle Signals in DNA Sequences. In Proceedings of the 8th International Conference on Intelligent Systems for Molecular Biology, pp. 269-278. 16. Silberschatz, A. and Tuzhilin, A. (1996). On Subjective Measures of Interestingness in Knowledge Discovery. In Proceedings of the First International Conference on Knowledge Discovery and Data Mining, pp. 275-281. 17. Smaha, S. E. (1988). Haystack: An Intrusion Detection System. In Proceedings of the 4th IEEE Aerospace Computer Security Applications Conference, Orlando, FL, pp.37-44. 18. Smyth, P. (2001). Breaking out of the Black-Box: Research Challenges in Data Mining. In Proceedings of the ACM SIGMOD International Workshop on Research Issues in Data Mining and Knowledge Discovery (DMKD'01). 19. Srikant, R. and Agrawal, R. (1996). Mining Quantitative Association Rules in Large Relational Tables. In Proceedings of the 1996 ACMSIGMOD Conference on Management of Data, pp.1-12. 20. Stedman, C. (1997). Data Mining for Fool's Gold. Computerworld, 31(48). 21. Teng, H. S., Chen, K., and Lu, S. C. (1990). Adaptive Real-Time Anomaly Detection Using Inductively Generated Sequential Patterns. In Proceedings of the IEEE Symposium on Research in Security and Privacy, Oakland, CA, pp.278-284. 22. Vaccaro, H. S. and Liepins, G. E. (1989). Detection of Anomalous Computer Session Activity. In Proceedings of the IEEE Symposium on Research in Security and Privacy, Oakland, CA, pp.280-289. 23. Warrender, C., Forrest, S., and Pearlmutter, B. (1999). Detecting Intrusions Using System Calls: Alternative Data Models. In Proceedings of the IEEE Symposium on Research in Security and Privacy, Oakland, CA, pp.133-145.

Refine both automated processing and human analysis. Use data mining techniques such as classification, clustering, and anomaly detection, to suggest new filter rules. Ensure that automated data processing can be done efficiently. 14. REFERENCES 1.

Agrawal, R. and Srikant, R. (1994). Fast Algorithms for Mining Association Rules. In Proceedings of the 20th International Conference on Very Large Databases, pp.487-499. 2. Almgren, M., Debar, H., and Dacier, M. (2000). A Lightweight Tool for Detecting Web Server Attacks. In Proceedings of the Network and Distributed System Security Symposium (NDSS'00), pages 157{170. 3. Barbar_a, D., Wu, N., and Jajodia, S. (2001b). Detecting Novel Network Intrusions Using Bayes Estimators. In Proceedings of the _rst SIAM International Conference on Data Mining (SDM'01). 4. Berry, M. J. A. and Lino_, G. (1997). Data Mining Techniques. John Wiley and Sons, Inc. 5. Brachman, R. J., Khabaza, T., Kloesgen, W., Piatetsky-Shapiro, G., and 6. Simoudis, E. (1996). Mining Business Databases. Communications of the ACM, 39(11):42-48. 7. Cohen, W. W. (1995). Fast Effective Rule Induction. In Proceedings 12th International Conference on Machine Learning, pp.115-123. 8. Dain, O. and Cunningham, R. K. (2001). Fusing Heterogeneous Alert Streams into Scenarios. In Proceedings of the ACM CCS Workshop on Data Mining for Security Applications.Debar, H., Dacier, M., Nassehi, M., and Wespi, A. (1998). Fixed vs. 9. Variable-Length Patterns for Detecting Suspicious Process Behavior. In Proceedings of the 5th European Symposium on Research in Computer Security, pp.115. 10. Debar, H., Dacier, M., and Wespi, A. (2000). A Revised Taxonomy for Intrusion Detection Systems. Annales des Telecommunications, 55(7-8):361-378. 11. Eskin, E. (2000). Anomaly Detection over Noisy Data Using Learned Probability Distributions. In Proceedings of the International Conference on Machine Learning (ICML). 12. Ester, M., Kriegel, H.-P., Sander, J., Wimmer, M., and Xu, X. (1998). Incremental Clustering for Mining in a Data Warehousing Environment. In Proceedings of the 24th International Conference on Very Large Databases (VLDB'98), pp.323-333.


50


A Lossy Medical Image Compression Scheme Using Inner Elliptic ROI and Biorthogonal Spline Wavelet R Loganathan#1, Dr. Y S Kumaraswamy *2 #

Research scholar, Department of Computer Science and Engineering, Sathyabama University Jeppiaar Nagar, Chennai, Tamilnadu, INDIA Prof. & HOD, Department of Computer Science and Engineering, HKBK College of Engineering Banaglore, Karnataka,INDIA [email protected] * Sr. Prof. & HOD, Department of MCA (VTU), Dayanada Sagar Institutions Kumarawsamy Layout, Bangalore, Karnataka, INDIA, [email protected] Abstract: With increasing popularity of telemedicine image compression in telemedicine is important for narrow band and high latency networks. In medical image compression the compression ratio is dependent on the image to be compressed among other factors. Various innovative wavelet compression techniques have been proposed which gives good compression compared to JPEG technique for the same image resolution. Wavelet transform becomes a first choice for lossy compression due to its embedding strategy wherein the large amplitude transform parameters are first processed and the wavelet coefficients are progressively refined across scales with quad trees. To improve compression without loss of important features, Region of Interest plays a crucial role in determining different levels of lossy compression on the medical image. In this paper we propose the inner elliptic region of interest, a novel method to identify region of interest and apply Biorothogonal Spline Wavelet for compression with different decomposition levels on the medical image with satisfying results.

image in a compressed manner. The compressed image is reconstructed using a decoder. Compression algorithms can be broadly classified into lossless and lossy compression systems. Techniques using lossless image compression do not change any pixel intensity in the image and hence when decompressed the original image with full energy retained can be constructed. However since pixel intensities are not changed, this limits the compression ratio. In lossy compression the originality of the image is lost and the amount of retained energy to be maintained is decided by the user. The encoders used for lossy compression typically convert the spatial data to frequency domain by mapping the pixels and reducing the transform coefficients by eliminating the majority of the insignificant coefficients. Though raw data and images are represented in bits, the compression method is significantly different in both the cases[4]. Joint photgraphic experts group (JPEG) has become a standard for compression in many areas and has been proposed for medical applications [5]. However JPEg has been found to suffer from blocking artifacts that become evident as the compression ratio increases[6].To reduce the artifacts newer compression techniques have been proposed based on wavelet transform with various modifications[7]. Shapriro's compression technique is one of the most popular technique and is extensively based on the wavelet transform and self similarity features in the images [8]. Various filter bank techniques have been proposed with wavelet transforms with good compression ratios [9]. Wavelets can be described as functions that split the data into different frequency components and analyse each component at a resolution matched to its scale. This is very useful in scenarios where the signal contains discontinuities or sharp change is amplitude. Wavelets are created by dilating and translating a single prototype function or wavelet (t )

Keywords—Medical Image Compression, Wavelet transform, Biorthogonal spline wavelet, lossy compression, Region of interest.

I. INTRODUCTION With large amount of medical imaging produced in hospitals transmission of these large data for telematic applications is a challenge. Medical image compression has become a necessity in telemedicine to reduce the size of data which is essential for bandwidth constrained networks [1]. Lossless compression is a viable solution, however with the amount of data generated in radiology the required compression levels is not achievable using lossless compression. For successful medical image compression the reconstructed image should preserve all the significant characteristics present in the original image[2] The idea behind image compression is to reduce the number of data required to represent the original image and remove redundant information so that when reconstructed, the originality lost is minimal and within acceptable levels. Image compression is highly dependent on the image to be compressed, similarity between neighbour pixels, and the human perception of the image information[3]. An encoder is used to exploit the redundancy in an image and represent the Sathyabama University

The mother or basic wavelet[10]

51

(t ) must satisfy


II. EXPERIMENTAL SETUP Labview was used to implement the proposed Inner Elliptic ROI algorithm. The basic idea of this algorithm is that in medical images the outer edge is generally not the region of interest for diagnosis and the second layer is the probable area from where the diagnosis is made.

The continuous wavelet transform of f(t) [11]with respect to the wavelet (t ) can be given by

The inverse of the continuous wavelet transform is given by

This section describes in detail the proposed Inner Elliptic ROI algorithm that has been implemented. Edge detection plays a very important role in image processing and defines the boundaries between various regions in a image. The quality of edge detection is dependent on the presence of areas with similar intensities, density of edges and noise.

Another type of wavelet is based on the binary scaling and dyadic translations to form the basis functions. For a signal f(t) the output becomes

j , k (t )

formed from the mother wavelet

The steps involved in computation of the Inner Elliptic ROI is given below. Step 1 : Smooth the image with using Gaussian approximation in both axes.

(t ) produce

the wavelet expansion functions[12] that can form an orthogonal basis defined by

The Gaussian function for one dimension is given by

Where j determines the dilation and the translation is specified by k. The two dimensional value is also known as discrete wavelet transform.

Where

represents the standard deviation.

Step 2 : Compute the gradient of the image which indicates the change in intensity for both the axis.

Though various compression ratios with maximum energy retained have been proposed in literature, improving the compression ratio while retaining the maximum energy hits the road block as medical images are highly sensitive to noise. To overcome the limitations of lossy compression different compression techniques on the same image based on the area of interest has been proposed.

Step 3: Maximum gradient value indicates the presence of edges. For each pixel compute the magnitude of the gradient and check if the gradient value is greater at one pixel's distance away in both direction perpendicular to the gradient. If the pixel gradient value is not greater compared to the others in either direction, ignore it.

Metrics used to compare the image compression techniques are the Peak signal to noise ratio which is a measure of peak error and the Mean square error which is the cumulative squared error between the original and the compressed image. The PSNR and the mean squared error is given by

Step 4: Select an upper and lower limit and eliminate pixels which do not fall within the upper and lower limit. Step 5: Two Locations of outermost pixels obtained in step 4 is used to create an imaginary rectangle such that {xi,yk} is the location of pixel on the top left side of the image obtained in step 4.

where I(x,y) is the original image, I'(x,y) is the decompressed image. The image is represented by M by N pixels.

{xi+n,yk+m} is the location of pixel on the bottom left side of the image obtained in step 4. This forms the outer layer of the ROI.

In this paper we propose a novel region of interest identification based on the characteristics of medical image and use the lossy biorthogonal spline wavelet transform for compression. Section II describes the proposed methodology and the experimental setup and section III analyses the results obtained. Details of Biorthogonal spline wavelet is available in[13,14,15]. Sathyabama University

Step 5 is repeated starting from location {xi+p,yk+q}such that the pixel distance in horizontal and vertical direction is above the set threshold value. The inner imaginary rectangle so created is used to compute the Inner elliptic ROI. 52


Table I : The decomposition and reconstruction values used in the experimental setup along with decomposition level.

Step 6: The length and height of the imaginary rectangle is computed. Step 7 : An ellipse is drawn in this area by

Image Decomposition Decomposition Reconstruction Area Level Value value Original 1 1 3 Image Region of 1 1 3 Interest Non Region of 8 3 9 Interest

Where (x0,y0) represents the centre of the inner rectangle and a=Height / 2 of rectangle b= Width / 2 of rectangle The area computed from the above method is the Region of interest. The results obtained from above methods are shown in figure I, II and III.

Table II: Compression percentage obtained for all images. % Compression First Image Second Image

Full Image

ROI

Non ROI

47.54

59.64

64.37

47.81

58.29

63.07

III. CONCLUSION In this paper we proposed a novel method to extract the region of interest and use the Biorthogonal spline wavelet transform for compression. The results obtained are promising with higher compression for Non ROI based images and compression at decomposition level of one for the ROI. However it is interesting to note that the overall compression of the split images is better than that achieved for the original image.

Figure I : The images used to test the proposed method

REFERENCES [1]

Figure II : The ROI obtained through the proposed method. [2]

[3]

[4]

Figure III : The non region of interest. [5]

The above images were compressed using Biorthogonal spline wavelet. The application built in Matlab accepts the input medical image to be compressed and displays the compressed and reconstructed image along with the compression ratio and retained energy. The Matlab program was designed to handle various low and high level decomposition levels. In this paper the investigations on the various decomposition and reconstruction levels used are tabulated in table I and the results obtained is tabulated in table II.


[6]

[7]

[8]

[9]

53

S.Van Assche,, D. De Rycke, W. Philips, and I. Lemahieu, “Exploiting interframe redundancies in the lossless compression of 3D medical images,”Data Compression Conference, 2000, Page(s): 575. Bullmore, E., J. Fadili, V. Maxim, L. Sendur, J. Suckling, B. Whitcher, M. Brammer and M. Breakspear, 2004. Wavelets and Functional Magnetic Resonance Imaging of the Human Brain. NeuroImage, 23(1): 234-249. Sonja Grgic, Mislov Grgic and Branka Zorkocihlar,“Performance Analysis of Image Compression Using Wavelets” IEEE Trans on Industrial Electronics, Vol 48, No 3, June 2001. Rafael C. Gonzalez and Richard E. Woods. Digital Image Processing. Pearson Education, Englewood Cliffs, 2002. Pennebaker W. B., Mitchell J. L. et al., JPEG-Still Image Data Compression Standard. New York: Van Nostrand Reinhold, 1993, pp.65-79 Ho B.K.T., Tseng V., Ma M., and Chen D. A mathematical model to quantify JPEG block artifacts. Proc. SPIE vol. 1897 pp.269-274, 1993 Goldberg M. A., Pivovarov M. et al., “Application of wavelet compression to digitized radiographs”, AJR:163, pp.463-468, 1994. Shapiro J.M.,, “Embedded image coding using zerotrees of wavelet coefficients”, IEEE Trans. Signal Processing, vol. 41, no. 12, pp. 3445-3462, 1993 Vetterli M. and Herley C., “Wavelets and filter banks: Theory and Design”, IEEE Trans. Signal Processing, vol. 40, no. 9, pp. 2207-2232, 1992.


spline wavelets. Appl. Comput. Harmon. Anal., 2:392– 397, 1995. [14] A. Cohen and I. Daubechies. A stability criterion for biorthogonal wavelet bases and their related subband coding scheme. Duke Math. J., 68(2):313–335, 1992. [15] I. Daubechies. Orthonormal bases of compactly supported wavelets. Comm.Pure Appl. Math., 41(7):909–996, 1988.

[10] Castleman K.R., Digital Image Processing. Prentice

Hall,Englewood Cliffs, New Jersey, 1996. [11] Lewis A.S. and Knowles G., “Image compression using the 2D wavelet transform”, IEEE Trans. Image Processing, 1, pp. 244-250, 1992. [12] Rioul O. and Vetterli M., “Wavelets and SignalProcessing”, IEEE Signal Processing Magazine, Vol. 8, pp.85-107, 1991. [13] T. Berger and O. Str¨omberg. Exact reconstruction algorithms for the discrete wavelet transform using


54


Medical Image Retrieval System Using Gain based Feature Extraction Sasi Kumar M, Research scholar, Dept. of CSE, Sathyabama University, Chennai, India

Dr.Y.S.Kumarswamy, Sr.Professor, HoD-MCA, Dayananda Sagar College of Engg. Bangalore, India Abstract: Medical image classification based on its attributes is crucial for storing and efficient retrieval from a Medical image archive. Modern medical imaging systems like Magnetic Resonance Imaging, Computer Tomography, Ultrasound techniques produce large volumes of data and efficient management of the same is challenging. Medical images generated from various devices plays a crucial role in the entire clinical process, be it diagnosis or treatment planning. A large hospital produces gigabits of image data every month. Effective archiving and efficient retrieval plays an important role for good utilization of the medical image archive. In this paper we propose a novel feature selection mechanism using the Discrete sine transform with information gain to investigate the classification accuracy of decision tree induction algorithms for the medical image retrieval problem. The proposed method is compared with the normal method of feature extraction using discrete sine transform. Results obtained from the proposed method has better classification accuracy and prove the proposed methodology is effective for medical image retrieval system.

Since medical images are complex in nature they require extensive image processing techniques for computer based diagnoses. Other activities that are extensively done in medical images are the removal of noise, identifying region of interest. Other applications of medical images include research and continuing education.

Image retrieval plays an important role for handling large amount of visual information in medical applications [7]. Three important factors need to be considered for effective image retrieval systems[8]. (a) The extraction of information from images to a multi dimensional feature vector, (b) Computation of distance metrics in a quantifiable manner and (c) Identify the images in database with lowest distance metrics from the query image. Descriptors for image content include histogram, color, texture, shape and spatial relationship. In medical Keywords— Image retrieval, Discrete Sine Transform, imaging color has been used effectively in dermatology[9]. Since most of the MRI images Random forest, BF tree, LAD tree captured is in greyscale, color may not be an effective I. INTRODUCTION method to solve the MRI medical image retrieval Since early 80's image database management was problem. Most of the work in image retrieval is based on an active area of research and older image retrieval systems extensively relied on textual annotation of similarity measures computed from low level image images[1][2]. Though early work did not concentrate features. This has motivated researchers to focus on on visual features of the image or on feature extraction, the utilization of data mining techniques based on images were organized by semantic queries. Since it is decision tree, Neural network, Bayesian network[10] difficult to generate descriptive texts automatically, and support vector machine classifiers. In this paper it is proposed to extract the frequency the process of annotation of images were an expensive vector from medical images such that the extracted task. energy is located atleast one pixel length from the next In the 90's digital imaging became popular and the co efficient using discrete sine transform. The amount of digital images produced increased and information gain obtained from these vectors are hence the need for management of digital images computed with respect to the dependent class variable became acute. This need drove researchers to to further reduce the feature vectors. Decision tree investigate on visual features of images for induction algorithms are used to classify the obtained development of automatic image classification and feature vectors for the given class. This paper is organized into the following sections, retrieval systems based on input query images. Early work concentrated on using visual features with text section II briefly introduces the Proposed Discrete Sine Transform, Section III describes the classification annotation[3][4][5] for the image retrieval problem. algorithm used in this paper, Section IV gives in detail Medical image retrieval is gaining wide interest due our experimental setup and section V discusses the to the creation of large numbers of medical images in result obtained. digital formats from various medical devices. Medical II. FEATURE VECTOR EXTRACTION image guided diagnoses has become the order of the day for most health care professionals. Medical The feature vector from each image was extracted images are by nature complex in nature and is using the discrete sine transform. Pixels which are one produced from various diverse sources like length away from each other are selected. The Computerized tomography(CT), Positron emission algorithm pseudo is given below: tomography(PET), X-Ray, Digital microscopes, I (Y ; X ) H (Y ) H (Y | X ) Magnetic resonance imaging(MRI) to name a few[6]. Sathyabama University

55


Random forest[11] is an ensemble of classification trees. To classify an object from an input vector, the input vector is classified in each of the trees in the forest. The forest chooses the classification having the most votes for that class. In random forest each tree is constructed using different boot strap sample and over one third of the cases are not used in the bootstrap sample and hence not used for the construction of the kth tree.

j l

H (Y | X )

P( X

x j ) H (Y | X

xj )

j 1

1. 2.

Compute Image size MxN for each alternate value 'i' in array M and array size less than M or M+1 for each alternate value 'j' in array N and array size less than N or N+1 compute DST(array[xi,yj]) Store computed value in one dimensional array Repeat from step 1 till all images are computed

3. 4. 5. 6.

The programming logic for random forest is given below 1. 2. 3.

The discrete sine transform (DST) is similar to the discrete Fourier transform (DFT) but with the difference of using only the real numbers. The discrete sine transform is represented by Sk

pk

N 1

pK

n 0

2

(n xn sin

1 )(k 1) 2 k N

4. 5.

0,1, 2,.........., N 1

6. 7.

k ,0

The BF tree[12] is a decision tree algorithm wherein the most important attribute is expanded first. The best attribute is first selected by computing Gini index and selection of attribute which leads to reduction of impurity to its maximum. The gini impurity measure that is to be computed at node ‘t’ is given by

N

Where xn is the original vector on N real numbers. is the Kronecker delta. DST operates on real data with odd symmetry and hence the output data are shifted by half a sample. The inverse of Discrete sine transform is given by

III

Sk

Input the number of training set N. Compute the number of attributes M For m input attributes used to form the decision at a node m25kHz Self-impedance :>1MU Capacitance dependence of the sensitivity : 0.06Mv/g(.OC) Operating- temperature range : -40…+150 C ) Permissible oscillation Sustained: 80g Short- term 400g

VI. Problems to be Tackled a) Concealing If all is set up well the main aim of the product is to prevent or alert theft of the two-wheeler. The sensor has to be strategically placed to detect force full violations on the motor vehicle. The camera of the cell phone has to be so placed that the user can get a clear view of the fusion region. The camera should also be hidden from the naked eye and in the event of damage to it alarm would be triggered and the user would be intimated of their motor vehicle’s position. Use of “one way mirror” can solve the issue of hiding the camera. A one way mirror has a reflective coating applied on a very this spare layer so thin that it is called a half silvered surface. The many half-silvered comes from the fact that the reflective molecules coat the glass so sparsely that only about half the molecules needed to make the glass an opaque mirror are applied. At the molecular


192


VIII. REFERENCES

level, there are reflective molecules speckled all over the glass in an even film but only half of the glass is covered. (5) The half-silvered surface will reflect about half the light that strike its surface, while letting the other half go straight through. Placing the camera lends behind the two way mirror perhaps near the front panel of the motor vehicle or any space preferred by the user.

1.

How stuff works, Wikipedia sensor manufacture

2.

“Mobile Communications Beyond 3G in the Global Context”

3.

“Next Generation Network Evolution:. WIRELESS NETWOPRKS

4.

The Draft IEEE 802. 16m System Description Document.

5.

David W.Carman, Peter S.Kruus, and Brian J.Matt. Constraints and approaches for distributed sen-sor network security. NAI Labs Technical Report#00010,September2000.

6.

D.W.Craman, P.S Kruus and B.J.Matt, Constraints and approaches for distributed sensor network security, NAILabs Technical Report No.00-010(2010)

b) Power Supply The power supply for the operation of the whole security system is not high. The vibration sensor and the microcontroller work on negligible power. It is the cell phone which requires the bulk of supply.[6] The motor vehicles battery can be used to charge the cell phone when it is going dry. c) Multiple Cameras: We can install multiple cameras and get better view of the motor vehicle surrounding. VII. CONCLUSION A lot of variations and modifications can be carried out in the proposed system for example multiple cameras, auto dial to police numbers etc.


193


Differentiated services network with Dynamic admission control algorithm Hemlata Pal (PG student) [email protected] Department of Electronics and Telecommunication Shri G. S. Institute Of Science & Technology, 23 Park Road, Indore, India the header of IP packets for packet classification purposes. The differentiated services framework enables quality-of-service provisioning within a ABSTRACT: Differentiated service network is used network domain by applying rules at the edges to to provide QoS (Quality of service) in an IP network create traffic aggregates and coupling each of these with the help of Dynamic admission control with a specific forwarding path treatment in the algorithm. Bandwidth broker, an important part of domain through use of a code point in the IP header.In figure.1 the shaded region shows basic Diff-serv network, performing all the controlling Diff –serv netwok. function of Diff-serv network. Admission control module is an integral part of bandwidth broker. The aim of admission control algorithm is to achieve good quality of service by increasing network bandwidth utilization and efficient resources allocation. NS2 simulator is being used to perform the task of achieving QoS in Differentiated service network with Dynamic admission control algorithm. I . INTRODUCTION Figure 1. Diff-Serve network

As the internet grows into commercial and global infrastructure, so that it can support many types of application such as network multimedia, VOIP, real time application and on demand media streaming. This application required to be transfer huge amount of data at a time with good quality of service. IP is unreliable protocol and it provides Best efforts delivery services. Because of these reasons IP itself is unable to fulfill the requirement of these applications. For supporting IP, one type of well known model is the Diff-serv Model.

DiffServ has only data plane .Bandwidth broker is needed for performing the task of controlling throughout the network. Bandwidth broker is basically a agent which act as the control plane for the DiffServ. DiffServ itself can’t achieve end to end quality of service, so that bandwidth broker is most important part for DiffServ. Only because of that Bandwidth broker proper resource management is possible. Bandwidth broker contain admission control module and in that admission control module admission control algorithm is defined. On the basis of that algorithm Bandwidth broker decides the incoming flow request is admitted or rejected.

Differentiated Services or DiffServ is networking architecture that specifies a simple and scalable mechanism for classifying, managing network traffic and providing Quality of Service (QoS) guarantees on modern IP networks DiffServ is a class-based mechanism for traffic management. Class-based mechanism means that in Diff-serv network incoming data packets are classified into different types of classes. The differential treatment is provided to each type of classes.. DiffServ can be used to provide lowlatency to critical network traffic such as voice or video while providing simple best-effort traffic guarantees to non-critical services such as web traffic or file transfers. DiffServ uses the 6-bit Differentiated Services Code Point (DSCP) field in


II . BANDWIDTH BROKER Bandwidth Broker is an agent that has some knowledge of an organization's priorities and policies and allocates bandwidth with respect to those policies. In order to achieve an end-to-end allocation of resources across separate domains, the Bandwidth Broker managing a domain will have to communicate with its adjacent peers, which allows end-to-end services to be constructed out of purely bilateral agreements. Bandwidth Brokers can be configured with organizational policies, keep track of 194


the current allocation of marked traffic, and interpret new requests to mark traffic in light of the policies and current allocation.

request otherwise request is denied. Figure.3. shows Diff-serv network with Distributed Bandwidth Broker.

When a flow request arrives, first it goes to the Bandwidth broker. Bandwidth broker has all the data about resources availability, path availability and link availability throughout the network. Bandwidth broker sent it to the admission control module where the admission control algorithm is defined. Admission control module checks the network stats and takes the decisions. If resources are available then it accepts the flow request otherwise it is rejected. There are two architecture are available for Bandwidth broker, one is the Centralized Bandwidth broker modal and other one is the Distributed Bandwidth broker modal.

CB – Central Bandwidth broker eB – Edge Bandwidth Broker

Figure 3. Distributed Bandwidth Broker

In Centralized Bandwidth broker modal only single one Bandwidth broker is used for a particular network domain. In Centralized Bandwidth broker modal, each resource reservation and provisioning and also admission control decisions are taken by single one Bandwidth broker. This single one Bandwidth broker is responsible for each and every decision making. Figure.2 shows Diff-serv network with Centralized Bandwidth Broker.

III . ADMISSION CONTROL Admission control is basically a criteria or set of rules by which we can decide the flow request is to be accepted or rejected. There are two types of admission control are possible, one is the static and another one is dynamic. In case of Static admission control, bandwidth allocation for each path and links are predefined in the static manner. When the bandwidth required by flow request is more than available at the path, but in this case bandwidth allocation is not changes even if there is unused bandwidth is available at some other path. In case of static admission control waste of bandwidth occur. There are two types of static admission control are algorithms are explained. PBAC (Parameter based admission control)-In this PBAC method, all the admission decisions for any incoming data stream are taken on the basis of parameters of the network. This is based on only certain traffic behavior assumption. If the incoming user request is not one that is taken into assumption, then it will not wo rk properly.

Figure 2. Centralised bandwidth broker In the hierarchical Distributed Bandwidth broker modal two types of bandwidth broker are used, one is the central bandwidth broker, other one is the edge bandwidth broker. In which responsibilities are divided into two groups for two type of bandwidth broker. When a flow request arrives first it goes to the edge router then it forwards it to the edge bandwidth broker. Edge bandwidth broker made admiddetability test and it checks network stats for availability of bandwidth and resources. If the bandwidth required by flow request is the available then it response to the flow request. If the bandwidth required by flow request is greater than the available bandwidth, then it request the chunk of bandwidth from Central bandwidth broker. If required chunk of bandwidth is unused or free on other path, then Central bandwidth broker provide it to the flow


MBAC (Measurement based admission control) - In this MBAC method, all the admission decisions and resource reservation and provisioning are based on the real time measurement of traffic. In which before taking any type decision first current situation of traffic is measured and on the basis of that decisions are taken. Bandwidth broker periodically measured link lode and bandwidth reservation is done on the basis of that periodically measured link load. If link load is below some specified threshold value, then user request is accepted otherwise it will be rejected. There are many drawback of static

195


algorithm such as waste of bandwidth and resources. To overcome this disadvantages we are using Dynamic admission control algorithm.

in which e1,e2 andC are Edge and Core routers between source and destination . At these routers differential treatment is provided to the incoming flow request. In this sinerio we are taken the simulation time is 100s , packet size of 1500 and link delay is 10ms .The policer type is Token bucket and for this policer type the value of CIR(committed information rate) is 1000000 and CBS(Committed burst size) is 10000 The bandwidth between S1-e1, e1-C ,and e2-D is 10 Mb and between C-e2 is 5 Mb.

IV . DYNAMIC ADMISSION CONTROLDynamic admission control algorithm uses the concept of Resource sharing pool. In this Resource sharing pool concept, when user request arrive then bandwidth broker not only checks the free or remaining available bandwidth and resources on the network but also checks the reserve bandwidth and resources on the network. If reserve bandwidth and resources are not use at peak level, some bandwidth and resources are unused, then bandwidth broker also taken into account these unused resources and bandwidth with the remaining available bandwidth and resources on the network for fulfill the requirement of newly arrive user request. Dynamic admission control mechanism is a two phase mechanism, path level admission control and link level admission control. Path level admission control is performed by bandwidth broker. Bandwidth broker perform path level admission control at the edge router. Bandwidth broker simply calculates bandwidth and provide it to the edge router. Edge router itself performs path level admission control. If the bandwidth required by user request is within the bandwidth of that path, then request is accepted. If the bandwidth required by user request is greater than the bandwidth of that path, then edge router can’t take any decision .At that time link level admission control is required. Link level admission control is performed on the basis of Dynamic admission control algorithm. If the path bandwidth is insufficient for user request then we perform link level admission control. In which case we takes the decision dynamically that means we checks the unused bandwidth on other path ,if available ,then it provide it to requested path to complete user request. After that network stats are updated .

Figure 4. Simulation topology for IP network

Figure 5. Simulation topology for Diff-Serv network

VI . EXPERIMENTAL RESULTS This paper simulates IP network without Diffserv network and IP network with Diffserv network by NS2 software . Table I shows the PHB table for the Diff-serv network. Table 1. Experimental results of Diff-serve network TABLE I. Experimental results of Diffserv network DScP

V . SIMULATION AND DISCUSSION The simulation of Diff-serv network is performed by using NS2 simulator. Experimental topology for IP network is shown in figure.4 in which n1,n2 and n3 are intermediate node between source and destination. At the intermediate node data is simply passes through node, there is no differential treatment is provided to the incoming flow request. In this sinerio we are taken the simulation time is 100s , packet size of 1500 and link delay is 10ms . The bandwidth between S1-n1, n1-n2 ,and n3-D is 10 Mb and between n2-n3 is 5 Mb Experimental topology for Diff -servnetwork is shown in figure.5


TotPkts

TxtPkts

Ldrops

Edrops

All

29991

22585

5886

1567

10

7457

7436

21

0

11

7402

3788

2844

770

20

7503

7466

37

0

21

7629

3848

2984

797

The graph shows the comparative analysis of IP network without Diff-serv and IP network with Diffserv for three parameters are as Troughput, Delay and Jitter.

196


network with the IP network .But when we are see the PHB table for Diff-serv network we can analyze that value of packet loss rate is higher because Diffserv network itself can’t achieve end to end QoS for IP network. When we include a effective admission control algorithm with Diff- serv network, then Packet loss rate, Delay. Jitter and Throughput values are improved.

VIII . REFERENCES

[1]

S. Blake, D. Black, M. Carlson, E. Davies, Z. Wang, and W. Weiss, An Architecture for Differentiated Service, IETF RFC2475, 1998.

[2]

Z. Zhang, Z. Duan, L. Gao, and Y. Hou, “Decoupling QoS Control From Core Router: a Novel Bandwidth Broker Architecture for Scalable Support of Guaranteed Service,” in Proceeding of ACM SIGCOMM’00, Stockholm, Sweden, 2000.

[3]

S. H. Jeong, H. Owen, J. Copeland, and J. Sokol, “QoS Support

for

UDP/TCP

based

Networks,”

Computer

Communications, vol. 24, pp.64–77, 2001. [4]

J. Qiu and E. W. Knightly, “Measurement-based admission control with aggregate traffic envelopes,” IEEE/ACM Transactions on Networking,vol. 9, 2001. [5] S. Chen and K. Nahrstedt, “An Overview of Quality of Service Routingfor Next-Generation High-Speed Networks: Problems and Solutions,”IEEE Networks, vol. Nov/Dec, pp. 64–79, 1998.

[6]

C. Bouras and K. Stamos. “An Adaptive Admission Control Algorithm for Bandwidth Broker”. Cambridge: 3rd IEEE International Symposium on Network Computing and Applications (NCA04), pp. 243-250, 2004.

VII . CONCLUSION AND FUTURE WORK The graphs shows that Throughput, Delay and Jitter values are improved when we are using the Dff-serv


197


Data Security in Wireless Sensor Networks Using A star topology Approach Sajana.Tiruveedhula# #

1

Department of computer science & engineering, JNTU-KAKINADA University # M.tech, (09F01D5814) St.Ann’s college of engineering &Technology, SACET, Chirala, Andhra Pradesh (A.P) [email protected]

Abstract: Security has become one of the major issues for data communication over Wired and Wireless sensor networks. Consider wireless sensor networks while transferring the data it can be attacked by different kinds of attacks such as compromised node, denial of service attacks. We agree that existing multi-path routing approaches are vulnerable to such attacks, mainly due to their deterministic nature. By considering the routing algorithm, it can compute the same routes known to the source, therefore all information sent over these routes. To secure the data from these attacks we can generate randomized multipath routes. These routes are taken by the shares of different packets change over time. While sending the data the randomized multi path routes include the same routes. In this paper, “instead of selecting self nodes “we generate dynamic multi path routing in which the shares of different packets are taken and send to the destination. We develop efficient topology for delivery the data in secure manner. The experimental results show that topology out performs traditional schemes in terms of CPU cost, minimization of retransmissions. Keywords— Randomized multi path routing, Star Topology, Message transmission, Data security, wireless sensor networks.

I. INTRODUCTION In wireless sensor network (WAN) while transferring the data it can be attacked by different kinds of attacks. In this paper we are specifically interested in Combating two types of attacks: Compromised node (CN) and denial of service (DOS). In the CN attack, an adversary physically compromises a subset of nodes to eavesdrop information, whereas in the DOS attack, the adversary interferes with the normal operation of the network by actively disrupting, changing, or even paralysing the functionality of a subset of nodes. These two attacks are similar in the sense that they both generate black holes: areas within which the adversary can either passively intercept or actively block information delivery. Due to the unattended nature of WSN’s, adversaries can easily produce such black holes. Severe CN and DOS attacks can disrupt normal data delivery between sensor nodes and the sink, or even partition the topology. A conventional cryptography-based security method cannot alone provide satisfactory solutions to these problems. This is because, by definition, once a node is compromised, the adversary can always acquire the encryption/decryption keys of that node,


and thus can intercept any information passed through it. Likewise, an adversary can always perform DOS attacks (e.g., jamming) even if it does not have any knowledge of the underlying cryptosystem. One remedial solution to these attacks is to exploit the network’s routing functionality. Specifically, if the locations of the black holes are known a priori, then data can be delivered over paths that circumvent (bypass) these holes, whenever possible. In practice, due to the difficulty of acquiring such location information, the above idea is implemented in a probabilistic manner, typically through a Two-step process. First, the packet is broken into M shares (i.e., components of a packet that carry partial information) using a (T, M) threshold secret sharing mechanism such as the Shamir’s algorithm. The original information can be recovered from a combination of at least T shares, but no information can be guessed from less than T shares. Second, multiple routes from the source to the destination are computed according to some multipath routing algorithm. These routes are node-disjoint or maximally node-disjoint subject to certain constraints (e.g., min-hop routes). The M shares are then distributed over these routes and delivered to the destination. As long as at least (M T + 1) or T shares bypass the compromised (or jammed) nodes, the adversary cannot acquire (or deny the delivery of) the original packet. II.BASIC IDEA In this paper, we propose a randomized multipath routing algorithm that can overcome the above problems. In this algorithm, multiple paths are computed in a randomized way each time an information packet needs to be sent, such that the set of routes taken by various shares of different packets keep changing over time. As a result, a large number of routes can be potentially generated for each source and destination. To intercept different packets, the adversary has to compromise or jam all possible routes from the source to the destination, which is practically not possible. Because routes are now randomly generated, they may no longer be nodedisjoint. However, the algorithm ensures that the randomly generated routes are as dispersive as possible, i.e., the routes are geographically separated as far as possible such that they have high likelihood of not simultaneously passing through a black hole.

198


III. POSSIBILITIES

minimize the end-to-end energy consumption under a given security constraint. 3. We conduct extensive simulations to study the performance of the proposed schemes under more realistic settings. Our simulation results are used to verify the effectiveness of our design. When the parameters are appropriately set, all four randomized schemes are shown to provide better security Performance at a reasonable energy cost than their deterministic counterparts. At the same time, they do not suffer from the type of attacks faced by deterministic multipath routing.

The main challenge in our design is to generate highly dispersive random routes at low energy cost. A naive algorithm of generating random routes, such as Wanderer scheme (a pure random-walk algorithm), only leads to long paths (containing many hops, and therefore, consuming lots of energy) without achieving good depressiveness. Due to security considerations, we also require that the route computation be implemented in a distributed way, such that the final route represents the aggregate decision of all the nodes participating in the route V.IMPLEMENTATION selection. As a result, a small number of compromised nodes cannot dominate the selection result. In addition, for efficiency purposes, we also require that the randomized route selection algorithm only incurs a small amount of communication overhead. , we also require that the route computation be implemented in a distributed way, such that the final route represents the aggregate decision of all the nodes participating in the route selection. As a result, a small number of colluding/compromised nodes cannot dominate the Fig. 1. Randomized dispersive routing in a WSN. selection result. In addition, for efficiency purposes, we also require that the randomized route selection As illustrated in the above fig. we consider a threealgorithm only incurs a small amount of phase approach for secure information delivery in a communication overhead. WSN: secret sharing of Information, randomized propagation of each information share, and normal IV.REQUIREMENTS AND SPECIFICATIONS routing (e.g., min-hop routing) toward the sink. More specifically, when a sensor node wants to send a 1. We explore the potential of random packet to the sink, it first breaks the packet into M dispersion for information delivery in WSNs. shares, according to a (T; M) threshold secret sharing Depending on the type of information available to a Algorithm. Each share is then transmitted to some sensor; we develop four distributed schemes for randomly selected neighbour. That neighbour will propagating information “shares”: purely random continue to relay the share it has received to other propagation (PRP), directed random propagation randomly selected neighbours, and so on. In each (DRP), non repetitive random propagation (NRRP), share, there is a TTL field, whose initial value is set by and multicast tree the source node to control the total number of random Assisted random propagation (MTRP). PRP utilizes relays. After each relay, the TTL field is reduced by 1. only one-hop neighbourhood information and provides When the TTL value reaches 0, the last node to baseline performance. DRP utilizes two-hop receive this share begins to route it toward the sink neighbourhood information to improve the using min-hop routing. propagation efficiency, leading to a smaller packet Once the sink collects at least T shares, it can interception probability. The NRRP scheme achieves a reconstruct the original packet. No information can be similar effect, but in a different way: it records all recovered from less traversed nodes to avoid traversing them again in the than T shares. future. MTRP tries to propagate shares in the direction of the sink, making the delivery process more energy Random Propagation of Information Shares efficient. To diversify routes, an ideal random propagation 2. We theoretically evaluate the goodness of these algorithm would propagate shares as dispersively as dispersive routes in terms of avoiding black holes. We possible. Typically, this means propagating the shares conduct asymptotic analysis (i.e., assuming an infinite farther from their source. At the same time, it is highly number of nodes) for the worst-case packet desirable to have an energy-efficient propagation, interception probability and energy efficiency under which calls for limiting the number of randomly the baseline PRP scheme. Our results can be propagated hops. The challenge here lies in the interpreted as the performance limit of PRP, and a random and distributed nature of the propagation low-bound on the performance of the more advanced share may be sent one hop farther from its source in DRP, NRRP, and MTRP schemes. Our analysis helps given step, but may be sent back closer to the source us better to understand how security is achieved under in the next step, wasting both steps from a security dispersive routing. Based on this analysis; we standpoint. To Tackle this issue, some control needs to investigate the trade-off between the random be imposed on the random propagation process. propagation parameter and the secret sharing 1 Purely Random Propagation (Baseline Scheme) parameter. We further optimize these parameters to Sathyabama University

199


In PRP, shares are propagated based on one-hop neighbour-hood information. More specifically, a sensor node maintains a neighbor list, which contains the ids of all nodes within its transmission range. When a source node wants to send shares to the sink, it includes a TTL of initial value N in each share. It then randomly selects a neighbor for each share, and unicasts the share to that neighbor. After receiving the share, the neighbor first decrements the TTL. If the new TTL is greater than 0, the neighbor randomly picks a node from its Neighbor list (this node cannot be the source node) and relays the share to it, and so on. When the TTL reaches 0, the final node receiving this share stops the random propagation of

forth by eliminating this type of propagation within any two consecutive steps. Compared with PRP, DRP attempts to push a share outward away from the source, and thus, leads to better propagation efficiency for a given TTL value. 4 . Multicast Tree-Assisted Random Propagation MTRP aims at actively improving the energy efficiency of random propagation while preserving the dispersiveness of DRP. The basic idea comes from the following observation of Fig. 1: Among the three different routes taken by shares, the route on the bottom right is the most energy efficient because it is the shortest end-to-end path. So, in order to improve energy efficiency, shares should be best propagated in the direction of the sink. In other words, their propagation should be restricted to the right half of the circle in Fig. 1. Conventionally, directional routing requires location information of both the source and the destination nodes, and sometimes of intermediate nodes. Examples of location based routing are the Greedy Perimeter Stateless Routing ASYMPTOTIC ANALYSIS OF THE PRP SCHEME

The random routes generated by the four algorithms in Sections 2 are not necessarily node-disjoint. So, a natural Question is how good these routes are in avoiding black holes. We answer this question by conducting asymptotic 2 Nonrepetitive Random Propagation analysis of the PRP scheme. Theoretically, such NRRP is based on PRP, but it improves the analysis can be Interpreted as an approximation of the propagation efficiency by recording the nodes performance when the node density is sufficiently traversed so far. Specifically, NRRP adds a “node-in- large. It also serves as a lower bound on the route” (NIR) field to the header of each share. Initially, performance of the NRRP, DRP, and MTRP schemes. this field is empty. Starting from the source node, Note that the security analysis for the CN and DOS whenever a node propagates the share to the next hop, attacks are similar because both of them involve the id of the upstream node is appended to the NIR calculating the packet interception probability. For field. Nodes included in NIR are excluded from the brevity, we only focus on the CN attack model. The random pick at the next hop. This nonrepetitive same treatment can be applied to the DOS attack with propagation guarantees that the share will be relayed a straightforward modification to a different node in each step of random propagation, leading to better propagation efficiency. Fig2. Implication of route dispersiveness on bypassing the black hole. (a) Routes of higher dispersiveness. (b) Routes of lower dispersiveness.

3. Directed Random Propagation DRP improves the propagation efficiency by using two-hop neighbourhood information. More specifically, DRP adds a “last-hop neighbor list” (LHNL) field to the header of each Share. Before a share is propagated to the next node, the relaying node first updates the LHNL field with its neighbor list. When the next node receives the share, it compares the LHNL field against its own neighbor list, and randomly picks one node from its neighbors that are not in the LHNL. It then decrements the TTL value, updates the LHNL field, and relays the share to the next hop, and so on. Whenever the LHNL fully overlaps with or contains the relaying node’s neighbor list, a random neighbor is selected, just as in the case of the PRP scheme. According to this propagation method, DRP reduces the chance of propagating a share back and


Fig.3. Packet interception area, a six-hop random propagation example For a given source sensor node, the security provided by the protocol is defined as the worst-case (maximum) probability that for the M shares of an information packet sent from the source, at least T of them are intercepted by the black hole. Mathematically, this is

200


defined as follows: Let the distance between the source s and the sink o be d s. 5.Energy Efficiency of the Random Propagation We assume that the energy consumption for delivering one bit over one hop is a constant q. Then, the average energy consumption for delivering one packet from source s to sink o depends on the average length (in hops) of the route. Note that each random route consists of two components. The first is a fixed N-hop component attributed to the random .

Fig 4: The total transmission distance after random propagation asymptotic assumption, when min-hop routing is used, the ratio between the number of hops from w -> o and from s -> o can be approximated by the ratio of the lengths of these two paths. This ratio can be calculated as follows. Suppose w is located in the ith ring (see Fig. 10). Let the distance between w and s be (i-1) Rh ¼ = T(Ri[B(1)]

a 1 , ifk f

1

(2. 3)

rk , a k 1 ; otherwise

Then, find rn according to b and a

1

by

As a result, {R1, R2, ..., Rn} produced in this way is a rn = f (b, a 1). (2. 4) set of two out of n visual cryptograms of random grids. Yet, it is not a set of VCRG-n, since Condition 2 in It is noticed that formula (2. 2) is a special case definition I fail. of formula (2. 4) by setting n = 2. After all rn's Rn Let B = {0, 1} be a binary set. Now introducing a corresponding to all b's B are computed, which function f : B × B B defined as follows: obtain Rn. Then, E = {R1, R2, ..., Rn} is reported as a set of VCRG-n of B. Sathyabama University

264


IV. HALFTONE TECHNOLOGY Halftoning or analog halftoning is a process that simulates shades of gray by varying the size of tiny black dots arranged in a regular pattern. Halftoning is a technique of image reproduction for devices with a limited range of tone levels (usually bi-level). With halftoning, an image is produced with a series of dots printed on the paper. If the dots are small enough, the eye cannot detect individual dot patterns; instead it integrates halftone dots and unprinted areas as varying shades. The varying intensities of the black dots produce a simulation of continuous tone image. Digital halftoning started around the 1920s and was used to display images on bi-level devices and reduce the transmission bandwidth. The printing industry adopted digital halftoning after the introduction of the electronic proofing and desktop publication. Digital halftoning is similar to halftoning in which an image is decomposed into a grid of halftone cells. Elements (or dots that halftoning uses in simulates shades of grays) of an image are simulated by filling the appropriate halftone cells. The more number of black dots in a halftone cell, the darker the cell appears. For example, in Figure 1 a tiny dot located at the center is simulated in digital halftoning by filling the center halftone cell; likewise, a medium size dot located at the top-left corner is simulated by filling the four cells at the top-left corner.

Fig 1.a

Fig 1.b

Fig 1.b) Digital halftoning

The large dot covering most of the area in the third image is simulated by filling all halftone cells. A. Error Diffusion Error diffusion is a major stochastic halftoning approach. It is often called spatial dithering. It is an adaptive algorithm that uses the threshold error feedback to produce patterns with different spatial frequency content, depending on the input values. Error diffusion sequentially traverses each pixel of the source image. Each pixel is compared to a threshold. If the pixel value is higher than the threshold, a 255 is outputted; otherwise, a 0 is outputted. Error diffusion is a neighborhood operation since it operates not only on the input pixel, but also its neighbors. Generally, neighborhood operations produce higher quality results than point operations. The process consists of a single pass over the input image: each pixel is processed sequentially for a binary thresholding, and Sathyabama University

The following Figure 2 depicts the schematic diagram and the error filter of error diffusion. As each pixel is corrected with the errors from its preceding neighbors, its value will change from its original value. This correct value, not the original input value, is then thresholded.

Figure 2 The process and error filter of error-diffusion (p denotes the current pixel)

Figure 1: Sample of halftoning Fig 1.a) Halftoning

the error between output and modified input values is computed. This error is diffused into yet to be considered input values as governed by an error filter (also referred to as an error matrix, diffusion matrix, or kernel). In this filter, the p represents the pixel you are currently scanning, and the numbers (called weights, for equally boring reasons) represent the proportion of the error distributed to the pixel in that position. Here, the pixel immediately to the right gets 7/16 of the error (the divisor is 16 because the weights add to 16), the pixel directly below gets 5/16 of the error, and the diagonally adjacent pixels get 3/16 and 1/16 as shown in figure 3.2. When scanning a line right-to-left, this pattern is reversed. This pattern was chosen carefully so that it would produce a checkerboard pattern in areas with intensity of 1/2. It is also fairly easy to calculate when the division by 16 is replaced by shifts.

Error diffusion forces total tone content to remain the same and attempts to localize the distribution of tone levels. Several larger error filters have been proposed that appear to create better looking images, such as minimized average error filter and stuki error filter. However upon inspection, they look better because the filter tends to sharpen more areas of flat gray are in fact less homogeneous than the original 4-element filter. The pseudo code for error diffusion algorithm as follows: B. Error Diffusion Algorithm Input: An H x W gray-level image, where p(x, y) denotes a pixel value of the position (x, y). Output: A halftone image. for y = 1 to W do for x = 1 to H do if(p(x, y) > 127) then do output = 255; else output = 0; error = p(x, y) – output; p(x+1, y) = p(x+1, y) + error x (7/16);

265


p(x, y+1) = p(x, y+1) + error x (5/16); p(x-1, y+1) = p(x-1, y+1) + error x (3/16); p(x+1, y+1) = p(x+1, y+1) + error x (1/16); p(x, y) = output; end end V. ENCRYPTING A GREY-LEVEL IMAGE There are three encryption algorithms are there to hide an image into random grids. The pseudo codes for the encryption and decryption as follows. A. Encryption Algorithm 1 Encrypting a gray-level image into a set of VCRG-n Input: an h × w gray-level image G and an integer n. Output: a set of n random grids E = {R1, R2, ..., Rn} constituting a VCRG-n of G. 1. Apply Error diffusion algorithm for B(x, y). 2. for (1 k n 1) do { Rk as a random grid, T (Rk) = ½ } 3. for ( each pixel B[i, j], 1 i h and 1 j w) do { a1 = R1 [i, j] for (2 k n 1) do { ak = f (Rk [i, j], a 1) //f (x, s) is defined in formula(2. 1) } Rn [i, j] = f (B[i, j], a 1) } 4. Output (R1, R2, . .., Rn) B. Encryption Algorithm 2 Encrypting a gray-level image into a set of VCRG-n Input: an h × w gray-level image G and an integer n. Output: a set of n random grids E = {R1, R2, ..., Rn} constituting a VCRG-n of G. 1. Apply Error diffusion algorithm for B(x, y). 2. for (1 k n 1) do { generate Rk as a random grid, T (Rk) = ½ } 3. for (each pixel B[i, j], 1 i h and 1 j w) do { a1 = R1[i, j] for (2 k n 1) do { ak = f (Rk[i, j], a 1) } if (B[i, j] = 0) Rn[i, j] = f (B[i, j], a 1) ( = f (0, a 1) = a 1) else Rn[i, j] = random_ pixel( ) } 4. Output (R1, R2, ..., Rn)

Input: an h × w gray-level image G and an integer n Output: a set of n random grids E = {R1, R2, ..., Rn} constituting a VCRG-n of G 1. Apply Error diffusion algorithm for B(x, y). 2. for (1 k n 1) do { generate Rk as a random grid, T (Rk) = ½ } 3. for (each pixel B[i, j], 1 i h and 1 j w) do { a1 = R1[i, j] for (2 k n 1) do { ak = f (Rk[i, j], a 1) } if (B[i, j] = 0) Rn[i, j] = random_ pixel( ) else

4.

1)

(=f (1, a

1)

= a

1)

The decryption algorithm for encryption algorithms 1, 2 and 3 are same. D. Decryption Algorithm Input: random grids R1, R2, . .., and Rn (i.e. encrypted images). Output: a reconstructed image B. 1. Perform OR operation between three generated grids R1, R2, . .., and Rn. i.e. B = R1 R2 ... Rn . 2. Output (Recovered image B) VI. EXPERIMENTAL RESULTS This section gives the results of encryption and decryption algorithms for a secret image.

C. Encryption algorithm 3 Encrypting a gray-level image into a set of VCRG-n Sathyabama University

Rn[i, j] = f (B[i, j], a } Output (R1, R2, . .., Rn)

Figure 3 output results (a) secret image (b) R1 (c) R2 (d) R3 (e) R1 R2 (f) R2 R3 (g) R1 R3 (h, i, j) R1 R2 R3 results for encryption algorithms 1, 2, 3 respectively

266


In this experiment n is taken as 3 so, R1, R2, R3 are the random grids generated for the secret image. Figure 3.e, Figure 3.f and Figure 3.g represents the resultant outputs by two encrypted grids and Figure 3.h, Figure 3.i and Figure 3.j represents the decrypted output by three encrypted grids. By this observation we can get the output if and only if all random grids are superimposed. VII. CONCLUDING REMARKS AND FUTURE SCOPE The security and efficiency of different encryption algorithms using VCRG verified theoretically. These VCRG-n Algorithms are most effective, reliable and maintains lossless recovery of the secret image. Moreover, these encryption algorithms can be easily hardwired by incorporating a 0/1 random number generator with T flip-flops or Exclusive- OR gates. There are many kinds of halftoning, which generate halftone images in different ways. So some effort need whether these halftoning techniques may make any difference on the visual effect of the stacked image.


REFERENCES [1]

O. Kafri, E. Keren, Encryption of pictures and shapes by random grids, Opt. Lett. 12 (1987)377–379.

[2]

Naor, M. and Shamir, A., (1995). “Visual cryptography,” Advances in Cryptology - EUROCRYPT’94, Lecture Notes in Computer Science 950, Springer- Verlag, pp. 1 12.

[3]

S. J. Shyu, Image encryption by random grids, Pattern Recognition 40 (2007)1014–1031.

[4]

C.- N. Yang, C.- S. Laih, New colored visual secret sharing schemes, Des.Codes Cryptogr. 20 (2000) 325–335.

[5]

C. Blundo, A. De Santis, D. R. Stinson, “Improved schemes for visual cryptography”, Des. Codes Cryptogr. 24 (2001) 255– 278.

[6]

S. J. Shyu, Efficient visual secret sharing scheme for color images, Pattern Recognition 39 (2006) 866– 880.

267


Secured Double Data Compression Based On Huffman With Sparse Storage Prof. Biku Abraham1, Prof. Dr. Varghese Paul2 1

M.G University, Kerala, India 1

[email protected]

2

CUSAT, Kerala, India

2

[email protected] information source as accurately as possible using the fewest number of bits. Huffman code is one of the earliest methods of data compression developed by Huffman (1952). It based on how often characters appear. Frequently occurring characters are encoded with short bit strings; characters that occur rarely are encoded with long bit strings. The goal of data compression is then achieved. Bentley et al. (1986) proposed another word based moveto-front (MTF) scheme as a data compression scheme, which exploits locality of reference. The scheme maintains a list of words sorted into least-recently used order. A word is encoded by its position in the dynamically changing list and will have a short code when it is near the front of the list. As result of locality, compression is achieved. The following simple example cited from Bentley et al. (1986) may illustrate the locally adaptive data compression. Data communication and data storage applications have benefited greatly from data compression methodology. By reducing the size of the transmitted data, the effective bandwidth of the communications channel can be increased. The obvious advantage for data storage applications is that smaller data require less storage space. Thus, the effective storage capacity of any storage medium is increased if the data are compressed. There is, however, with current technology, another important implication of data compression on storage technology [19].

Abstract: The most important challenges of data compression are security of text during the transit, speed of encryption and the storage space given the limited memory available in the computer. Results of the study presents an alternative method which needs less storage space with high speed and necessary security aspects required. The method attempts to utilize a two tier coding for the text compression. The first level starts with Huff coding and in the second stage it proceeds to convert the binary coding in a matrix form. The optimal level at which the storage space is minimized is achieved by storing only the positions of binary value “1” which is sparsely distributed in the matrix created out of original Huffman coding . Results of encryption and decryption shows a significant reduction in the file sizes which clearly provide the usefulness and importance of the method applied in the research. Key words - Data compression, Sparse storage, Caesar cipher, Vigenere cipher, Huffman coding.

I.

INTRODUCTION

This work shows lossless compression, search and decompress the data in text format in a computer which has a limited memory. Word-based compression over natural language text has shown to be a good choice to trade compression ratio and speed, obtaining compression ratios close to 30% and very fast decompression. Additionally, it permits fast searches over the compressed text using Boyer-Moore type algorithms. Such compressors are based on processing fixed source symbols (words) and assigning them variable byte-length code words, thus following a fixed-to-variable approach [18]. Nowadays the tremendous growth of text databases as well as data types is increasing continuously. This creates an interest on text compression techniques which reduces the storage space by keeping all the files in compressed form. Many lossy and lossless compression techniques available in the market. There are mainly three categories. They are lossless substitution, statistical and dictionary based data compression techniques. The statistical compression technique based on an estimated probability of characters. An example of this category is Huffman coding. In Huffman coding the most common symbols in the file assigned the shortest binary codes, and the least common assigned the longest codes. Word based compression techniques, plain Huffman and Tagged Huffman were presented later. The main idea of this work is the arrangement of text data to find and exploit patterns, statistics and regularities. The goal of data compression is to represent Sathyabama University

II.

RELATED WORK

Kenji Hamano proposes a new data compression scheme based on T-codes using a dictionary method such that all phrases added to a dictionary have a recursive structure similar to T-codes. This scheme can compress the Calgary Corpus more efficiently than known schemes based on T-codes and the UNIX compress, a variant of LZ78 [2]. Gregory derived lower bounds on rate-distortion performance of LDGM codes over the BES using two different methods and then showed that these lead to the same result. The two methods used are valid to any sparse graph code [3]. Danny presents a technique that compresses a string w by enumerating all the substrings of w. The substrings are enumerated from the shortest to the longest and in lexicographic order. The compression efficiency come close to that of the best PPM variants[4]. In “Adding compression to a full text retrieval system” by justin Zobel et al. proposes a semi static word based compression model, the space needed to store the text is under 30% of the original requirement. The experiments showed that it not only saves space, it can also yield faster

268


word appears n/(i H) times, where H is a constant that makes the frequencies to add up to n.

query processing. The disadvantage of semi static approach is that two passes over the data are required for compression [17].

The symbols in the text are organized in a so called Huffman tree. The Huffman method gives a tree that minimizes the length of the compressed file. Results of the work reveal an alternative method which needs limited storage space with high speed and also security aspects required. The work is based on a two tier coding scheme for the text compression. The first level starts with Huffman coding and in the second stage it takes the text file replaced with Huffman code and stores the binary value as a sparse representation. The optimal level at which the storage space is minimized is by storing only the positions of binary value “1” which is sparsely distributed in the matrix created out of original Huffman coding . Results of encryption and decryption achieved shows a significant reduction in the file sizes which clearly provide the usefulness and importance of the method applied in the research.

In “RLH- Bitmap compression technique based on run-length and Huffman encodings”, by Micha stabno et al. proposes a technique of compressing bit map indexes for application in data ware houses. This technique called run-length Huffman based on run length encoding and Huffman encoding. The experimental evaluation is based on the size of the bitmap indexes, query processing time and update time of bitmap indexes [8].

III.

METHODOLOGY

A.

The compression procedure.

1.

Read the text file character by character.

2.

Apply encryption on the text file to make it more

Shannon defined the entropy H as the average amount of information obtained from the value of a random variable. Suppose we have a random variable X, which may take on the values x1,x2………xn and that the corresponding probabilities of each outcome are P(x1),Px2)…….P(xn). In a sequence of K occurrences of X, the outcome xj will on average be selected KP(xj) times. Therefore the average amount of information obtained from K outcomes is

secure by Caesar cipher shift 5 techniques. 3.

Again apply Vigenere code on the encrypted text.

4.

Apply Huffman coding on encrypted text.

5.

Store the Huffman codes as sparse matrices. That is only 1 bits are storing in the memory.

6.

Now the file is double compressed and secured.

7.

End.

B.

Implementation

KP1log(1/P1)+…………+KPnlog(1/Pn) Dividing by K yields the average amount of information per outcome for the random variable, referred to as the entropy of X, and designated by H(x): H(x) =

Pi log(1/Pi)

IV. We have implemented all the previous improvements and integrate them with the compression algorithm. The proposed compression algorithm has been implemented in C and tested on various files. The results are shown in Table 2. The efficiency of the algorithm depends on the data set. The algorithm reduces the size of the data to less than 30% of the text size.

The Secured Data Compression based on Huffman with Sparse storage (SDCHS) algorithm we presented here based on Huffman and encryption technique aims to improve the compression ratio as well as the security. The results showed that by applying the multistage process of compression we can achieve a very good compression ratio. The very well known programs GZIP and BZIP were selected as industry standard representatives. The GZIP program is based on LZ77 and use same method as very well known program WINZIP. The BZIP program is based on burrows wheeler transformation. The following table shows the coding result of eight symbols after the first stage. This gives entropy of 1.974 and average length is 2.184. After the first stage the efficiency of coding is 0.903. By research statistical method shows better compression than the dictionary methods. We compared the result with many latest compression methods and it gives good compression performance in almost all cases.

To apply encryption in this work, two types of encoding are done. Traditional ceasar cipher and vigenere cipher. Initially apply ceasar cipher shift 5 to the text and then vigenere cipher. For large sized texts, the effective statistical compression technique is Huffman coding. The basic idea of Huffman coding is to assign a unique code word to each different character of text and compression can be achieved by assigning shorter codes to characters with higher frequencies. Compression proceeds in two passes over the text. The encoder makes a first pass over the text to obtain the frequency of each different text word and performs actual compression in a second pass.

TABLE 1 CODING RESULT OF 8 SYMBOLS AFTER FIRST STAGE

An important consideration is the size of the text vocabulary. An empirical law widely accepted in IR is Heap’s law, which states that the vocabulary of a text of n words is of size V=O(n ), where 0< and elements respectively. These elements are passed as the payload to specific WSDL ports that are implemented by security token services. This framework does not define specific actions; each binding defines its own actions. When requesting and returning security tokens additional parameters can be included in requests, or provided in responses to indicate serverdetermined (or used) values. If a requestor specifies a specific value that isn't supported by the recipient, then the recipient MAY fault with a wst:InvalidRequest (or a more specific fault code), or they MAY return a token with their chosen parameters that the requestor may then choose to discard because it doesn't meet their needs. The requesting and returning of security tokens can be used for a variety of purposes. Bindings define how this framework is used for specific usage patterns. Other specifications may define specific bindings and profiles of this mechanism for additional purposes. In general, it is RECOMMENDED that sources of requests be authenticated; however, in some cases an anonymous request may be appropriate. Requestors MAY make anonymous requests and it is up to the recipient's policy to determine if such requests are acceptable. If not a fault SHOULD be generated (but is not required to be returned for denial-of-service reasons). The [WS-Security] specification determines time references in terms of the dateTime type defined in XML Schema. It is RECOMMENDED that all time


references use this type. It is further RECOMMENDED that all references be in UTC time. Requestors and receivers SHOULD NOT rely on other applications supporting time resolution finer than milliseconds. Implementations MUST NOT generate time instants that specify leap seconds. Also, any required clock synchronization is outside the scope of this document. V. REQUESTING A SECURITY TOKEN The element (RST) is used to request a security token. This element SHOULD be signed by the requestor, using tokens contained/referenced in the request that are relevant to the request. If using a signed request, the requestor MUST prove any required claims to the satisfaction of the security token service. If a parameter is specified in a request that the recipient doesn't understand, the recipient SHOULD fault. The syntax for this element is as follows: ... ... ... This is a request to have a security token issued. /wst:RequestSecurityToken/@Context This optional URI specifies an identifier for this request. All subsequent RSTR elements relating to this request MUST carry this attribute. /wst:RequestSecurityToken/wst:TokenType This optional element describes the type of security token requested, specified as a URI. That is, the type of token that will be returned. message. VI. RETURNING A SECURITY TOKEN To return or response to a security token request we use element (RSTR). It should be noted that any type of parameter specified as input to a token request MAY be present on response in order to specify the exact parameters used by the issuer. Specific bindings describe appropriate restrictions on the contents of the RST and RSTR elements. In this specification the RSTR message is illustrated as being passed in the body of a message. However, there are scenarios where the RSTR must be passed in conjunction with an existing application message. In such cases the RSTR (or the RSTR collection) MAY be specified inside a header block. The exact location is determined by layered specifications and profiles; however, the RSTR MAY be located in the header if the token

316


is being used to secure the message (note that the RSTR SHOULD occur before any uses of the token). The combination of which header block contains the RSTR and the value of the optional attribute indicate how the RSTR is processed. It should be noted that multiple RST elements can be specified in the header blocks of a message. It should be noted that there are cases where an RSTR is issued to a recipient who did not explicitly issue an RST (e.g. to propagate tokens). In such cases, the RSTR may be passed in the body or in a header block. The syntax for this element is as follows: ... ... ... The syntax for this element is as follows: ... ... ... The following describes the attributes and elements listed in the schema overview above: /wst:RequestSecurityTokenResponse This is the response to a security token request. /wst:RequestSecurityTokenResponse/@Context

This optional element specifies the type of security token returned. This is an extensibility mechanism to allow additional elements to be added. If an element is found that is not understood, the recipient SHOULD fault. VII.

This Trust Model is used by a Web service for additional challenges to a requestor to ensure message freshness and verification of authorized use of a security token.So this scheme is simple and can realize security across domains by evaluating the comprehensive trust correctly and effectively.

REFERENCES [1]

HongMei Ge, Microsoft & Slava Kavsan, RSA Security, "Web Services Security Policy Language," IBM, Microsoft, RSA Security, VeriSign, December 2002.

[2]

Heather Hinton, IBM & Richard Levinson, Sonic Software "Web Services Policy Framework," BEA, IBM, Microsoft, SAP, Sonic Software, Verisign, September 2004.

[3]

Malik, Z., Bouguettaya, A.: Reputation-based Trust Management for Service-Oriented Environments. VLDB Journal 18(4), 885–911 (2009)Trust Assessment for Web Services under Uncertainty 485

[4]

Maximillien, E.M., Singh, M.P.: Conceptual Model of Web Service Reputation.SIGMOD Record 31(4), 36–41 (2002)

This optional URI specifies the identifier from the original request. That is, if a context URI is specified on a RST, then it MUST be echoed on the corresponding RSTRs. For unsolicited RSTRs (RSTRs that aren't the result of an explicit RST), this represents a hint as to how the recipient is expected to use this token. No values are pre-defined for this usage; this is for use by specifications that leverage the WS-Trust mechanisms. /wst:RequestSecurityTokenResponse/wst:TokenType


CONCLUSION

317


Delegation and Revocation Using Role Based Access Control Nirmalrani V1, Sakthivel P2 1

Department of Information Technology, 2Department of Electronics and Communication Engineering 1 Sathyabama University, Chennai, India, 2 Anna University, Chennai, India 1

[email protected], [email protected]

Abstract: Global education system is a key area in Information Technology. It facilitates the developers to provide various learning systems with low cost. The needs for effective information sharing in secured manner have been largely ignored in e-learning. Information sharing usually occurs in broad, highly dynamic network based environments. Accessing the resources in a secured manner is a vital challenge in this environment. Hence this paper aims to build a new rule-based framework to identify and address issues of sharing information through RoleBased Access Control (RBAC) management. The basic implementation of delegation and revocation depicted in RBAC involves individual users being associated with roles, as well as roles being associated with permissions (each permission is a pair of objects and operations). As such, a role is used to associate users and permissions. A user in this model of the virtual university environment is a human being, such as a lecturer or professor. A role is a job functions or job title within the organization associated with authority and responsibility. There are two types of roles: regular role and administrative role. This paper only addresses regular roles in the organization. Roles have hierarchy structure in the RBAC model. The framework includes a role-based group delegation granting model, group delegation revocation model, authorization granting, and authorization revocation. This paper also analyzes various revocations and the impact of revocations on role hierarchies.

Therefore, effective and efficient communication with distant collaborators is required for the collaboration between and within a virtual university environment. This paper aims to develop a policy based framework for information sharing in a distributed collaborative virtual university environment with role based delegation. The inclusion of role based delegation and revocation allows users themselves to delegate role authorities to others to process some authorized functions and later remove those authorities. Role based delegation and revocation models are developed with comparison to establish technical analysis, laboratory experiments, support hierarchical roles and multistep delegation. The models are implemented to demonstrate the feasibility of the framework and secure protocols for managing delegations and revocations.

Keywords: E-learning, RBAC, Role based Delegation, Revocation, Rule Based Framework

RBAC enables managing and enforcing security in large scale and enterprise wide systems and its applications depend on specific system requirements. In RBAC models, permissions are associated with roles, users are assigned to appropriate roles, and users acquire permissions through roles. Users can be easily reassigned from one role to another. This project proposes a delegation framework, which includes the structure of role based delegation, role based group delegation and revocation models.

Delegation is the process whereby an active entity in a distributed environment grants access resource permissions to another entity. The most common delegation types are user-to-machine, user-to-user and machine-to-machine delegations. Authorization decisions are made based on the identity of the resource requester.

I. INTRODUCTION E-Learning is a significant modern education learning approach and tool performed over electronic devices. E-Learning has the potential to become a lower cost and efficient education tool and one of the key e=commerce applications. As an example in many countries e-learning concepts are implemented in virtual universities. These virtual universities are becoming strongly networked and fundamental changes in the organization of education are occurring. The collaboration within a virtual university is found to be a beneficial and enjoyable component of offering a course. Furthermore, the collaborative partners in the virtual university are often distributed and their collaboration occurs across highly dynamic internet based environments and formally accessing the subject materials in a secure manner poses a vital challenge.


II. ROLE BASED DELEGATION AND REVOCATION FRAMEWORK The basic elements and relations in RBAC are depicted in Fig1. RBAC involves individual users being associated with roles, as well as roles being associated with permissions (each permission is a pair of objects and operations). As such, a role is used to associate users and permissions. A user in this model of the virtual university environment is a human being, such as a lecturer or professor. A role is jobs function or job title within the organization associated with authority and responsibility.

318


to a particular user. Thus, the user gets a role (or several roles), and then the role (or roles) gives him predefined permissions. The roles indirection is similar to groups in UNIX and other operating systems and privilege groupings in database management systems. Though, groups can include only users as their members, RBAC can contain collections of users, permissions, and other roles “in a single access control model in terms of roles and role hierarchies, role activation, and constraints on user / role membership and role set activation.

There are two types of roles: regular role and administrative role. We only address regular roles in this paper. Roles have hierarchy structure in the RBAC model. Senior roles inherit all permissions from junior roles. Therefore, a senior-junior relationship exists in the figure. Permission is an approval for a particular operation to be performed on one or more objects. As shown in the figure, User_name, Role_name, Perm_name, Oper_name, and Object_name are attributes of User, Role, Permission, Operation, and Object, respectively. Four relationships between users and roles, between roles and permissions, between roles and roles, and between operations and objects are many to many. The security policy of the organization determines role membership and the allocation of each role’s capabilities.

RBAC controls the users’ access to information and system resources based on users’ activities in the system, and requires the roles’ identification in the system. Such a model is supposed to have a set of basics elements such as users, roles, permissions, operations, and objects, and relations between these elements [1]. A set of actions and responsibilities related to a particular activity can defined a role, then permissions to access objects are specified for roles, and afterward, users are assigned to appropriate roles.

This paper gives a design for Role based access framework for role delegation, revocation and group based delegation. Here every user can delegate his / her task to their juniors with certain constraints. Using the delegation and revocation framework the user can achieve different type of delegation such as group delegation, role delegation and partial delegation. The director will monitor all the task of all users. The advantages of this framework are:

Organizations may require various numbers of roles and access rules. In most organizations roles are quite constant while users and tasks which are assigned to them can be impermanent, and reassignment is essential. So, RBAC is a most suitable approach to provide secure association and access, because “RBAC provides a powerful mechanism for reducing the complexity, cost, and potential for error in assigning permissions to users within the organization” [2]. Since RBAC has role hierarchies, where a given role can enclose all of the permissions of some other roles, it is a way to go for organizations where roles have overlapping permissions. The major requirements and challenges of the role based delegation within the collaborative virtual university are

Higher official will go for task monitoring easily Role Delegation and revocation can be done quickly We can achieve intra department task delegation in an efficient manner.

Group based delegation means that a delegating user may need to delegate a role to all members of a group at the same time Multistep delegation occurs when a delegation can be further delegated. Single step delegation means that the delegated role cannot be further delegated. Revocation schemes are an important feature of collaborative virtual university systems. They take away the delegated permissions. There are different revoking schemes; among them are strong and weak revocations, cascading and no cascading revocations as well as grant dependent and grant independent revocations. Fig1. Delegation Architecture.

Constraints are an important factor in RBAC for laying out higher level organizational policies. They define whether or not the delegation or revocation process is valid.

III. RBAC ALGORITHM The basic idea of RBAC is to give permissions to the users indirectly by using roles which are assigned


319 312


Partial delegation means that only subsets of the permissions are delegated while total delegation means that all permissions are delegated. Partial delegation is an important feature because it allows users only to delegate required permissions. The well known least privilege security principle can be implemented through partial delegation.

The same user can be an original user of one role and a delegated user of another role. Also, it is possible for a user to be both an original user and a delegated user of the same role. The original user assignment (UAO) is a many-to-many user assignment relation between original users and roles. The delegated user assignment (UAD) is also a many-to-many user assignment relation between delegated users and roles. Fig2. shows the relationships of user and role in UAO and UAD.

This paper proposed a rule based framework for role based delegation. The delegation framework including delegation granting and revocation models and group based delegation.

A delegation relation (DELR) exists in the rolebased delegation model, which includes three elements: original user assignments UAO, delegated user assignment UAD, and constraints. The motivation behind this relation is to address the relationships among different components involved in a delegation. In a user-to-user delegation, there are five components: a delegating user, a delegating role, a delegated user, a delegated role, and associated constraints. Fig3. shows the role based delegation model.

A. Role Based Delegation Model An important concept within RBAC is a session, which involves a mapping between a user and possibly many roles. A session is always associated with a single user and each user may establish zero or more sessions. There may be hierarchies within roles. Senior roles are shown at the top of the hierarchies. Senior roles inherit permissions from junior roles. Let x > y denote x is senior to y with the obvious extension to x _ y. Role hierarchies provide a powerful and convenient means to enforce the principle of least privilege since only required permissions to perform a task are assigned to the role. Table 1 expresses an example of user-role assignment. There are two sets of users associated with a role r: 1) Original users are those users who are assigned to r and 2) Delegated users are those users who are delegated to r. TABLE 1 USER-ROLE ASSIGNMENT Role_Name

User_Name

DIRECTOR

TOMMY

REGISTRAR

RICHARD

HOD

JOHN

ASST. HOD

NIKE

PROFESSOR

TONY

Fig3. Role Based Delegation Model.

B. Role-Based Group Delegation The scope of the model is to address role-based group delegation with role hierarchies and constraints that support multistep and partial delegations. There are two kinds of group delegation: user-group delegation and group-group delegation. We first discuss user-group delegations, which consist of original user-group and delegated usergroup delegations. The new relation of user-group delegation is defined as delegation group relation (DELUGR), which includes: original user assignments UAO; delegated user assignments UAD; delegated group assignments GAD; and constraints. In a user-group delegation, there are five components: a delegating user (or a delegated user); a delegating role; a delegated group; a delegated role; and associated constraints. A user-group delegation relation is a one-to-many relationship on user assignments. It consists of original user group delegation (ORIUGD) and delegated user group delegation (DELUGD). Fig4. illustrates components and their relations in the

Fig2. UAO and UAD Relationships.


313 320


dominance, and only the direct delegating user can remove the delegation relationship.

role based delegation model. Hence, have the following elements and functions for user-group delegation

The features of scheme DWLD when user U1 wants to revoke role r from user U2 are: 1. 2. 3. 4.

U1 does not grant role r to U2, Role r may still stay with U2 if users other than U1delegate r to him, Roles granted by users other than U1 are intact, and The delegation structure may need to be updated since roles delegated by U2 have to remain.

2) Dependent Strong Local Delete (DSLD)

C. Role-Based Delegation Revocation

The Dependent Strong Local Delete (DSLD) is different from DWLD in the dominance aspect. It does not have resilience and propagation but does have dominance, and only the direct delegating user may take away the delegation relationship.

Revocation is a significant function in role-based group delegations.

The features of scheme DSLD when user U1 wants to revoke role r from user U2 are:

Fig4. Role-based Group delegation model

1. 2.

1) Revocation dimension: Dependency:

3.

Dependency refers to the legitimacy of a user who can revoke a delegated role. Dependent revocation means that only the delegating user can revoke the delegated user from the delegated role. Independent revocation allows any original user of the delegating role to revoke the user from the delegated role.

4.

U1 does not grant role r to U2, Implicit role r0 that is senior to role r is removed, Roles other than r and its senior role are intact, and The delegation structure may need to be updated since roles delegated by U2 have to remain.

3) Dependent Weak Global Delete (DWGD): Resilience: The Dependent Weak Global Delete (DWGD) is different from DWLD in the propagation aspect. It does not have resilience and dominance but does have propagation, and only the direct delegating user may remove the delegation relationship.

This dimension describes revocation through deleting or negative authorization. The effect of a role deleting revocation from a user is acted; another user may grant the user the same role that was just revoked, and as a result, the revocation has no effect on the user. The negative authorization has high priority in this dimension.

The features of scheme DWGD when user U1 wants tore revoke role r from user U2 are: 1. 2. 3. 4.

Propagation: This dimension distinguishes revocations according to a delegation structure of role locations. There are local revocation and global revocation in the dimension. The local revocation only happens to the direct delegation relationship, while the global revocation affects all other users authorized by the revoked user.

4) Dependent Weak Local Negative (DWLN): The Dependent Weak Local Negative (DWLN) is the scheme with negative authorization in resilience dimension. When a negative authorization of a role is granted to a revoke as weak and local revocation, the negative authorization will remain with the

D. Revoking Delegation 1) Dependent Weak Local Delete (DWLD) The Dependent Weak Local Delete (DWLD), as the first revocation scheme, is the easiest operation. It does neither have resilience, propagation, nor


U1 does not grant role r to U2, Role r delegated by U2 to users is removed, Roles other than r are intact, and The delegation structure may need to be updated since roles other than r delegated by U2 have to remain.

314 321


revoke until the authorization is removed, and the revoke cannot act in the role even though the role is delegated to her/him. The negative revocation scheme is different from other revocation schemes in the resilience dimension. We use black for the role with negative authorization. IV. ASP .NET IMPLEMENTATION This section presents the implementation of the group delegation with ASP .NET technology. The implementation of the role-based delegation framework in the virtual university includes many components, for example: 1.

Fig6. Constraint Based Delegation

V. CONCLUSION

The structure of each university involved in the virtual university

2.

Role hierarchy of staff and group hierarchy

3.

Constraints in delegation and revocation.

This paper has discussed a role-based delegation model for virtual university learning environments and its implementation with ASP .NET. We have analyzed a delegating framework including delegating authorization and revocation with constraints and extended it to include group-based delegation. The revocation dimensions and delegating revocations are discussed. The advantages of the work in this paper are given as follows:

A. Implementation Screen shots 1) Role Assignment:

1.

Assigning the roles to the individual users at virtual universities is shown in Fig5.

2.

3.

Provides a framework for efficient information sharing in a virtual university since users can delegate (or revoke) roles to (or from) users and groups without the administrators’ interaction. The framework supports not only group delegations and various revocations but also multistep delegations, constraints, and partial delegations. Analyzes the impact of revocations with role hierarchy.

This paper is not discussing how to analyze constraints in the delegation model at a virtual university and the workflow analysis of delegation processes in the role-based delegation model at virtual university. REFERENCES [1]

[2]

[3]

Fig5. Role Assignment

[4]

2) Constraint Based Delegation: [5]

The Constraints which are used for delegating the roles to other users are show in Fig6.


[6]

315 322

Hua Wang, Yanchun Zhang and Jinli Cao, “Effective Collaboration with Information Sharing in Virtual Universities”, IEEE Transactions on Knowledge and Data Engineering, Vol. 21, No. 6, PP. 840 – 853, June 2009. G. Ahn, B. Mohan, and S. Hong, “Towards Secure Information Sharing Using Role-Based Delegation,” J. Network and Computer Applications, vol. 30, no. 1, pp. 4259, 2007. M. Arenas and L. Libkin, “A Normal Form for XML Documents,” ACM Trans. Database Systems, vol. 29, no. 1, pp. 195-232, 2004. G. Ahn and R. Sandhu, “Decentralized User Group Assignment in Windows NT, ” J. Systems and Software, vol. 56, no. 1, pp. 39-49, 2001. G. Ahn and R. Sandhu, “Role-Based Authorization Constraints Specification, ” Information and System Security, vol. 3, no. 4, pp. 207-226, 2000. E. Barka and R. Sandhu, “Framework for Role-Based Delegation Models and Some Extensions,” Proc. 16th Ann. Computer Security Applications Conf. (ACSAC ’00), pp. 168-177, 2000.


Groundwater Level Prediction Using Fuzzy and ANFIS Method M.Kavitha Mayilvaganan1, K.B.Naidu2 1

Research Scholar, 2Professor, Department of Mathematics, Sathyabama University, Chennai. [email protected] very precise measurement, and moreover a precise model can be very complicated and uneconomical in the development time. Secondly the variables involved in the problem are fuzzy in nature. Therefore, a fuzzy logic can provide better solutions in simple way [1, 6 and 7].

Abstract:Fuzzy-logic, a soft computing technology of Artificial Intelligence, nowadays has a great concentrated applications importance in water engineering. It is an excellent mathematical tool to handle uncertainty of the system arising due to the fuzziness or vagueness. Groundwater is of major importance to civilization, because it is the largest reserve of drinkable water in regions where humans can live. The estimation of the groundwater level is one of the important aspects to understand the mechanism which comprises groundwater resources and to predict what might happen under various possible future conditions. The soft computing techniques viz. Fuzzylogic modeling and Adaptive Neuro Fuzzy Inference System were used in present investigation. These systems begin with some basic rules that describe the process. Two models have been developed, one with Fuzzy rule based models and the other with ANFIS models in the prediction of ground water level of a watershed. On the basis of performance criteria ANFIS yielded the better results out of the two models developed.

II. FUZZY LOGIC Fuzzy set theory, which has been proposed in 1965 by Lofti A. Zadeh (1965) [10], is a generalization of classical theory. Fuzzy logic representations found on Fuzzy set theory try to capture the way humans represent and reason with real world knowledge in the face of uncertainty. Uncertainty could arise due to do generality, vagueness, Ambiguity, chance or incomplete knowledge. A Fuzzy set can be defined mathematically by assigning to each possible individual in the universe of discourse, a value representing its grade of membership in the fuzzy set. This grade corresponds to the degree to which that individual is similar or compatible with the concept represented by the Fuzzy set. In other words, fuzzy sets support a flexible sense of membership of elements to a set [2 to 5]. The range of the model input values, which are judged necessary for the description of the situation, can be portioned into fuzzy sets. The process of formulating the mapping from a given input to an output using fuzzy logic is called the fuzzy inference. The basic structure of any fuzzy inference system is a model that maps characteristics of input data to input membership functions, input membership functions to rules, rules to a set of output characteristics, output characteristics to output membership functions, and output membership function to a single valued output or a decision associated with the output. In rule based fuzzy systems, the relationship between variables are represented by means of fuzzy if-then rules e.g. “if antecedent proposition then consequent proposition”[8].

Keywords: Fuzzy logic, ANFIS, training, learning, groundwater level modeling, Observation wells, watershed

I. INTRODUCTION During the last decade of 19th century, the artificial neural network and thereafter fuzzy logic techniques have become popular in data forecast of time series particularly in the application where the deterministic approach presents serious drawback, due to the noisy or random nature of data. These learning based approaches, which can be considered an alternative to classical methods, exploit the statistical relationships between inputs and outputs, without explicitly considering the physical process relationships, which exist between them. Although fuzzy logic attempts to simulate human “Vagueness” of reasoning, in practice many characteristics of this approach, such as ability to learn and generalize, the ability to cope up with noise, the distribute processing, which maintains robustness can be of great help in many engineering tasks. Moreover, in general, this technique can be included in overall concept of soft computing approach [9]. Again the problem of groundwater modeling does not require a Sathyabama University

323


III. STUDY AREA The Thurijapuram watershed covers geographical area of 151.38 sq. km and is located in between 12o12’58” and 12o 21’11” North latitudes and 78o59’45” and 79o9’28” East longitudes (Fig. 1.) It is mainly situated in Thiruvannamalai district of Tamilnadu, India. It is mainly located in Thurinjapuram block and partially falls into two other blocks (Chengam and Thiruvannamalai). Thurinjalar is one of the major tributaries of Ponnaiyar major river originating from Kavuttimalai reserve forest in Chengam Taluk of Tiruvannamalai district. Thurinjalar River, which is the major stream draining the area, exhibits only sporadic flow during the rainy season. The drainage characteristics are very good. Bedrock is peninsular gneiss of Archean age. The Thurinjapuram area can be classified as ‘‘hard rock terrain’’. The predominant soil types in this river basin are Entiso, Inceptisols, Vertisol and Alfisols. The soil in this minor basin is observed to have good infiltration characteristics. Hence groundwater recharge is possible in this area.

Fig. 1 Study area IV. METHODOLOGY

The climate is semi-arid. May is the hottest month with a maximum temperature of up to 41° C and December is the coolest month with a maximum of 21.6° C. Hydro meteorological data were collected from Kilnatchipattu weather station maintained by State Ground & Surface Water Resources Data Centre, W.R.O, P.W.D. The economy of the Thurinjapuram sub watershed depends mainly on agriculture. Data from three observation wells, which have been monitored on a monthly basis by the Department of Groundwater, are available in the Thiruvannamalai Groundwater subdivision.

B. Fuzzification of Input and Output Data Since in the present study the data were qualitative form rather to linguistic, hence these need to be fuzzified first. Antecedent rainfall RF (k) and water levels WL (k) are taken as inputs and the future water level WL (k+1) is an output. Input and output data were fuzzified in to fuzzy subsets, in order to cover the whole range of changes. The criterion of defining fuzzy subsets is based on subjective perception of specific linguistic level by relevant experts.

A. Data The input data used for water level prediction are monthly Rainfall and Ground water (level in the observation well) data of Thurinjapuram watershed in Tamilnadu, India, and one month ahead groundwater level as output. For the present study monthly water level data for three observation wells (23112, 23142, and 23143) during 1985 to 2008 has been collected from Thiruvannamalai Groundwater subdivision. In the same period monthly Rainfall data were collected from Kilnatchipattu Raingauge station.

Fig.2.Membership function for RF(k) Sathyabama University

324


C. Fuzzy Rules On the basis of the available data, we define relation between fuzzy inputs and out puts, these are fuzzy rules for the analysis. To combine input data we used fuzzy intersection rule for fuzzy sets. There are different rules for each model. Fuzzy rules are as in Fig.5. D. Defuzzification Method The result obtained from the implication is in the form of a fuzzy set. This is defuzzified to get a crisp output. In the present study, we used centroid method to defuzzify the data, which is given by algebraic expression:

Fig.3. Membership function for WL(k)

E. ANFIS Model The acronym ANFIS derives its name from adaptive neuro-fuzzy inference system. In this model the fuzzy system is configured in a parallel fashion based on a corporative relationship a conceptual ANFIS consists of primarily five components: inputs and output database, a Fuzzy system generator, a FIS, and an adaptive neural network. The Sugeno-type FIS, which is the combination of a FIS and an adaptive neural network, was used in this study for the prediction purposes (Fig.6).

FIG.4.MEMBERSHIP FUNCTION FOR WL(K+1)

Fig.5.Fuzzy rules Fig.6 ANFIS model RF(k) is divided into three subsets, as low, moderate and high (Fig. 2). WL(k) is divided into three fuzzy subsets, as low, moderate and high (Fig. 3). WL(k+1) is divided into five subsets v.low, low, moderate, high and v.high (Fig. 4).


V. MODEL DEVELOPMENT A simple model was developed by taking two parameter antecedent rainfall RF(k) and water levels WL(k) are taken as inputs, and the future water level WL(k+1) is an output using the representation. WL (k+1) = f{R (k), WL (k)}.

325

Proceedings of National Conference on Computing Concepts in Current Trends, NC4T’11 11th & 12th Aug. 2011, Chennai, India Estimated groundwater level (m)

Fuzzy Logic Rule base model and Adaptive Neuro Fuzzy Inference System are used to prediction of this model (FLM, ANFIS). VI. RESULTS AND DISCUSSION F. Prediction of Ground Water level Using Fuzzy Logic

4 2 0 2

4

6

8

10

12

10 5

20 07

20 06

20 05

20 04

20 03

20 02

0 20 01

Groundwater level (m)

15

Numbe r of Data (Years)

Estimated groundwater level (m)

Actual

20 07

20 06

20 05

20 04

20 03

20 02

6

Well No.23143

12 10 8 6 4 2 0 20 01

8

Obse rve d groundwate r le ve l (m)

We ll No.23112 Groundwater level (m)

2

R = 0.9452

10

0

Models were constructed using fuzzy logic toolbox with MATLAB 7.6. The membership functions are constructed using the observed data. Fuzzy Inference System prepared using Fuzzy logic toolbox in MATLAB. Rules which are constructed from observed data are then added in fuzzy logic rule editor. The FIS was then evaluated to obtained output data which is predicted value of the water level for the model. Predicted groundwater level values and observed groundwater level values were then plotted on ordinate against years on abscissa, to evaluate the model performance, on qualitative basis. Fig.7 show the close agreement between observed and predicted value can be easily observed.

FLM Predicted

14 2

R = 0.8546

12 10 8 6 4 2 0 0

2

4

6

8

10

12

14

16

Observe d groundwater level (m)

Numbe r of Data (Years) Actual

12

FLM Predicted

Estimated groundwater level

Fig.7 Observed and predicted values of water level of three wells 12

G. Prediction of Ground Water level Using ANFIS

2

R = 0.8862

10 8

The Sugeno type Fuzzy Inference System is used to construct the ANFIS model. The hybrid ANFIS model with 3 subsets of membership functions of Gaussian shape for input and linear output membership function gives the best results. The 150 epochs were given to train the model. The observed and predicted values of water level for all the three wells were plotted (Fig.8). These figures show that there exist close relation between observed and predicted values of water level using ANFIS technique.

6 4 2 0 0

2

4

6

8

10

12

Obse rve d groundwater level (m)

20 07

20 06

20 05

20 04

20 03

20 02

12 10 8 6 4 2 0 20 01


Well No.23141

Number of Data (Years) Actual


FLM Predicted

326


H. Comparison between Fuzzy Logic Rule Based Models and ANFIS Models The values of regression coefficient (R²) range from 0 to 1. Small values indicate that the model does not fit the data well. R² values of ANFIS model is very much high when compared to Fuzzy model. Which indicates that ANFIS Models performed better than Fuzzy logic rule based models. Scatter plot between observed and predicted groundwater level is also given for the goodness of fit.

20 07

20

06

20 05

20 04

03 20

20

20 02

12.00 10.00 8.00 6.00 4.00 2.00 0.00 01


Well No.23112

Number of Data (Years )


Actual

ANFIS Predicted

12 2

R =1

10 8

VII.

6

Soft computing techniques like Fuzzy Logic rule based models and ANFIS are reliable and more accurate than conventional methods. The performance of Fuzzy Logic rule based models and an ANFIS models was found to be satisfactory on the basis of performance evaluation of models. On the basis of performance evaluation of models, ANFIS model performed better than Fuzzy Logic rule based models. The present study yielded good results and has shown superior performance and the application of modeling ground water level prediction.

2 0 0.00

2.00

4.00

6.00

8.00

10.00

12.00

Observed groundwater le vel (m) We ll No.23141 12.00


CONCLUSIONS

4

10.00 8.00 6.00 4.00 2.00

20 07

20 06

20 05

20 04

20 03

20 01

20 02

0.00

Numbe r of Data (Ye ars)


Actual

VIII.

ANFIS Predicted

12

[1]

2

R = 0.9969 10 8

[2]

6 4 2

[3]

0 0.00

2.00

4.00

6.00

8.00

10.00

12.00

Obse rve d Groundwate r level (m)

[4]

15 10 5

[5] 07 20

20

06

20 05

20 04

20 03

20 02

0 20 01


Well No.23143

Number of Data (Years)


Actual

[6]

ANFIS Predicted

16 2

14

R = 0.9601

12

[7]

10 8 6 4 2 0 0

2

4

6

8

10

12

14

[8]

16

Obs erved groundwater level (m)

[9]

Fig.8 Observed and predicted values of water level of three wells


[10]

327

REFERENCES

Bahat, M., Inbar, G., Schneider, M. A Fuzzy Irrigation Controller System, Artificial Intelligence,Vol. 139, 2000, pp. 137–145. Bardossy, A., Disse, M. Fuzzy Rule Based Models for Infiltration, Water Resources Research, Vol. 29, No 2, 1993, pp. 373–382. Bardossy, A., Duckstein, L. Analysis of Karstic Aquifer Management Problem by Fuzzy Composite Programming, Water Resources. Bulletin, Vol. 28, No 1, 1992, pp. 63–73. Coppola, Duckstein, L., Dvis, D. Fuzzy Rule Based Methodology for Estimating Monthly Ground Water Recharge in a Temperate Watershed, Journal of Hydrologic Engineering, Vol. 7, No 4, 2000, pp. 326–335. Fontane, D. G., Gates, T. K., Moncada, E. Planning Reservoir Operations with Imprecise Objectives, Journal of Water Resources Planning and Management, Vol. 123, No 3, 1997, pp. 154–168. Lohani, A. K., Goel, N. K., Bhatia, K .K. S., TakagiSugeno. Fuzzy Inference System for Modelling Stage-Discharge Relationship, Journal of Hydrology, Vol. 331, 2006, pp. 146–160. Lohani, A. K., Goel, N. K., Bhatia, K. K. S. Deriving Stage Discharge Sediment Concentration Relationships Using Fuzzy Logic, Hydrological Sciences – Journal-des Sciences Hydrologiques,Vol. 52, No 4, 2007, pp. 793–807. Panigrahi, D. P., Mujumdar, P. P. Reservoir Operation Modelling with Fuzzy Logic, Water Resources Management, Vol. 123, No 3, 2000, pp. 154–168. Rajasekaran, S., Pai Vijayalakshmi, G. A. Neural Networks. Fuzzy Logic and Genetic Algorithms.New Delhi: Prentice Hall of India Pvt. Ltd., 2007. 10. Zadeh, L. A. Fuzzy sets, Information and Control, Vol. 8, No 3, 1965, pp. 338–353.


Integrating Web Services and Agent Technology with DIGIGEO Agent 1

Lalitha M, 2Nisha S.

1

PG Scholar , M.C.A, Panimalar Engineering College PG Scholar , M.C.A, Panimalar Engineering College 1 [email protected], [email protected] (World Wide Web) due to its exponential growth Abstract:-The advent of Internet has brought down enabled substantial progress in new information geographical distances but proportionally increased society functions [9, 3] such as online commerce. the digital geographical distances with masses of data Electronic commerce entails business-to business, and information that demands more than a mind to business-to-customer and customer-to-customer comprehend. Every digital megabyte is a mile of transactions. It encompasses a wide range of issues physical distance. Getting the wrong information is including security, trust, reputation, law, payment like taking a wrong turn. It wastes time, energy, mechanisms, advertising, ontologies, electronic productivity and money. This growth has led to the product catalogs, intermediaries, multimedia demanding needs of effectively using Digital shopping experiences, and back office management. Geography. Agent technologies can be applied to any of these Intelligent agents can play a role in ensuring this areas. Software agents help automate a variety of doesn’t happen. They are pieces of software focused activities, mostly time consuming ones, and thus on organizing and getting the best out of our digital lower the transaction costs. Software agents differ geography. They will become extensions of our will, from “traditional” software in that they are aiding in the fulfillment of our needs and wants. personalized, social, continuously running and semiThey will venture into the digital wilderness and autonomous[12] . In this way, e-commerce is bring back to us the list of options we demand. This becoming more user-friendly, semi-intelligent and paper focuses on developing an intelligent agent human-like. These qualities are conducive for DIGIGEO that have been applied to tourism optimizing the whole buying experience and industry, especially in the Indian Railways and revolutionizing commerce, as we know it today [11]. discusses on how it deals with the limitation of handling time delay of searching information in the tourism sector and dealing with the problem of II. INTELLEGENT AGENTS online ticket booking. A model of the Agent has been proposed along with deployment and a Intelligent agents are software entities that can performance evaluation result has been added to execute functionalities in an autonomous, pro-active, highlight on the effectiveness of the system. This social and adaptive fashion. These functionalities paper also gives a brief introduction to intelligent include searching, comparing, learning, negotiating agents and electronic commerce, followed by a and collaborating and endorsing. Subsequently, review of agent technologies involved in buying and based on these functionalities, seven types of agents selling and the methodologies to endorse future have been identified: collaborative agents, interface digital geography with intelligent agents. agents, mobile agents, information/Internet agents, reactive agents, hybrid agents and smart agents. Provisionally in electronic commerce, interface Keywords:- DIGIGEO, Agent, Ontology, Eagents and information or Internet agents will be Commerce, Digital Geography, RSA, WSP heavily used. Theory and development on the use of intelligent agents has focused on two areas - making agents 'smarter' (that is, enlarging their I. INTRODUCTION functionalities) and creating a framework (an The world economy is going digital and a electronic marketplace) in which intelligent agents handful of business pioneers already provide can operate. Hence in the near future, these developments make the digital geography a reality. exemplars of the tactics and operation of 21st century business competitors. In recent years the Internet 2


328


III .ANATOMY OF AN INTELLIGENT AGENT

Inferential capability: can act on abstract task specification using prior knowledge of general goals and preferred methods to achieve flexibility.

A basic software agent stands on three pillars autonomy, reactivity, and communication ability. The notion of autonomy means that an agent exercises exclusive control over its own actions and state. Reactivity means sensing or perceiving change in their environment and responding. the ability to communicate with other entities: human users, other software agents, or objects.

Temporal continuity: persistence of identity and state over long periods of time. Personality: the capability of manifesting the attributes of a believable character such as emotion.

Intelligent agents are similar to objects in a number of ways. Agents can be organized into generalization and specialization hierarchies, to exploit inheritance (eg, a risk assessment agent represents a specialization of a personal assistant agent class). An agent can advertise its services using a variety of means, and how it implements these services should be transparent (ie, encapsulation of behavior). Different agents can respond to the same service request differently without the requester needing to know about such differences.

Adaptivity: being able to learn and improve with experience. Mobility: being able to migrate in a self-directed way from one host platform to another. V. CLASSIFICATION OF AGENTS Agents may be classified as:

IV. ATTRIBUTES OF INTELLIGENT AGENTS

Mobility (static, mobile) – the ability to move around.

Based on the requirements of a particular problem, each agent might possess to a greater or lesser degree the following attributes [8,5,7]:

Presence of symbolic reasoning model a) Deliberative – from the deliberative thinking paradigm, which holds that agent posses symbolic reasoning models, they engage in planning and negotiation with other agents in order to achieve their goals.

Autonomy: goal-directedness, proactive and selfstarting behavior. Reactivity: the ability to selectively sense and act. Collaborative behavior: can work in collaboration with other agent to achieve a common goal.

b) Reactive – do not have any internal, symbolic models of their environment; instead they respond in a stimulus response manner to the present state of the environment in which they are embedded.

“Knowledge-level” communication ability: the ability to communicate with human and other agents with language more resembling humanlike speech than symbol-level protocols.

VI ELECTRONIC COMMERCE Electronic commerce is sharing business information, maintaining business relationships, and conducting business transactions by means of communication networks [14]. E-commerce includes the relationship between companies (business tobusiness), between customers (customer to customer), as well as between companies and customers (business-to customer). Business-tobusiness segment currently dominates the Ecommerce while consumer oriented segment is significantly lagging behind and current estimates place it at less than 10 percent of the total volume, even tough they are all experiencing an exponential growth [14].

Figure 1: Anatomy of an Intelligent Agent


329 322


They must be adaptive to meet the needs of the moment and bring productivity to an increasingly overwhelmed business user and self-service to customers. VIII GOVERNMENT INITIATIVES IN ENDORSING DIGITAL GEOGRAPHY Government initiatives have made governments participants in e-commerce to endorse digital geography in a number of ways. Through these activities, governments function as consumers and suppliers of e-commerce services, thus transforming their role from one of pure policy or regulatory oversight to one of participation. In so doing, governments have the potential to make a positive contribution to e-commerce development. Governments that recognize this opportunity may gain a competitive advantage by using e-government operations to assist in driving e-commerce growth. The various initiatives taken by the government in endorsing digital geography includes Internet Tax Freedom Act, Electronic Signatures in Global and National Commerce Act, Digital Millennium Copyright Act etc.,

Figure 2: Pillars of E-Commerce The three pillars of e-commerce are electronic information, electronic relationships, and electronic transactions. The most highly touted applications of Ecommerce are consumer-oriented. They include activities related to buying and selling goods or services, banking, and stock brokerage, accompanied by on-line advertising. Customer can also gain a lot from e-commerce such as investing little effort to find a product or service, the best (lowest) price, the latest updates etc with the help of intelligent agents that has converted the perspective of geography digitally.

Limitations Of Intelligent Agents In Online Transactions E-Commerce is the conduct of commerce in goods and services over the Internet. It includes: consumers using the Internet to purchase goods and services online; as well as businesses selling and communicating with other businesses through the Internet

VII. PERSPECTIVE OF DIGITAL GEOGRAPHY Digital Geography according to the Government is perceived as carrying out commerce online. The new realities of advancements in information technology have created new imperatives for digital geography with the support of government initiatives to enhance commerce worldwide through digital transactions. Today's business systems must provide enterprise (and interenterprise) reach so that islands of disparate information can be integrated into a meaningful whole bringing down the geographical distances. Digital Geography must be able to cope with the overwhelming complexity of distributed technology and an interenterprise information base. They must be open to survive a network-centric ecosystem. Rapid applications development goes without saying, and applications must be designed to embrace constant changes. Digital Geography System with the support of intelligent agents must be knowledge-based (not just information-based) if they are to cope with the incompleteness and ambiguity of real business processes, e-commerce and workflows.


Few technologies have the many benefits ecommerce does, whether taking a small business to never before seen global proportions or opening up millions of new customer markets. But still these online transactions pose lot of limitations that includes the following: Potential time delay in information availability of tourism sector whilst making an extensive search. Problems with online ticket bookings. Credit Card security is a serious issue if vulnerable Costs involved with bandwidth and other computer and server costs Extensive database and technical knowledge and experience required Customer apprehension about online Credit Card orders 330 323


handling requests for information search. Its identifiable feature is to book tickets online based on the user’s priority and choices.

Constantly changing technology may leave slow businesses behind Some customers need instant gratification, and shipment times interrupt that

For instance if User X makes a request for booking 4 tickets online, the existing system of ticket booking just takes the request and allots space on a random schedule without checking for the user’s choice of seating. Whereas the DIGIGEO agent applies techniques based on knowledge-based decision making to allot space based on the user’s request and choices. The proposal highlights on a system that effectively identifies the group booking of tickets with the ID proof submitted for booking the tickets and thus stores the group details in a separate buffer that acts as a temporary storage media. With the help of the stored data then a search is performed on the database that contains details about the availability of tickets and an effective search algorithm is defined to meet the user preference.

Search utilities far surpasses the speed used to find products through catalogs Encourages competition between small and large online retailers Intelligent Agent- Digigeo Tourism has grown to be a world’s largest economic sector and still races continuously to expand its growth to become a powerful tool for economic growth, poverty reduction and for the conservation of natural and cultural resources. Tourism industry in India is on a great boom at the moment. India has tremendous potential to become a major global tourist destination and Indian tourism industry is exploiting this potential to the hilt. Travel and tourism industry is the second highest foreign exchange earner for India, and the government has given travel & tourism organizations export house status. The Indian Tourism industry is being utilised as a powerful tool to facilitate international understanding and helps in building cross-cultural horizons[15] . According to the Travel & Tourism Competitiveness Report 2009 brought out by World Economic Forum, India is ranked 11th in the Asia-Pacific region and 62nd overall in a list of 133 assessed countries in 2008, up three places since 2007.

IX MODEL OF DIGIGEO Due to the growing importance of the tourism sector as a leading contributor to the whole Indian economy developing optimized agents to efficiently carry out online transactions have become a challenge to the government and the public sector service industries. DIGIGEO is an intelligent agent specially proposed and designed for the Indian railway reservation booking, where the role of the agent is to optimize the transaction in terms of time and cost with more importance towards customer satisfaction.

As the importance of travel and tourism industry in India has a prospective growth in the near future, our proposal is to design an intelligent agent DIGIGEO that deals with the limitation of handling time delay of searching information in the tourism sector and dealing with the problem of online ticket booking. The name of the agent is christened as DIGIGEO that means the intelligent agent that helps is manipulating and endorsing the effectiveness of digital geography. Role Of Digigeo In Tourism Sector Figure 3: Model of DIGIGEO

DIGIGEO is the proposed intelligent agent in the tourism sector that enhances the productivity of online ticket booking and gathering vital information. This agent implements a knowledge-based decision making integrated with e-commerce approach in maintaining information integrity and effectively


When the consumer needs a service of ticket booking he makes a request to the Indian Railway System through a user interface by providing the necessary details. The booking system comprises a ticket booking package that uses knowledge base

331 324


system, that helps in efficiently allocating tickets via a decision making process that binds an intelligent agent which helps in identifying the group of people under a single booking and matching out seat allocation continuously based on the customer preference to the maximum possible, if allocation could not be done as per the customer choice a status indication is given to the user about the availability and based on his confirmation booking is done. X. IMPLEMENTATION & EVALUATION OF DIGIGEO The above said model consists of the following functional components: 1. Client System (Service Consumer), 2. Mobile Agent (DIGIGEO), 3.Centralized Server (Knowledge Server), 4. Registry Stationary Agent (RSA), 5. Web Service Provider (WSP). The Client System is implemented in JSP/Servlet technology, and many users can be accommodated without having java runtime environment (JRE) or the MA platform (MAP) installed on their device. The only requirement is a browser to access the Client System. The Client System offers services to clients like: account creation, user login/logout, service invocation policies profile editing, and control of existing agents. Moreover, the administrator is allowed to add/remove/edit user properties/profiles. Finally, users’ service invocation policy profiles are serialized and stored into the server’s database that enables the seamless and transparent provision of services. The DIGIGEO is the representative of the user in the fixed network and is capable of roaming, finding and executing services and delivering results to the user. The DIGIGEO may also spawn clones that execute the selected Web Service (WS) in parallel to minimize the total processing time. Clones can migrate and invoke simultaneously the chosen WS and return to the client system with the results. The MA has the following components: (1) data state, (2) code, (3) migration and cloning policies, (4) matching engine, and, (5) policy management component

the WS expressed in Web Service Description Language (WSDL). A. Service Description This section provides a functional description of the DIGIGEO model using a service scenario. As shown in the use case view, the Client System wishes to find and invoke a certain WS. After a successful registration, the Client system sets the desired criteria for the WS. The user may also define the DIGIGEO service invocation policies and force the latter to follow a certain policy while roaming throughout the network. Subsequently, an agent is created to represent the user in the fixed network and dispatch his service requests. The created Agent DIGIGEO is equipped with the user’s unique ID, service invocation and agent behavioural policies. Such policies are passed to the MA in XML format and stored into his policy repository. The trigger service has the authority to change these policies, according to the messages that the event service may receive from the Client System or other network entities. When DIGIGEO arrives at the registry, it communicates with the RSA, which will query the registry on behalf of the DIGIGEO. The latter finds the service(s) that meet the user needs. The DIGIGEO, after acquiring the results decides on the next step according to its specified service invocation and agent behavioural policies. When the Client system obtains the results, they may ask DIGIGEO to repeat one of the above scenarios by changing, if necessary, its policies, or may cancel the execution of the agent. The Client System may also, at any time, search for the agent, instruct to return or cancel its execution at runtime. Figure 6 shows plotting the migration time of the Mobile Agent from one WS provider to another, against the service result size with the help of 2nd order polynomial interpolation lines. The line labeled “With RSA” indicates

Registry Stationary Agent (RSA) is a stationary agent that acts as a broker between the DIGIGEO and the service registry. RSA implements part of the registry’s functionality and serves DIGIGEO’s requests. By using RSA in the WS registry, the agent does not have to be aware of the implementation specific functionalities of the registry, and, as such, different service registries can be used as long as RSA acts between WS registry and DIGIGEO. Web Service Provider (WSP) provides the WS to interested clients. It maintains a description of


Figure 4: Deployment of DIGIGEO model

332 325


generate itineraries by considering network information published in the WS description, network status and topology. REFERENCES [1] Anthony Chavez, Daniel Dreilinger, Robert H. Guttman, Pattie Maes, A Real-Life Experiment in Creating an Agent Marketplace, Proceedings of the Second International Conference on the PracticalApplication of Intelligent Agents and MultiAgent Technology (PAAM'97), London, UK, April 1997. [2] Aleksander Pivk, Electronic trading with application in Java, Graduation thesis, University of Ljubljana, Slovenia, June 1999. [3] Baldomir Zajc (editor), Proceedings of ERK’99. [4] Dick Stenmark, Intelligent Software Agent, A study, April 1999, http://w3.informatik.gu.se/ ~dixi/agent/agent.htm. [5] O. Etzioni, D.S. Weld, Intelligent agents on the Internet: Fact, fiction, and forecast, IEEE Expert, Vol.10, No.4, pp. 42-49, 1995. [6] Franc Solina, Ales Leonardis, Proper scale for modeling visual data, Image and Vision Computing, 16(2), pp.89-98, 1998. [7] S. Franklin, A. Graesser, Is It an Agent or just a Program? A Taxonomy for Autonomous Agents, In Proceedings of thethird International Workshop on Agent Theories, Architectures, and Languages,New York: Speinger-Verlag. [8] Jeffrey M. Bradshaw, Software Agents, AAAI Press, Menlo Park California, 1997. [9] Matjaz Gams, Information Society and the Intelligent Systems Generation, Informatica 23, 4, pp. 449-454, 1999. [10] Moses Ga, Agents in E-commerce (guest editor), Communications of the ACM, Vol. 42, No.3, pp.7980, ACM Press, March 1999. [11] Pattie Maes, Robert H. Gutmann, Alexandros G. Moukas, Agents that Buy and Sell: Transforming Commerce as we Know It, Communications of the ACM, Vol. 42, No.3, pp.81-91, ACM Press, March 1999. [12] Robert H. Gutmann, Alexandros G. Moukas, Pattie Maes, Agents as Mediators in Electronic Commerce, Electronic Markets, Vol.8, No.1, pp. 22-27, January 1998. [13] Y. Shoham, An Overview of Agentoriented Programming, ed. J.M. Bradshaw, Software Agents, AAAI Press, Menlo Park, California, 1997. [14] Vladimir Zwass, Structure and Macrolevel Impacts of Electronic Commerce: From Technological Infrastructure to Electronic Marketplaces, ECommerce paper, 1998, http://www.mhhe.com/business/mis/zwass/e cpaper.html. [15] Alvin Hung-Chih Yu,Duarte Morais,Garry Chick(2005), Service Quality In Tourism: A Case Study Of The 2001study Tour Of Taiwan. http://www.fs.fed.us/ne/newtown_square/publications /technical_reports/pdfs/2006/341%20papers/yu341.pd f [16] Anurag Dugar (2009), Challenges and Strategies – Enhancing Competitiveness of Indian Tourism Industry, Conference on Global Competition & Competitiveness of Indian Corporate http://dspace.iimk.ac.in/bitstream/2259/499/1/421432+.pdf

Figure 5: Use Case-Client System Perspective Performance Evaluation 1200

700

Migration time(in ms)

1000

560

800

Migration Time without RSA 450

600

Migration Time with RSA(ms)

400

230 170 200

200 100

100

50

0 1

10

250

300

150

100

1000

10000

100000

Size of service requests(in Kbps)

Figure 6: Performance Evaluation Results that WS results are delivered through the RSA. The line labelled “NO RSA”, means that no such agents are provided. It can be observed, from Figure 6, that agents in the “With RSA” system have constantly less migration time from the “No RSA” system. Moreover, it can be observed that as the size of the service results increases the previously stated difference becomes more obvious, and the “With RSA” system performs better than the other systems. XI. CONCLUSION This paper highlights on proposing an intelligent agent that effectively handles online business transaction by proportionately endorsing the ability of digital geography that enhances the way in which business transactions can be done effectively with reference to the service sector in the tourism industry of India. Future work includes the study of agent mobility. The agent infrastructure that has been proposed takes network events into account. Network events (e.g., node failures) occurring while the service invocation is underway, may force the agent to dynamically reschedule its itinerary accordingly. The agent will implement routing algorithms that Sathyabama University

333 326


ARM based Advanced Microcontroller Based Architecture Mrs.Kavitha.v#, Mrs.P. Meena priya Dharshini*

ECE Department [email protected] CMR institute of Technology,Bangalore,India [email protected]

Abstract- The latest advancement in semiconductor technology is the exponential increase in the number of transistors in a chip. This allows us to implement more functions on a single chip. But, the problem that arises due to the increase in number of transistors is the complexity and time to market. Even though the complexity is high the time to market and the cost should be less. For this to happen we have a concept of IP reusability. This created demand for widely accepted bus architectures. AMBA or the Advanced Microcontroller Bus Architecture was thus introduced. The AMBA is a bus that enables IP re-use and thus meeting the essential requirements of flexibility and compatibility. This paper discusses a bus protocol called the AMBA (Advance Microcontroller Bus Architecture) from ARM.

The Advanced Microcontroller Bus Architecture (AMBA) protocol is an open standard, on-chip bus specification that details a strategy for the interconnection and management of functional blocks that makes up a system-on-chip (SOC). There are three distinct buses that are defined within the AMBA specification: • The Advanced High-performance Bus (AHB) • The Advanced System Bus (ASB) • The Advanced Peripheral Bus (APB). The Advanced High-performance Bus (AHB) is a high performance bus used for high speed devices. This bus has a high clock frequency. The AHB allows connection between processors, external memory, some peripheral devices etc.

Keywords—AMBA, microcontroller, master, slave, protocol

I.

INTRODUCTION

The Advanced System Bus (ASB) is a system bus used for connection between processor, external memory, some peripheral devices etc. This is an alternative bus where high performance features are not required.

The latest advancement in semiconductor technology is the exponential increase in the number of transistors in a chip. This allows us to implement more functions on a single chip. But the problem that arises due to the increase in number of transistors is the complexity and the time to market. Time to market is that which allows the company to sell the product or in this case the chip into the market.

The Advanced Peripheral Bus (APB) is for low peripheral devices. This APB is optimized for minimal power consumption and reduced complexity. APB is used along with the ASB or AHB.

Even though the complexity is high the time to market and the cost should be less. For this to happen we have a concept of IP reusability. IP reusability means that the existing design can be reused for a new application. This created demand for widely accepted bus architectures.

II.

A TYPICAL AMBA BASED MICROCONTROLLER

In this section a typical AMBA based microcontroller consists of a system bus that can either be an AHB or ASB as shown in the figure below.

The Bus architecture describes interfaces for the components placed on the bus. These components may be memory, processors, peripherals etc. It also provides protocols for on-chip communication and signals transmission between the components. It also reduces cost of verification because standard test methodology will also be described along with bus architecture. AMBA or the Advanced Microcontroller Bus Architecture was thus introduced. The AMBA is a bus that enables IP re-use and thus meeting the essential requirements of flexibility and compatibility.

Figure 1 A typical AMBA based microcontroller Sathyabama University

334


The AHB or ASB bus can be used to connect processors, external memory, RAM, DMA controller etc. The APB bus is used to connect the various peripheral devices like the keyboard, printer, timers etc. A bridge is used to connect the high performance AHB bus to the less performance APB bus. The AHB bus has a higher bandwidth compared to the APB bus. Hence a bridge is a de-facto in between the two buses. It can either be an AHB to APB bridge or an ASB to APB bridge. III.

The slave’s responds back to the master during a read or write operation. The slave also responds to the active master whether the transfer was a success or failure.

Block Diagram of an AMBA system

The following figure shows a typical AMBA system. It consists of the following components Master, Slave, arbiter and Decoder.

Figure 4 A typical AMBA AHB Slave The decoder is used to decode the address for each transfer. It also provides the select signal to the slave involved for transfer.

Figure 2 Block Diagram of an AMBA system The AMBA AHB system consists of many master and many slaves. It is used for on chip communication that is within a SoC. As shown in the above block diagram, we consider 3 masters and 4 slaves. The master can be any processor or RAM etc. A slave can be a keyboard, printers, external memory etc.

The arbiter ensures that only one master can access the bus at a time for performing an operation. The arbiter protocol is generally fixed, but any other algorithm can also be used such as the highest priority or fair access can be implemented. The AMBA AHB is a new generation bus that is high performance system bus. It supports multiple masters as well as high bandwidth operation. AMBA AHB has the following features

AMBA AHB system design contains the following components:

burst transfers split transactions single-cycle bus master handover single-clock edge operation Wider data bus configurations (64/128 bits).

The master will be able to perform the read and write operation with the help of some control signals to perform these operations. The below figure show the interface diagram of a AMBA AHB Master and the Slave. All the signals that the master and slave utilizes during a read or a write transfer is specified in this diagram.

All signals are prefixed with the letter H, thus making the AHB signals different compared to the other signals.For example HCLK, HRDATA, etc. But for an ASB the signals are prefixes with the letter B. And for APB it is prefixed with the letter P. IV.

HOW DOES THE AMBA AHB MASTER WORK?

The AMBA Master is used to perform a read and write operation by using the control and address information. Only one master can utilize the bus at a time. A master has to use the bus, at which point it sends the request to the arbiter. The arbiter checks to see if the bus is free. If the bus is free then it grants

Figure 3 A typical AMBA AHB Master


335


the master to use the bus. If the bus is not free then the master has to wait and again after sometime request for the use of the bus.

If we want to introduce wait states the signal called HREADY can be made low. Suppose at some point the slave is not ready to accept the write data, maybe it is doing another process. So some wait states are introduced during the transfer.

When HREADY is high it causes the transfer to take place. When this signal is low, then the transfer can be extended for more number of clock cycles.

The main advantage of the AMBA protocol is that it is able to transfer data in the form of bursts. This allows faster transfer to take place. The burst operation can be either of four, eight, sixteen bit transfer in an AMBA AHB protocol. It can also be of undefined length transfer and also single type transfer. There are many advantages by using the burst transfer. This allows us to reduce the time taken for each consecutive transfer.

This can be referred to as wait states. This condition occurs when the master transfers a data but the slave is not ready to accept it, so some wait states are introduced. As soon as the master gets the access of the bus it checks to see if the transfer is read operation or a write operation. There is a 1 bit signal to indicate this called the HWRITE. If this signal is asserted it means that it is a write operation else it will be a read operation. For a write operation the master writes some data on the specified address location. The address is placed on the 32 bit address bus. The write data is carried on the 32 bit data bus given as HWDATA. This data will be written at the address location. This is a simple write transfer with no wait states.

When you want to access a single memory location at a time, what you do is that you just specify the memory location (and the data to be written in case of a write operation) and that it will be done. Again if you want to access another location you do the same. You know when you give address this way (one at a time), then each time the access will take same time. But since we know memory accesses are very slow compared to CPU speed, so a lot of time is wasted in case you want to read continuous memory locations and you have to provide those continuous locations one at a time.

The figure below shows a basic read and write transfer of an AMBA AHB protocol.

If the memory read access takes say N memory cycles then if you want to read say M locations and you provide one address at a time then it will take N*M memory cycles for the entire process. But this takes a lot of time. Therefore it is better to go for BURST MODE, where you just need to give input once and then increment the address when data are continuously accessed.

Figure 5 Basic Write Transfer of Master

Figure 7 Four-beat incrementing burst Figure 6 Basic Transfer for read operation


336


The above figure is a graph of a four – beat incrementing burst transfer. Suppose there are many slaves named as Slave 1, Slave2, and Slave3 etc. Each slave has a particular address.

In the above figure we can see that when it’s a burst transfer it is a sequential transfer type. If the signal is not in burst mode it is indicated by nonsequential transfer type. But for the figure below the transfer type is nonsequential as its not in Burst format.

Figure 8 Eight-beat incrementing burst So when a transfer has to take place the master sends the address of that particular slave and then the slave responds back as ready. And then the transfer takes place, it can be a read operation or write operation. If the next transfer is by the Slave2 and then by Slave3 we can call this as consecutive transfer. Instead granting the bus to master and then receiving the ready signal we can just increment the address so that the transfer is continuous. Thus improving the speed and also reducing the time needed for the transfer.

Figure 10 Transfer Response nonsequential Once the transfer is complete the slave shows the response of the transfer. The transfer can either be a success or a failure. If it is a success the slave gives a OKAY signal. If it’s a failure the slave sends an ERROR signal and then sends a RETRY or a SPLIT signal. This signal tells the master that the transfer could not complete and it has to continue to attempt transfer.

This type of transfer is called Increment transfer given as INCR in the AMBA AHB protocol. There is another type of transfer called the wrapping bursts given as WRAP in AMBA AHB protocol. For wrapping bursts, if the starting address is not aligned with the total number of bytes then the address wraps itself when the boundary has reached. For example, if the start address of the transfer is 0x34, then it consists of four transfers to addresses 0x34, 0x38, 0x3C and 0x30. Every transfer can be classified into one of four different types, as indicated by the HTRANS[1:0] signals. The type of transfer can be either sequential or nonsequential. Sequential transfer indicates the beginning of a burst operation.

Figure 11 Slave Response OKAY In a normal operation the master has to complete all its transfer if it’s a burst operation before the arbiter grants the bus to another master. But there are chances where the arbiter can break the burst and grant the bus to another master. In such cases the master has to request to the arbiter again to complete its transfer. Each transfer will have a number of control signals that provide additional information about the transfer. These control signals have the same timing as the address bus. However,

Figure 9 Transfer Response sequential


337


V.

they are constant throughout a burst of transfers. The size of the transfer is very important. The data transferred can either be an 8 bit data or 16 data and so on. The size is indicated by the signal HSIZE[2:1], which is a 2 bit bus.

In this paper the various signal of an AMBA AHB protocol is discussed and analyzed. The signals are simulated in verilog to study the functionality and behavior of the signals of AHB Master and Slave. Some of the applications of AMBA bus architecture is far beyond microcontroller devices, and is now widely used on a range of ASIC and SoC parts including applications processors used in modern portable mobile devices like smartphones.

The signal for knowing the size of the data is given by HSIZE[2:0]. This signal is a 3 –bit value. The transfer can be either of 8bit, 16bit, 32bit. In this project I will be showing a 32bit transfer. After a master has started a transfer, the slave then determines how the transfer should progress. No provision is made within the AHB specification for a bus master to cancel a transfer once it has commenced. Whenever a slave is accessed it must provide a response which indicates the status of the transfer. The HREADY signal is used to extend the transfer and this works in combination with the response signals, HRESP[1:0], which provide the status of the transfer.

REFERENCES [1] AMBA Specification Rev 2.0 [2] http://www.arm.com [3] A High Throughput AMBA AHB Protocol MS. USHA A. JADHAV, PROF. M. M. JADHAV [4] Using formal techniques to Debug the AMBA System – on – chip Bus Protocol Abhik Roychoudhury, Tulika Mitra, S.R. Karri School of Computing National University of Singapore [5] www.auroravlsi.com/product_briefs/aub3000_brief.pdf [6] Verification of AMBA Using a Combination of Model Checking and Theorem Proving - Hasan Amjad – Computer Laboratory – University of Cambridge

The slave can complete the transfer in a number of ways. It can: complete the transfer immediately insert one or more wait states to allow time to complete the transfer signal an error to indicate that the transfer has failed delay the completion of the transfer, but allow the master and slave to back off the bus, leaving it available for other transfers. IV ADVANTAGE By using amba ahb bus reducing the cost of verification and reduce time to market High-performance, and high clock frequency system modules. AHB supports the efficient connection of processors, on-chip memories and off-chip external memory interfaces with low-power peripheral macrocell functions. AHB is also specified to ensure ease of use in an efficient design flow using synthesis and automated test techniques.


CONCLUSION

[7] AU-B3000: AMBA AHB Bus Master Core – Aurora VLSI [8] AHB Example AMBA System Technical

338


Zigbee Based Wearable Personal Healthcare and Emergency Aid System Venkata Ramesh Mamilla1, N.Venkata Ramana2,Dr.G.Lakshmi Narayana Rao3 1Associate Professor, 2Assistant Professor,3Principal 1,2 QIS College of Engineering &Technology QIS Institute of Technology, Ongole,Andhra Pradesh Abstract: This paper deals with the design of the hand held hardware with sensors, Zigbee transceiver based wearable personal healthcare and emergency aid system, which can continuously monitor personal body status in a real-time manner and automatically issue the alert for medical aids in case of emergency. This paper also focused about the power supply selector circuit for peripheral and controller in system. To write the programmer for the main controller using embedded ‘c’for 8051.This paper also involves about the design of a driver for sending or receiving message. Key words: Zigbee transceiver, Sensors, Power supply, Controller, Embedded I INTRODUCTION For patients whose health is on the line the benefits are even greater. They have increased access to specialized doctors. They don't have to stick around the hospital any longer. This ease in mobility allows them to do their own work while still under the doctor's care. Safety is another issue that is helped here because the rate of mistakes can significantly be decreased. Also, patients can be picky when making changes in their daily lives when signing up for a treatment. With Wireless Technologies, their healthcare can be less intrusive, for example in the case of wearable sensors [Malan04]. They don't have to show up to the hospital for a blood pressure check. It can be done while they are working through wearing wireless sensors that transmit this information in real-time to their doctors. Negligent mistakes by doctors and nurses costs hospitals, insurance companies and the government a large sum of money each year. These costs can be reduced by reducing the number of mistakes. Efficient and secure data handling and resource management is another area where wireless networks can help. Deploying huge machines around the hospital can be very expensive and time consuming. With Wireless Technology interfaces can be designed such that access to the machines can be provided from anywhere in the hospital. This allows for rapid and flexible deployment. By increasing the doctors' and nurses' efficiency, Sathyabama University

339

hospitals can provide care for more patients and increase their profits. Sensor Networks technologies such as Zigbee [ChevrollierJuly05] are being combined with WBANs [JovanovMar05] to form smaller scale networks that can be placed on human clothing (or other objects) and provide unobtrusive access to their health information. Sensor Networks are also increasingly being used in natural sciences for example in monitoring wild life or other natural phenomenon. Due to lower power requirements they can be deployed for a long period of time. Due to limited range, they are deployed in large numbers and thus form a distributed network covering a large portion of space. A good example of an application of sensor networks in the medical field is the Code Blue project [Malan04] being developed at Harvard. Other experimental applications include forest fire detection and path tracking using ad hoc sensor networks [Fok04]. Since sensors networks devices are very cheap, they can be deployed anywhere in large numbers. Some of the wireless sensor networks based devices are very sophisticated. They operate on their own operating system called the TinyOS. Therefore they can be programmed over the air, making their management very easy. 2.0 PIC16F877A MICROCONTROLLER

Figure 1: Microcontroller


3.1 Features:

2.1 Features: Operating Speed Max 20 MHz, Voltage-(2-5.5)v

Output Current up to 1A

Memory:

Output Voltages of 5, 6, 8, 9, 10, 12, 15, 18, 24V

Flash Program RAM

8Kx14 Words, 368 Bytes,

Thermal Overload Protection Short Circuit Protection

EEPROM Data Memory 256 Bytes Low power, Technology

High

speed

Flash/EEPROM

Output Transistor Safe Operating Area Protection 4.0 TEMPERATURE SENSOR LM35

It has 5 Ports for Internal and External usage It has on chip Timers. 3 Timers are avail It has in built Analog to Digital Converter In built Multiplexer availability for signal Selection It has serial as well as Parallel Communication facilities

The LM35 series are precision integrated-circuit temperature sensors, whose output voltage is linearly proportional to the Celsius (Centigrade) temperature. LM35 temperature sensor is used to measure the temperature.

In built Capture, Compare and Pulse width modulation 1.0 POWER SUPPLY

Figure 3: LM35 sensor This section gives step-by-step instructions along with photos to the construction of IR Proximity Switch. Because this is a very simple circuit, only a schematic for the sensor is shown in Figure 3.

Figure 2: power supply Power supply is a reference to a source of electrical power. A device or system that supplies electrical or other types of energy to an output load or group of loads is called a power supply unit or PSU. The term is most commonly applied to electrical energy supplies, less often to mechanical ones, and rarely to others. A 230v, 50Hz Single phase AC power supply is given to a step down transformer to get 12v supply. This voltage is converted to DC voltage using a Bridge Rectifier. The converted pulsating DC voltage is filtered by a 2200uf capacitor and then given to 7805 voltage regulator to obtain constant 5v supply. This 5v supply is given to all the components in the circuit. A RC time constant circuit is added to discharge all the capacitors quickly. To ensure the power supply a LED is connected for indication purpose.


340

5.0 HEART BEAT SENSOR The heart beat sensor, also named pulse sensor, can detect the heart pulse non-invasively with the Photoplethysmography (PPG) technique, where a pair of infrared light emitter and detector capture the variations of the reflected light by the body surface. The wavelength of the infrared sensor was around 900 nm to optimize the measurement. This design has adopted reflective sensing instead of penetration method, which can reduce power consumption than human body penetration and be applied on wider parts on body, such as on the wrist and neck.


within the ISM 2.4 GHz frequency band and are pin-for-pin compatible with each other. 7.0 PRINCIPLE OF WORKING & METHODOLOGY The sensors used to get the parameters of human being. The parameters are converted into digital value by on chip Analog to Digital Conversion(ADC) module of the micro controller. The micro controller sends the value to monitor section via Zigbee transceiver. In monitor section the computer compare the value with predefined value. If the value is exceeds, the GSM send the message alert .

Figure 4: Heart beat sensor 5.1 Heart beat measurement

7.1 Transmitter section Sensor1 Figure 5

SCU MCU

Sensor2

The skin may be illuminated with visible (red) or infrared LEDs using transmitted or reflected light for detection. The very small changes in reflectivity or in transmittance caused by the varying blood content of human tissue are almost invisible

MAX232

Zigbee Transceiver

Figure 7 : Block diagram of Transmitter section of a System 7.2 GSM MODEM

6.0 ZIGBEE

Figure 8 : GSM Modem

Figure 6: ZIGBEE The XBee®/XBee-PRO OEM RF Modules interface to a host device through a logic-level asynchronous serial port. Through its serial port, the module can communicate with any logic and voltage compatible UART; or through a level translator to any serial device (For example: Through a Digit proprietary RS-232 or USB interface board). The XBee and XBeePRO OEM RF Modules were engineered to meet IEEE 802.15.4 standards and support the unique needs of low-cost, low-power wireless sensor networks. The modules require minimal power and provide reliable delivery of data between devices. The modules operate Sathyabama University

341

GSM Modem provides full functional capability to Serial devices to send SMS and Data over GSM Network. The product is available as Board Level or enclosed in Metal Box. The Board Level product can be integrated in to Various Serial devices in providing them SMS and Data capability and the unit housed in a Metal Enclosure can be kept outside to provide serial port connection. The GSM Modem supports popular "AT" command set so that users can develop applications quickly. The product has SIM Card holder to which activated SIM card is inserted for normal use. The power to this unit can be given from UPS to provide uninterrupted operation. This product provides great feasibility for Devices in remote location to


stay connected which otherwise would not have been possible where telephone lines do not exist

Software Tools: Embedded C

7.3 Application areas 7.0 CONCLUSION Mobile Transport vehicles. LAN based SMS servers Alarm notification of critical events including Servers Network Monitoring and SMS reporting Data Transfer applications from remote locations Monitor and control of Serial services through GSM Network Integration to custom software for Warehouse, Stock, Production, Dispatch notification through SMS.

In this paper, a novel wearable personal healthcare and emergency aid system, which can continuously collect personal health status, periodically send the status reports to healthcare centre, and rapidly issue the alerts for medical aid in case of emergency. The system can be attached on human body but will not affect the users’ daily life and activities. Employing Bluetooth devices and on-site data processing and storage on a mobile phone, this system can effectively reduce the product cost and communication energy consumption.

AMR- Automatic Meter Reading

8.0 REFERENCES

7.4 Monitoring Section

[1] “Hong Kong Population Projections 2004–2033”, The Government of the Hong Kong Special Administrative Region, Census and Statistics Department, June 2004.

Zigbee Device

[2]http://app.hkatvnews.com/content/2006/09/28/atvnews 94 990.htm. [3]http://www.rthk.org.hk/rthk/news/expressnews/20050 509/ 20050509_55_225452.html

GSM [4] Elite Care, http://www.elite-care.com/

Figure 9 : Monitoring Section The system does not need any specific caregivers and is easily used. System employs the body temperature sensors to continuously collect user’s vital signals. Also,the use tiny zigbee device is to transmit the raw sensory data. From the users’ point of view, the device attached on the body is just a zigbee set. This will not cause inconvenience to the users’ life. The use a zigbee transceiver is to receive the sensory data from the sensors to perform on-site data processing and storage, which avoid the continuous connection to the healthcare centre and reduce transmission cost. The mobile phone can use the attached GSM module to periodically send the health reports to the medical centre and issue timely alerts for medical aid in case of emergency. So that System can satisfy the requirements of personal healthcare and emergency aid system in an effective ,simple , and low-cost manner.


342

[5] D. Malan, at el. “CodeBlue: An Ad Hoc Sensor Network Infrastructure for Emergency Medical Care”, International Workshop on Wearable and Implantable Body Sensor Networks, Apr. 2004 [6]http://www.schsa.org.hk/eng/service/pel_network. Htm [7]

U. Anliker, at el. “AMON: A wearable multiparameter medical monitoring and alert system”, IEEE Transactions on Information in Biomedicine 8 (2004),415-427.

[8] N. Chevrollier and N. Golmie, "On the Use of Wireless Network Technologies in Healthcare Environments," Proceedings of the fifth IEEE workshop on Applications and Services in wireless networks (ASWN 2005), June 2005 Paris, France, pp. 147- 152.


A Sensor Based Real-Time Monitoring System for Heartbeat and Respiration Rate Sadiq Ali1, Satheesh2 12

Electronics and Communication Engineering, Ramappa Engineering College Warangal,India 1 [email protected] [email protected]

Abstract—A remote compact sensor system for the detection of human vital signs (heartbeat and respiration rate) is presented. The frequency band of 24 GHz is employed for remote sensing. For the compact size, the developed sensor uses a circularly polarized electromagnetic wave with a single antenna. The sensor system is composed of radio-frequency circuits, a signal conditioning block, a data-acquisition unit, and a signalprocessing part. The peak detection of the power spectral density with a tracking algorithm is utilized for the real-time detection of human vital signs. The measurement result is compared with the commercial fingertip sensor. The comparison result shows excellent agreement.

than communications. It is appropriate to use such a high-frequency band for the development of the compact sensor system as shown in Fig 1. In this paper, the remote compact sensor for the detection of human vital signs is presented. To reduce the sensor size, the circularly polarized EM wave is utilized. The design of RF circuit blocks and their measurement results will be discussed. And base band blocks, such as a signal conditioning block and signalprocessing algorithm, will be explained in detail. For the validity purpose of the developed sensor, the measurement results will be compared with the commercial fingertip sensor.

Keywords— PDA,EM waves,WLAN,ECG signal

I. INTRODUCTION RECENTLY, the health-care sensor for the remote monitoring of the human heartbeat and respiration rate receives an attraction [1]. Presently, most commercial sensors have to be attached to the human body. It is inconvenient to use in daily life. The remote monitoring of human cardiac and respiratory activities are desirable since they help treat patients in emergency circumstances. This remote sensor can be equipped in the home for long-period monitoring of the patient and in the bed for managing comfortable sleeping. If the sensor is applied to the mobile application, the sensor needs to portable and compact, maintaining the accuracy of the detection. For the ex vivo detection of the vital signs, the radio frequency (RF) can be used. By transmitting the RF signal to the heart and receiving the reflected waves, the heartbeat and the respiration rate can be detected. In previous reports, 2.4 GHz, 5 GHz, and 10 GHz have been used for the detection [2]. The 5-GHz RF architectures in [3] and [4] have utilized the double-sideband transmission and employed separate antennas for transmission (TX) and reception (RX). The advantage of the double-sideband transmission was discussed in [5]. The architecture used two separate free-running oscillators and employed linearly polarized waves. Two separate antennas operating at 5 GHz inherently increase the size of the radar system. For the compact size of the antenna, the operating RF frequency has to be increased since the antenna size is inversely proportional to the operating frequency. The frequency band of 24 GHz can be considered, which is reserved internationally for industrial-scientific-medical (ISM) purposes other Sathyabama University

Figure 1. Patient Monitor System Architecture

II. METHODS A.

Overview of the System The aim of this study is to design and implement a mobile system for monitoring vital signs, and to facilitate the continuous monitoring of patients during transport. Fig. 2 shows the architecture of the proposed system. The telemedicine system consists mainly of two parts—1) the mobile unit, which is set up around the patient to acquire the patient’s physiological data, 2) the management unit, which enables the medical staffs to telemonitor the patient’s condition in real-time. The management unit is from either a fixed computer within an existing hospital network or a mobile laptop via WLAN. The major design requirements of the mobile unit are as follows:

343


1) it should be portable and lightweight, which means easy to carry; 2) it should have power autonomy of more than 60 min to support patient transport; 3) it should have a user friendly interface; 4) it should collect and display critical biosignals, including threelead ECG, HR, and SpO ; 5) it should record patient information and data; and 6) it should support wireless communication. On the other hand, the design requirements of the management unit are as follows: 1) it must have an easy-to-use interface; 2) it must display critical biosignals and analysis of data; 3) it must record, retrieve, and manage patient information and data (local database); and 4) it must be

B. Mobile Unit The The mobile unit in this study is comprised of a designed vital-sign signals acquisition module and a Pocket PC (HP iPAQ H5450). Multiple vital-sign parameters, which include the three-lead ECG, SpO , and HR, can be measured by this unit. Fig. 2 shows the design architecture of this mobile unit. This signals acquisition module acquires the three-lead ECG and dual-wavelength photoplethysmographic (PPG) signals, and converts them into digital data. Through an RS232 connection, the Pocket PC receives the physiological data and computes the SpO and HR parameters. According to user commands, the mobile unit can display waveforms in real-time, store data locally, and trigger an alarm. With regard to remote monitoring, the Pocket PC transfers these physiological data to a remote management unit in real-time by its built-in WLAN device. 1) Module for Acquiring Vital-Sign Signals: Fig. 3 shows the diagram of the designed vital-sign signals acquisition module. The vital-sign signals acquisition module consists of ECG signal conditioning circuits, pulse oximeter analog circuits, and a microcontroller. This module is powered by

Figure 2. Architecture of wireless telemedicine system

connectable to the Internet to transmit data and distribute information. Furthermore, at the consultation terminals such as wireless PDAs or laptops, the medical staffs can use them either to monitor the physiological parameters and waveforms of a remote patient online or to access his or her case history through the wireless connection to the management unit. Wireless connection in the studied hospital has been established by WLAN technology (IEEE 802.11b) [6] with speeds up to 11 Mb/s. An access point acts as a wireless bridge for the network data to be transmitted to and received from the existing wired hospital network. With multiple access points linked to a wired network, it allows efficient sharing of network resources throughout an entire building. The distance set between each access point was less than 30mbecause of the radius of indoor coverage for typical WLAN and regional geography limitation. Devices with WLAN interface can roam among the access points. The transmission of data between a mobile unit and a management unit is implemented by the client server architecture. In the proposed design, the mobile unit serves as the client end and the management unit serves as the server end. Communication depends on the transmission control/Internet protocol Sathyabama University

Figure 3. Design Architecture in Mobile Unit

four rechargeable AA batteries and is packaged as a jacket of the Pocket PC. The core control unit of the module is an 8-bit microcontroller, PIC16F877, which has an on-chip eight-channel 10-bit analog-to-digital converter (ADC). The three-lead ECG signals were amplified with a gain of 700, filtered (0.5–50 Hz), and then fed into the inputs of the ADC in the microcontroller. The pulse oximeter analog circuits were designed based on the principles of spectrophotometry and optical plethysmography to measure SpO [7], and a Nellcor oxygen sensor (DS100A, finger probe) was used to measure the PPG signals. The signals determined by two light-emitting diodes (infrared and red) are first demultiplexed, then separately amplified, and finally separated into dc and ac components (IRAC, IRDC, RAC, and RDC), which are used to calculate pulse rate and the oxygen saturation in the blood. The microcontroller digitizes the signals with a sampling frequency of 200 Hz and transmits the ECG and PPG data to the Pocket PC through the serial port. The baud rate is 115.2 kb/s. Optical coupling is used in the serial communication

344


to separate the power supply of the signal acquisition module from that of Pocket PC, reducing power interference. 2) Program on the Pocket PC: A system program, developed by Microsoft embedded visual C++, was installed on the Pocket PC to monitor the vital signs. This program records users’ information and displays the HR, SpO , ECG, and PPG waveforms sent from the signal acquisition module. Raw data can be stored into the built-in memory of the Pocket PC and transmitted to a remote management unit via the WLAN. In long-term store-and-forward mode, the raw data are stored into the extended secure digital (SD) memory (256 MB) of the Pocket PC. The waveforms are plotted in window with an area of 200 150 pixels. The amplitude resolution is 0.04 mV/pixel for the three-lead display and 0.0125 mV/pixel for the singlelead display. When the frame displays 4 s of ECG data, the temporal resolution is 0.02 s/pixel. Besides, the sound reflecting each heart beat can be pronounced by the speaker of Pocket PC. In addition, this program is installed in the medical staffs’ PDAs for receiving and displaying the physiological parameters and waveforms of a remote patient under monitoring through the wireless connection to the management unit.

receives the data from the mobile unit, displays HR, SpO , three-lead ECG, and PPG waveforms on the terminal screen, and stores the data in the local database. In this work, a MySQL database system is set up to manage the raw data of ECGs and PPGs, patients’ information, and the doctors’ diagnosis. The database can also be accessed from authorized terminals through the hospital network and the Internet. Moreover, the vital-sign signal can be delivered in real-time to a mobile platform for sharing data. The waveforms are plotted in a 600 448 pixels window, which shows 6 s of ECG data. The default resolutions of amplitude and time are approximately 0.015 mV/pixel and 0.01 s/pixel, respectively. The program also supports the selection of leads, the replay of waveforms, analysis of raw data, and the scaling of amplitude and time. Both mobile unit and management unit have an alarm setting window which enables the medical staff to set up the alarm threshold of SpO and HR individually according to the physiological status of the patient. When the recorded vital signs are beyond the preset limits, the mobile unit would trigger an alarm automatically and a warning message window will pop-up on the screen. D. Evaluation of System The system was evaluated in the following phases. 1) Technical Verification: First, the developed pulse oximeter was calibrated by an index pulse oximeter simulator (Bio-Tek product; SpO range: 35%–100%; HR range: 30–250 bpm), whereas the accuracy of the ECG monitor was verified by the medSim 300 Patient Simulator (Dynatech Nevada, Inc.). Then, the functions of the PDA-based pulse oximeter and the ECG monitor, as well as the transmission of data between the mobile unit and the central management unit

Figure 4. Architecture of the Management Unit

C. Management Unit Fig. 5 shows the architecture of the management unit. The management unit consists mainly of a fixed personal computer or a laptop, and the management program. The management unit can be set in many spots depending on different applications of telemonitoring. It is normally located in the nurse’s station, and provides a user-friendly interface for telemonitoring a patient’s vital-sign signals. The management terminal can receive patients’ physiological data from the remote mobile units via the WLAN or the Internet. The management program is implemented on a Windows 2000 platform and developed by the Borland C++ builder. The program Sathyabama University

Figure 5. Architecture of the Management Unit

were tested. Twenty healthy volunteers, including eleven males (with an average age of 29.7 11 years old) and nine females (with an average age of 29.6 10 years old), were involved in the test. Three-lead ECG signals and PPG signals were acquired simultaneously. All results were recorded locally and were transmitted to the remote central management unit for 5 min to confirm the quality of the signals and the error rate of data transmission between the two units. Two different probes, one of the designed pulse oximeter and the other of the commercial pulse

345


oximeter BCI-3304 (product of BCI, Inc.) were connected to different fingers of the same volunteer and then operated simultaneously to compare SpO and HR readings over 5 min. During the first minute, the volunteers breathed normally. They were then required to hold their breath for one minute, and then breathe normally until the end of the test. 2) Clinical Test and User Survey: The complete system was demonstrated at National Taiwan University Hospital (NTUH). In the test scenario, each patient was transported from the intensive care unit (ICU) to a radiographic examination room. The mobile unit was placed beside the patient, which enables the medical personnel to observe the patient’s physiological condition and check the connection of the electrodes. The mobile unit transmits the patient’s vital-sign signals to the management unit via the WLAN, allowing medical staffs to monitor online the patient’s data during transport. During the radiographic examination, the mobile unit was placed next to the patient, and a laptop was set up at the control station to monitor the real-time data from the mobile unit. According to the test scenario, a survey was conducted to elicit the operators’ opinions on the wireless PDA-based physiological monitoring system in three areas—1) mobility (size and weight), 2) usability, and 3) performance of the overall system on intrahospital transport. A questionnaire with a fivepoint Likert scale (from 5 = completely satisfied to 1 = completely unsatisfied) was used to rate the performance of the overall system on intrahospital transport. Also, the mobility and usability of the wireless PDA-based monitoring system were compared with the currently used monitoring device (Agilent M3046A) in intrahospital transport at NTUH. The satisfaction of mobility was evaluated in relation to two statements of weight and size and that of usability was estimated by easy operation and easy monitoring. The outcomes of these four statements were represented as a ten-point scale (from 1 = completely unsatisfied, to 10 = completely satisfied). Intrahospital transport scenarios were tested over a one month period in the emergency department. Fifty medical personnel, including 30 nurses and 20 doctors, used the wireless patient monitoring system and answered the questionnaire. The staffs included 14 males and 36 females with ages in the range 23–50 years old (mean SD ) and with 1–25 years (mean SD ) of experience in emergency medical care.

end IC is designed. It is comprised of LNA, singleended mixer, and Lange coupler. LO is applied using an external signal source by which the frequency and the output power can be adjusted. LNA is implemented with 3-stages. At first stage, the transistor is degenerated with microstrip line inductor to match the gain and the noise optimum point at the same time. 2nd and 3rd stage is designed without degeneration to increase the gain. The designed LNA satisfies unconditionally stable condition. To downconvert he reflected signal to baseband, singleended mixer is designed. Single-end mixer operates according to square law due to the nonlinearity of the active device. Therefore, the base of the transistor is biased near pinch-off to increase the nonlinearity of the transistor. A capacitor is connected in parallel with the load resistor to suppress high frequency components such as LO and RF leakage signals at output port. To reduce the loading effect, emitter follower is used as a buffer amplifier. LNA and mixer were simulated with Agilent ADS 2002C. Lange coupler for a polarizer is designed using 2.5D EM simulator of Agilent Momentum 24 GHz Doppler radar receiver front-end IC was fabricated using 6inch InGaP/GaAs HBT HS process at Knowledge*on foundry. The large signal model of the transistor was performed using VBIC (Vertical Bipolar InterCompany) model. IV. RESULTS

Figure 6. ECG signal and its spectrum signal

. Figure 7. Respiration Rate spectrum signals (a) with error at the 60 Hz signals (b) without error at any frequency

III.

RADAR RECIVER AND FRONT -END RECIVER AND FRABICATION

To demonstrate the proposed circularly polarized radar system, 24 GHz Doppler radar receiver frontSathyabama University

V. CONCLUSION A mobile patient monitoring system was designed, developed, and tested. A pulse oximeter was

346


integrated with a threelead ECG monitor on a wireless PDA platform, which provides real-time and storeand-forward modes. The monitor in the new system has a significantly reduced size and weight, and thus, improves the portability of the monitoring device. Besides,WLAN also greatly increases the flexibility and usability for telemonitoring. The clinical evaluation reveals that this mobile patient monitoring system is user-friendly, convenient, and feasible for patient transport.

[3]

Y. Xiao, C. Li, and J. Lin, “A portable noncontact heartbeat and respiration monitoring system using 5-GHz radar,” IEEE J. Sensors, vol. 7, no. 7, pp. 1042–1043, Jul. 2007.

[4]

C. Li, Y. Xiao, and J. Lin, “A 5 GHz double-sideband radar sensor chip in 0.18um CMOS for non-contact vital sign detection,” IEEE Microw. Wireless Compon. Lett., vol. 18, no. 7, pp. 494–496, Jul. 2008.

[5]

C. Li, Y. Xiao, and J. Lin, “Experiment and spectral analysis of a low-power Ka-band heartbeat detector measuring from four sides of a human body,” IEEE Trans. Microw. Theory Tech., vol. 54, no. 12, pp. 4464–4471, Dec. 2006.

[6]

M. S. Gast, “802.11 Wireless Networks: The Definitive Guide,”O’Reilly, 2002.

[7]

J. P. de Kock and L. Tarassenko, “Pulse oximetry: Theoretical and experimental models,” Med. Biol. Eng. Comput., vol. 31, pp. 291–300, May 1993.

REFERENCES [1]

M. Steffen, A. Aleksandrowicz, and S. Leonhardt, “Mobile noncontact monitoring of heart and lung activity,” IEEE Trans. Biomed. Circuits Syst., vol. 1, no. 4, pp. 250–257, Dec. 2007.

[2]

J. Clerk Q. Zhou, J. Liu, A. Høst-Madsen, O. Boric-Lubecke, and V. Lubecke, “Detection of multiple heartbeats using doppler radar,” in Proc. IEEE ICASSP, Toulouse, France, May 2006, pp. 1160–1163.


347


Real-Time Application for Data-Acquisition Embedded System Based on Web Server G Anil Kumar1, B.Prasanna Jyothi2 12

Electronics and Communication Engineering, Ramappa Engineering College, warangal,India 1

[email protected]

2

[email protected]

Abstract: In this paper, we present the principles of a low operational cost but flexible Internet-based data- cquisition system. The main core of the system is an embedded hardware running a scaled-down version of Linux: a popular choice of operating system for embedded applications. The embedded device communicates through General Packet Radio Service (GPRS), which makes it accessible from anywhere in the world through a web server built into the embedded device. In addition, GPRS provides a bidirectional real-time data transfer allowing interaction. The proposed system eliminates the need for server software and maintenance. A novel approach is introduced to minimize the operational costs while operating with a large amount of data. The system is demonstrated to be suitable for different embedded applications by attaching several real time modules tharough appropriate interfaces. Keywords— Data-acquisition, embedded interaction, Internet, Linux, real time.

system,

I. INTRODUCTION DATA-ACQUISITION systems with remote accessibility are in great demand in industry and consumer applications. In some applications, human beings have been replaced by unmanned devices that will acquire data and relay the data back to the base [1]. There are data-acquisition and control devices that will be a substitute for a supervisor in a multisite job operation. A single person can monitor and even interact with the ongoing work from a single base station. An acquisition unit designed to collect data in their simplest form is detailed in [2], which is based on Linux [3], which is a popular choice for embedded PC systems. A similar system in [4] provides dataacquisition with no concern for remote access. Data collection for postprocessing on a vehicle’s position for an advanced traffic survey is discussed in [5]. Some applications adding remote accessibility are detailed in [6] and [7], which are built to collect and send data through a modem to a server. Although these are well-built systems that serve the purpose for a specific task, the user cannot interact with the system. Another unidirectional data transfer is presented in [8], which uses the Global System for Mobile Communications (GSM): a popular wireless choice for connectivity between the data-acquisition units and clients. Similar types of Internet-based systems, such as those in [13]– [16], are designed to gather a bulk of data before serving them upon request. In these applications, data are compiled in a central


server and are then served to the clients via the Internet. The client framework is in a central server and has all the applications. A person that needs to access any data must first access the server. An indirect access to the data- acquisition unit makes the system unattractive for real-time control applications, where direct interaction with the system may be required. The need to maintain an additional server will also increase the setup costs and the costs to maintain the acquisition systems, such as regular maintenance costs, system updates, etc. Therefore, the central server has to be eliminated for a realtime system. The closest to this idea is published in [17]. In this system, a reliable bidirectional Point-to-Point Protocol (PPP) link for real-time control and surveillance via a GSM network is formed. However, there is still no effort to minimize the operational costs (including the costs to transfer a large amount of data). In addition, this system is based on an industrial PC, thus making it an expensive solution. Interaction with the embedded unit is also an important issue. In [18], an embedded PC card placed on the Internet allows limitedinteraction through commands sent through Transmission Control Protocol/IP (TCP/IP) and User Datagram Protocol. In this paper, we propose a GPRS-based portable low-cost data-acquisition system, which can establish a reliable bidirectional connection for data-acquisition. The proposed system uniquely reduces the costs occurring from frequently requested data and eliminates the need for a wellestablished server. The system uses a dummy server for static information, thus optimizing the transfer of large data. The user can directly log in and interact with the embedded device in real time without the need to maintain an additional server. The system is modularly built, allowing different modules to be added. In addition, it is flexible to accommodate a wide range of measurement devices with appropriate interfaces. II. INTER-ACTIVE DATA ACQUISITION SYSTEM The general principles of Internet-based control systems have been modeled in [19]. Interactive Internet-based systems provide a way to monitor and adjust using standard web browsers and a PC. The target systems can be monitored andcontrolled

348


independent from the location and the plat form since standard web browsers can be used on the client side. A typical data-acquisition system is made up of three components connected to each other via the Internet, as shown in Fig. 1. The dataacquisition system needs to relay the acquired information to the requesting clients. The clients also need to send commands. If necessary, this is implemented through a server, and then, an enormous amount of data transfer time would be consumed. Thus, alternative methods need to be explored.

Figure 1. Genral diagram of data acquisition and control system

A. Establishing a Direct Communication Link Between the Client and the Embedded Device GSM and GPRS [20] are developed for cellular mobile communication. A GPRS connection with unlimited duration of connectivity is charged only for the data package transfers and adopted in several mobile remote control/access systems [13], [14], [16]. GPRS becomes a cost-effective solution only if the data transfers can be optimized. Once a GPRS connection has been established, queried data can be relayed to the client via a central server [13]–[16]. Using a central server to relay the acquired data has some disadvantages. First, a central server needs a client interface framework. An additional data transfer corresponds to time delays before the data are made available to the client. In addition, since the server acts as a relay, no direct bidirectional communication between the client and the embedded system can be established. This makes the system unsuitable for real-time control applications. The basic idea behind real-time processing is that the embedded system is expected to respond to the queries in time. Real time should be fast enough in the context in which the system is operating and reliable as well. Realtime system correctness depends not only on the correctness of the logical result of the computation but also on the result delivery time [21]. This method also increases the data transfer cost as the number of clients increases due to the access amount of data transfers ia GPRS. Direct communication, on the other hand, enables access to only relevant information in the embedded system by preprocessing the data. The embedded system should also andle the web services. This eliminates the need for a central server and reduces the amount of data sent from the remote unit since only the Sathyabama University

queried data will be transferred. In the proposed system, the GPRS architecture and protocols are compliant with [20]. This system is configured to be virtually online at all times in a GSM network. An admin script is executed after the boot of the operating system, initiating the GPRS connection software module. A PPP connection is established by a GPRS modem that works at 900/1800/ 1900 MHz operating frequencies. A PPP daemon (PPPD) is used to manage the PPP network connections between the client and the embedded module. The PPPD is responsible for setting up the GPRS parameters, such as the connection speed and compression. To directly access an embedded system, the IP address of the embedded device should be made available to the client side. There are two choices available. A static (hard-coded) IP could be used, or the remote device should initiate a connection by reporting its IP. This choice is quite straightforward and simple. Although the usage cost remains unchanged, it requires a static IP setup by the service provider and involves monthly recurring costs. The static IP is preferred for its simplicity in designing a system; however, its overhead may be impractical. The other choice is to use a dynamic IP assigned through a Dynamic Host Configuration Protocol (DHCP) server of the GSM provider for every connection established. However, this IP needs to be known by any client requesting an access to the embedded server. One solution is to broadcast this IP to

Figure 2. Folder structure for FTP server

a dummy FTP server (where the bulky static information such as image data is also kept). The FTP server is a dummy server and does not require regular software updates or maintenance. The folder structure of the FTP server is shown in Fig. 2. A script on the embedded device is configured to update its IP address on the FTP server in Hypertext Meta-Language as an index.htm file, under a folder uniquely named by its hostname. This script simply parses the current IP for that embedded device and sends an htm file with the IP information of the embedded device to the FTP server. Once this file is in place, a direct connection can be established

349


with the desired embedded device by a simple query. An example embedded system, named mozart (Fig. 2), can be queried from the FTP server by a simple command. The web browser processes the (index.htm) file in the specified folder as default; therefore, a file name is not needed for referencing. The DHCP approach is more flexible and works better compared with the static approach as a cost-effective solution, despite the necessity for a script running on the embedded server, one-time broadcasting its IP to the FTP server. The hypertext file placed on the FTP server by the embedded system and queried by the client. With this mechanism in place, the embedded system updates it IP information on the FTP server upon every reboot, which causes an IP refresh from the GSM service provider. B. Data management in the system The Internet server is used to decrease the management costs by sending all the pictures (logo, picture, bar graphics, etc.) to the client through a server on the Internet. Text data such as coordinates, temperature, and altitude are served from the embedded system. If bulky data are going to be

functionality. The acquisition units on the device can be varied with no limitation on their functionality and can be added by using appropriate interfaces.

D. Software and Operating System Choice The Linux 2.4 kernel series [23] with TCP/IP stack included has been chosen as the operating system for the embedded board. Only the bare minimum is installed, including the basics such as console tty, serial ports, kernel side of the PPPD, and support for memory and math emulation. The running kernel is around 1MB of code built into a Flash memory. A scaled-down version of Linux has been used to reduce the memory footprint and the complexity. The software running on the embedded system at the highestlevel is named the manager code, which will be explained in Section III with a sample implementation. In the design, the manager code controls the execution of

Figure 3. FTP server as viewed by the client

sent, the embedded module is set to send the image only once via GPRS and placed on an FTP server. This approach eliminates the transfer of large data through GPRS more than once, thus reducing the transfer costs, particularly if more than one client is involved or multiple requests to the same data are needed, as shown in Fig. 3. A user interface, which is brought up on establishing a direct connection, has links to the Common Gateway Interface (CGI) and Bourne Again Shell (BASH) script files executed on the embedded system. The code is compiled into the CGI format to be installed in the embedded board through a cross-compiler platform [22]. BASH scripts are directly triggered by the applications. C. Hardware The embedded system used in this work is an X86based on and alone unit with four serial ports and a parallel port with16-MB onboard removable Flash memory as shown in the Fig. 4. One of the serial ports is used in the application design stage for debugging purposes, and this port is designed to host more devices with a multiplexer unit. The other serial ports used by the modules are used to test the system Sathyabama University

Figure 4. Block diagram of the embedded system with sample devices attached

other applications and is triggered once all the components of the operating system are up and running. The flowchart of this code is given in Fig. 7, representing a sample operation of the GPS unit that checks the execution if the speed limit has been exceeded. The periodic operations and routine tasks are organized by a manager code. If a new dataaccess application is considered in future developments, its program can easily be added to the manager code as a periodic operation. III. SAMPLE APPLICATION A camera, a temperature sensor, and a GPS are integrated into the embedded board to form a sample application, as shown in Fig. 8. These units and their interaction with the embedded board are briefly explained in the following discussion. In addition, the time delays at each operation stage are discussed to demonstrate the effectiveness of the proposed method. A time delay of 0.57 s, on average, isneeded for any control command to be sent. This delay is related to the GPRS service of the

350


GSM service provider. For this sample application, a complementary metal–oxide– semiconductor camera with a built-in JPEG codec controller chip has been chosen. The camera acquires bulk image data; therefore, it is a good module to demonstrate the effectiveness of the system. It compresses and transfers the image from the camera to the serial port. The communication with the camera is established over an RS232 communication protocol using an asynchronous package transfer method. Before taking a snapshot, the camera is

upload duration. Although this may seem like a large delay, it may be improved with a faster and more expensive camera. The transfer of text data takes an average of less than 1 s ( 1 kB). The solid line in Fig. 9 illustrates the case where the client is served by the embedded system and an FTP server. As the number of clients increases, the data to be served remain constant with respect to the clients for the proposed system since the bulk of data is relayed through

Figure 6. Manager code for operational principles Figure 5. Data management in the proposed system

synchronized by sending an appropriate number of synch data packages. After the synchronization, both the embedded board and the camera wait until they receive an acknowledgement from the other side before sending another request or data. This protocol is executed in an average of 3.4 s for each picture, which can be considered as an adequate rate for most applications. Here, the bottleneck is the camera; hence, the speed of data transfer can further be improved by using a camera with a faster which eventually takes a snapshot. The embedded board receives the data from the camera port then stores them into the Flash memory externally added onto the embedded unit. The available data storage in 16MB Flash memory is 6.6 MB, which is suitable for over 420 pictures. The picture is uploaded (a 16-kB JPEG picture is transferred in an average of 17.8 s) to a dummy FTP server, as described in Section II- A. The server on the Internet is not maintained and only used for storage space. Since, for our application, a history is not required to be kept, the client(s) accessing the picture download(s) the most recent snapshot from the FTP server. All the queries to visualize the current picture are sampling rate. The duration comparison of using a dummy server with respect to direct access is shown in Table I. The scenario of serving a single picture to a maximum of four clients is shown in Fig. 9. The x-axis represents the number of client accesses to a picture. For the camera application, the client can take a snapshot and visualize the picture on the screen (an average of 0.57 + 3.4 + 17.8 = 21.77 s for a 16-kB picture) in less than 22 s, including the time delay of command execution,camera operation, and picture


the FTP server. If the FTP server is not used, then for each client, the data will directly be served through GPRS repeatedly, causing a linear increase in the response time requirements. In this case, the connection time is proportional to the amount of data to be transferred through GPRS. With the FTPserver-supported connection, a constant cost is obtained with respect to the number of clients, and the usage cost of the system is reduced. An actual snapshot taken by the embedded system is shown in Fig. 10. In this application, the camera visually shows the location to the client. For a dispatcher application, for instance, the client can visualize the environmental or cargo conditions. The snapshot script used in this application is given in Fig. 11. The routing information of the vehicle can be collected for the analysis of speed violations, altitudes traveled for sensitive cargo, etc. The GPS module used in the application is an OEM GPS UV40, serving the NMEA 0183 [24] format raw data with a simplex communication protocol operating on one of the serial ports. The program transfers the selected GPS data to the memory after compiling a bulk of raw data. Useful information is parsed from these raw data within the embedded server by detecting the starting points of the NMEA data as reference points while reading into the serial buffer. A raw GPS data sample is shown in Fig. 12. The embedded board starts a periodic operation of acquiring the raw data from the GPS module just after the boot time of the embedded Linux. An admin script is responsible for periodically executing the GPS program and sending the qualified GPS data to the Flash memory. In order not to exceed the Flash memory limit, the newest GPS data are exchanged with the oldest data using the memory as a FIFO buffer. This provides

351


up-to-date GPS data to be available at the FIFO upon logging into the system. This can be critical in some real-time applications. Information such as the current altitude and location can be logged. The system can be set to archive the past data associated with a certain date if logging is necessary. The altitude measurements are accurate up to 20 m, which is also within the accuracy limits for a typical GPS. The sampled raw GPS data are processed, as shown in Table II. The ernestGPGGA information is parsed from the raw data and stored in a file to be further processed by the CGI scripts. The system can be set to visually track the current location of the embedded system on a map. In Fig. 13, a snapshot displays the location information for the vehicle on a map using the processed data obtained from the GPS. An icon that

Figure 7. Components of embedded system

represents a vehicle is inserted into the location using basic frames in html, utilizing the latest coordinate information from the GPS data on the map image retrieved from the FTP server. The resolution can be increased by increasing the number of frames used in html. The GPS accuracy of the measurements is less than 15 m, which is typical for a civilian GPS module. The error in accuracy is known to be due to atmospheric effects, such as the ionosphere or multipath effects. Due to the necessity of using pure html code due to web server limitation in embedded systems, the generation of the pages through CGI scripts is simple yet informative. In addition, a temperature measurement integrated circuit with a very low cost temperature measurement chip (namely, DS1620) is used to collect ambient temperature in certain time intervals at an accuracy of 0.5 C. This chip is attached to the parallel port of the embedded board. A daemon is initiated at boot time to sample and display the temperature every 30 s for a time interval of 15 min. The output graph generated for temperature is shown in Fig. 14. This enables the user on the client to constantly monitor the temperature. Fig. 15 shows a snapshot displaying the altitude information. The embedded system is accessible via a web server built into the device. This is nothing more than a CGI, which accepts data, processes them, and returns an answer. A small web server, named Boa Server, which is


particularly targeted for embedded systems, is used in the embedded system, which keeps the memory requirement minimum. Unlike most other web servers, Boa does not use multiple threads or processes to serve multiple clients at the same time but uses smart usage of the select function. This reduces kernel scheduling activities. The sample system is low cost, including the main module, the GPS, the camera, and the temperature sensor. It was necessary to design and implement some of the necessary interfaces for the RS232 communication with the main module. IV.

CONCLUSION

In this application, a low-cost, Internet-ased data acquisition and control system has been designed and implemented that should find interest from researchers. The application possibilities are virtually unlimited by attaching modules with appropriate interfaces, although the usage of the system is demonstrated with only a few sample devices. Compared with other applications, this system has advantages in terms of allowing direct bidirectional communication and reducing overhead, which can be vitally important for some real-time applications. The operational costs have been reduced by relinquishing the storage of large data to an FTP server on the Internet. The system is designed to support both static and dynamic IPs. A method to distribute the IP information has been developed. This cost-minimization effort is a big concern for mobile systems using wireless communication methods and has not been discussed before. The overall cost advantage of the system in terms of the components used makes it an attractive choice for data- acquisition applications. The power demand of the device is still in the process of being improved by putting the attached devices into sleep mode at times when they are not in use to conserve power. REFERENCES [1]

C. E. Lin, C.-W. Hsu, Y.-S. Lee, and C. C. Li, “Verification

of unmanned air vehicle flight control and surveillance using mobile communication,” J. Aerosp. Comput. Inf. Commun., vol. 1, no. 4, pp. 189–197, Apr. 2004. [2] K. Jacker and J. Mckinney, “TkDAS—A data acquisition system using RTLinux, COMEDI, and Tcl/Tk,” in Proc. Third Real-Time Linux Workshop, 2001. [Online]. Available: The Real Time Linux Foundation: http://www.realtimelinuxfoundation.org/eve nts/rtlws2001/papers.html E. Siever, A. Weber, S. Figgins, and R. Love, Linux in a Nutshell. ebastopol, CA: O’Reilly, 2005. [4] Q. Zhou, W. Wu, and Y. Ma, “The embedded data acquisition system for Mössbauer spectrum,” in Proc. Third Real-Time Linux Workshop Embedded Linux Expo Conf. Real-Time Embedded Comput. Conf.,Milan, Italy, Nov. 2001, pp. 26–29. [5] J. E.Marca, C. R. Rindt,M.Mcnally, and S. T. Doherty, “A

352

Proceedings of National Conference on Computing Concepts in Current Trends, NC4T’11 11th & 12th Aug. 2011, Chennai, India [14] G. Zhenyu and J. C. Moulder, “An Internet based telemedicine system,” in Proc. IEEE EMBS Int. Conf. Inf. Technol. Appl. Biomed., 2000, pp. 99–103.

GPS enhanced in-vehicle extensible data collection unit,” Inst. Transp. Studies, Univ. California, Irvine, CA, Uci-Its-AsWp-00-9, 2000 W. Kattanek, A. Schreiber, and M. Götze, “A flexible and cost-effective open system platform for smart wireless communication devices,” in Proc. ISCE, 2002. [7] J. E. Marca, C. R. Rindt, and M. G. Mcnally, “The tracer data collection system: Implementation and operational experience,” Inst. Transp. Studies, Univ. California, Irvine, CA, Uci-Its-As-Wp-02-2, 2002. [8] E. Bekiroglu and N. Daldal, “Remote control of an ultrasonic motor by using a GSM mobile phone,” Sens. Actuators A, Phys., vol. 120, no. 2, pp. 536–542, May 17, 2005. [9] C. E. Lin and C.-C. Li, “A real-time GPRS surveillance system using the embedded system,” J. Aerosp. Comput. Inf. Commun., vol. 1, no. 1, pp. 44–59, Jan. 2004. [10] C. Xiaorong, S. Zhan, and G. Zhenhua, “Research on remote [6]

data

acquisition

system based

on GPRS,” in Proc. 8th

ICEMI, 2007, pp. 2-20–2-23. [11] M. A. Al-Taee, O. B. Khader, and N. A. Al-Saber, “Remote monitoring of vehicle diagnostics and location using a smart box with Global Positioning System and General Packet Radio Service,” in Proc. IEEE/ACS AICCSA, May 13–16, 2007, pp. 385–388. [12] C. Zhang, J. Ge, H. Yu, and X. Zhang, “ET gravimeter data collecting system based on GPRS,” in Proc. 8th ICEMI, Jul. 18–Aug. 16, 2007, pp. 2-86–2-92. [13] A.Sang, H. Lin, and C. E. Y. Z. Goua, “Wireless Internetbased measurement architecture for air quality monitoring,” in Proc. 21st IEEE IMTC, May 18–20, 2004, vol. 3, pp. 1901–

[15] J. Dong and H. H. Zhu, “Mobile ECG detector through GPRS/Internet,” in Proc. 17th IEEE Symp. CBMS, Jun. 24–25, 2004, pp. 485–489. [16] P.Wang, J.-G.Wang, X.-B. Shi, andW. He, “The research of telemedicine system based on embedded computer,” in Proc. 27th IEEE Annu. Conf. Eng. Med. Biol., Shanghai, China, Sep. 1–4, 2005, pp. 114–117. [17] E. Lin, C.-C. Li, A.-S. Hou, and C.-C.Wu, “A real-time remote control architecture using mobile communication,” IEEE Trans. Instrum. Meas., vol. 52, no. 4, pp. 997–1003, Aug. 2003. [18] T. Motylewski, “The industrial data-acquisition system with embedded Rt-Linux and network server technology,” in Proc. Third Real-Time Linux Workshop, 2001. [Online]. [19] S.

Tan, and

and architecture

X. Chen, “Requirements design for Internet-based

control systems,” in Proc. Int. Comput. Softw. Appl. Conf., Dev. Redev., 2002, pp. 75–80. [20] C. Bettstetter, H.-J. Vögel, and J. Eberspächer, “GSM phase 2+ General Packet Radio Service GPRS: Architecture, protocols, and air interface,” IEEE Commun. Surveys Tuts., vol. 2, no. 4, pp. 2–14, Third Quarter 1999.

1906.


H. Yang, L. S.

specification

353


Efficient Calculation of Energy and Medium : A Cross-Layer Way V.Narendar1, T.Swapna2 1

Electronics and Communication Engineering, Ramappa Engineering College Warangal,India 1 [email protected] 2 Electronics and Communication Engineering,Ramappa Engineering College Waragal,India 2 [email protected]

Abstract: This paper addresses Rayleigh fading networks, and in particular, wireless ad-hoc and sensor networks over Rayleigh fading channels. First, we will model Rayleigh fading networks and show how to map the wireless fading channel to the upper layer parameters for cross-layer design. Based on the developed fading network model, we will consider two scarce resources of wireless networks, namely energy and medium, and develop a cross-layer way to improve their efficiency. In particular, we will first study the energy-efficiency and introduce a new parameter, Energy Cost Factor, as the counterpart of Transport Capacity in wireless transmission. The new parameter will be used to design energy-efficient networks. As to the medium resource, we will bring forward the Medium Resource Space, which not only organizes various medium resources in a systematic way but also considers a third dimension related to space reuse and internode interference. Finally, we will give a general discussion on the cross-layer design and show how power control and route selection jointly contribute to improving the resource efficiency. A few particular routing algorithms will also be studied in detail.

I. INTRODUCTION Wireless networks, such as ad-hoc and sensor networks, are composed of a set N of nodes sharing a common wireless medium. Such networks have been attracting intense attention for their huge potential in application. However, there are still many open problems at present, which hamper the effective design of high-quality wireless networks. In this article, we will deal with some of them, giving both theoretical results and practical design strategies. It is widely acknowledged that fading features largely in wireless transmission. How to model the wireless fading channel and map it to the upper layer parameters is of primary importance in cross-layer network design. To the best of our knowledge, however, this has not been properly studied yet in the context of wireless ad-hoc and sensor networks. Most of the previous studies either focused on the physical layer techniques to combat the adverse wireless medium, or dealt with various protocols in the upper layer, which followed the methods of wired networks Sathyabama University

without paying enough attention to the inherent differences between the underlying media. For instance, the most widely used means to relate the physical layer with the network layer is the “Disc Model”, by which the network topology is affected by the physical parameters. In this model, two nodes can successfully communicate if and only if they are within the transmission coverage area of each other, which is a disc-like area and whose radius is a constant determined by the transmit power. Obviously, this is an oversimplified model, which neither takes channel fading into consideration nor captures the boundless nature of wireless transmission [1]. So far, there has been an effective way to integrate the physical layer characteristics into the upper layer design. Recent work in [2] has shed some lights on the problem, where fading was considered in the network design. However, the assumption of determining the transmit power by route reliability requirement in [2] is somehow unrealistic, and may compromise the value of the results in practice. In this paper, we will model the Rayleigh fading networks in a more practical way by including both the physical layer characteristics and upper layer operations, and show how to combine them in the context of cross-layer design. Based on the developed fading network model, we will consider two scarce resources of wireless networks, namely energy and medium, and develop a cross-layer way to improve their efficiency. Energy is widely recognized as a scarce resource in wireless networks [3] and various energyaware routing [4] and topology control algorithms [5] were proposed. However, most of the previous works are based on the simple “Disc Model”. Owing to the lack of well-matched physical models, the energy consumption functions adopted by previous works are usually inaccurate. In this paper, we will study the energy consumption problem based on the Rayleigh fading network model, which will lead to more practical results. On the other hand, in many energy aware routing algorithms, the total energy consumption of a whole multi-hop route is used as the criterion,

354


which typically requires central control [3]. In contrast, if a node can decide its most “economical” step based on local information, some greedy algorithms can be implemented in a distributed way while maintaining a comparable performance with their centralized counterparts. In this context, energy is the resource being consumed, and a step is said to be the most “economical” if it completes the largest amount of information transport with the same energy. How to measure the information transport quantity is of paramount importance here. Obviously, the number of packets, as in wired networks, is not a complete answer. In fact, information transmission in wireless networks is greatly different from that in wired networks. For instance, the geographical location of network nodes plays an important role. While in wired networks, the high-quality of wired links actually conceals the effect of geographical distance between nodes, distance is crucial in determining the reliability of a wireless link. In their milestone work [6], Gupta and Kumar brought forward the concept of Transport Capacity, which takes the distance between the source and destination into account in measuring the traffic load. This concept unveils the fundamental difference between wireless and wired transmissions, and inspires our definition of the new parameter Energy Cost Factor for measuring energy efficiency. We will illustrate this new measure with a few practical design examples and show how to design energyefficient networks by optimizing the proposed Energy Cost Factor. Medium resource or radio bandwidth is another scarce resource in wireless networks. Compared with point-to-point wireless link, the medium resource in wireless networks is more elusive. This is largely due to two highly related phenomenons in medium access, namely space reuse and internode interference. Space reuse expands the medium resources by reuse channels at separated locations, while internode interference limits the degree of space reuse. But how to measure the medium resource extension due to space reuse and how to make use of it efficiently are still open problems. In this paper, we will bring forward the concept of Medium Resource Space, which not only organizes various resources in a systematic way but also considers the third dimension related to space reuse and internode interference. Based onthe discussion on Medium Resource Space, we will further study how to design medium-efficient networks.

different factors, which leads to different design goals. As discussed above, energy and medium are two scarce resources in wireless networks, which should be exploited with caution. In light of these, we will give a general discussion on the crosslayerdesign of wireless fading networks, and show how power control and route selection jointly contribute to improving the resource efficiency. Though, the energy-efficiency and the medium-efficiency are not compatible all the time, both of them can be improved by the same procedure, which demonstrates a cross-layer way. II. POWER AWAR ROUTING (PAR) The proposed algorithm maximizes the network lifetime & minimizes the power consumption during the source to destination route establishment. This algorithm takes special care to transfer both real time and non real traffic by providing energy efficient and less congested path between a source and destination pair. Algorithm focuses on 3 parameters: 1) Accumulated Energy of a path 1

i

1

E ij

Ei

(1)

Ei is the residual energy of an intermediate node i and Eij is the total energy of a path from node i to 2) Status of Battery Lifetime (B_S) node j 3) Type j of Data to be transfer: a. Non Real Time (NRT) b. Real time (RT). A. Parameters on each node Each node has 3 variables: Node_ID, Battery Status (B_S) and Traffic Level (T_L), Number of Weak Nodes (WNs). Battery status is further divided into 3 categories: 1) If (Battery Status < 20%) Then Set B_S = 1

It is widely acknowledges that the design of wireless fading networks should follow a cross-layer way. For instance, as will be shown in this paper, the route selection on the network layer plays an important role in determining the energy and medium consumption. Meanwhile, wireless networks are constrained by Sathyabama University

j

355

2) If (20%

Battery Status

Then Set B_S = 2 3) If (Battery Status Then Set B_S = 3.

60%)

60% )


B. Parameters to concern during Route search

Nt , no. of packets transmitted by the node after time t.

At the time of route discovery, a route request (RREQ) packet broadcasted by the source. The header of the RREQ packet includes 1) if (WN= =0)

Where is the total energy of a path from node i to node j as given in equation (1). Is number of intermediate hops along the path.

Select the path with min{T_T_L} , acknowledge the source with the selected path.

III. ENERGY CONSUMPTION MODEL

Otherwise,

Energy consumption of a node after time t is calculated using equation (3):

Select a path with less no. of WNs. else-if (T_O_L = = RT)

Ec ( t )

Nt *

Nr * Let N different values of R are received,

(3)

where R

Where Ec(t ) , energy consumed by a node after time t.


If (N = = 0)

356

2


Send negative acknowledgement informing that no such path is possible for RT. Also inform about the availability of best NRT path. If source will be interested, it may use it for forwarding its data. else-if (N = = 1) Acknowledge the source with this path. else-if (N > 1) Select the path with Min {T_T_L} and acknowledge the source with the selected path. IV. MEDIUM RESOURCE SPACE Let us begin with the point-to-point packet transmission. Generally, packets are carried out by signals that span a band of frequency and last for a period of time. The number of packets delivered by the system is linearly proportional to the product of the spectrum bandwidth and the time duration. Thus, the frequency axis and the time axis come up as the first two dimensions of the Medium Resource Space. In multiple access networks and broadcast networks, the two dimensional resource space spanned by time and frequency can be shared by multiple users, by means of FDMA, TDMA or more generally CDMA [7]. Each user occupies a unit combination of frequency and time, which will be referred to as a channel hereafter. However, the total number of packets that can be transmitted is the same as the pointto-point case. This naturally leads to the notion that the total amount of medium resource also keeps the same as before. But are these two kinds of networks the same with each other? If not, what is the difference between them? When packet transmission further extends to wireless networks with multiple transmitters and receivers, more packets can be transmitted at the same time as a result of the socalled “space reuse”. That is, the same channel can be used simultaneously by different node pairs if they are separated far away. Clearly, the Medium Resource Space expects an expansion beyond that of the point-to-point case. A primary concern here is how to describe this expansion. To address the problems mentioned above, two highly related questions are of fundamental importance. That is, in multiuser wireless networks, Sathyabama University

• What is the medium resource beyond time and spectrum? • How is the new medium resource consumed? To the first question, the space seems a natural candidate at first glance, since it holds the wireless medium and plays an important role in the “space reuse” of wireless channels. However, space itself is not a “proper” resource from the viewpoint of wireless transmission. On the one hand, space is not an information carrier. In other words, if there is no wireless node deployed in an area, information transmission will not take place in that region, no matter how large its area is. On the other hand, space can not be occupied definitely by some transmitter or signal, as a result of the boundless nature of wireless propagation. This makes space an unwieldy representation of resource. In fact, the so-called “medium access” is to access the resource at the receiver, rather than the space itself. Each receiver provides a set of wireless channels at its air interface, and a transmitter carries out packet transmission by monopolizing a channel at its intended receiver. In other words, the medium resource exists at the antenna of wireless receivers. It is the receivers scattered in the space that reuse wireless channels, and thus expand the medium resources. With this in mind, we define the set of receivers as the third dimension of the Medium Resource Space, and measure it by the number of receivers. With respect to the second problem, a node may receive signals on a given channel from either an intended transmitter or others that interfere. Once the received signal from a transmitter is strong enough, say above a given threshold v, the receiver will not be able to successfully detect other transmitters’ signal on the same channel any more. In this case, the channel at the receiver is said to be occupied, by either an intended transmitter or interferers. From the viewpoint of a transmitter, the number of receivers it occupies over a channel, denoted by NR, can be used to measure its medium resource consumption on the third dimension. Quantities similar to NR were also studied in [8] [9], but in a different context. Now, let’s reconsider the multiple access and the broadcast networks from the new view point of Medium Resource Space. Actually, they are quite different, though they do transmit the same number of packets with the same spectrum and time. In the multiple access network, there is only one receiver. Therefore, the medium resource keeps the same as the point-to-point case, and the multiple transmitters only share the channels at the single receiver. In contrast,

357


the broadcast network has multiple receivers, and thus the medium resource multiplies accordingly. However, when the transmitter sends a packet to an intended receiver, all other receivers will receive the same signal but regard it as interference. That is, when some receiver is occupied on a channel, so are other receivers on the same channel. As a result, only the same number of packets can be transmitted as the point-to-point case. In conclusion, in wireless networks, the receivers scattered in the space multiply the medium resource, while the transmitters play the role of consumers of the new resource. MAC protocols are to arrange the transmitters to utilize the resources at the receivers in an effective way. But specific MAC protocols are beyond the scope of this paper and will not be discussed in detail. V. RESULTS Total energy consumption is the difference of the total energy supplied to the network and the residual energy with the network, in Joules. The initial energy supplied to the network in each scenario is 5000 Joules. Scenario 1: Nodes: 100, Pause Time: 0 sec, No. of Sources: 10-90 As Shown in Fig.1 Total energy consumption for AODV is less than DSR form low traffic condition to high traffic and the performance of PAR is better than both AODV and DSR as it is consuming less energy as compare to other two protocols for varying number of sources. Scenario 2: Nodes: 100, Pause Time: 500 sec, Sources: 10-90 A scenario for 100 nodes and 500 pause time has been evaluated for varying no. of sources from 10-90 and the results are shown in Fig.2. As fig depicts, in the initial stage of the simulation PAR consume more energy as compare to DSR but later on it has less energy consumption as compare to AODV and DSR, while AODV and DSR do not have a clear edge over other in terms of energy consumption. The smooth curve is obtained for PAR in terms of energy consumption, which shows proper distribution of energy among nodes.

Figure 1. Total Energy Consumed Vs No. of Sources

4.2. No. of Exhausted Nodes This is the no. of nodes that die-out at the end of each simulation run, due to the consumption of all the 100 Joules of energy supplied to them at the start of the simulation. Scenario 1: Pause Time: 0 Sec; Sources: 10-45, nodes: 100 It can be observed from Fig.3 that for 0 pause time and various no. of sources, a random death of nodes has been observed of the total nodes till the end of simulation run in case of AODV, DSR and PAR. No clear edge can be defined in terms of traffic or number of sources between AODV and DSR but PAR outperforms both the protocols throughout the simulation. As it can be seen in the fig that for less number of sources (10-17) total deaths reported in DSR are more as compare to AODV and PAR but for number of sources (17-22) AODV is poor as there are more dead nodes reported as compare to DSR, PAR is still better. But for all the cases of more than 30 sources at a time DSR is better than AODV as less deaths are reported as compare to AODV. But PAR is good for a large network of 100 nodes in heavy traffic conditions (more than 30 sources).

Figure 2. Total Energy Consumed for pause time 500 seconds, 100 nodes Sathyabama University

358


Although this scheme can somewhat enhance the latency of the data transfer but it results in a significant power saving and long lasting routes. This scheme is one of its types in adhoc networks which can provide different routes for different type of data transfer and ultimately increases the network lifetime. The process of checking the proposed scheme is on for more sparse mediums and real life scenarios and also for other metrics like Path optimality, Link layer overhead, total energy consumed etc. REFERENCES

Figure 3. Exhausted nodes for 100 nodes,pause time 0 seconds [1]

“IEEE Std 802.11-1997 Information Technologytelecommunications and information exchange between systems-local and metropolitan area networks-specific requirements-part 11: wireless lan medium access control (MAC) and physical layer (PHY) specifications." [Online]. Available: http://ieeexplore.ieee.org/servlet/opac?punumber=5258, Nov. 1997.

[2]

M. Haenggi, “On routing in random Rayleigh fading networks," IEEE Trans. Wireless Commun., vol. 4, no. 4, pp. 1553-1562, July 2005.

[3]

J. Li, D. Cordes, and J. Zhang, “Power-aware routing protocols in adhoc wireless networks," IEEE Wireless Commun., vol. 12, no. 6, pp. 69-81, Dec. 2005.

[4]

Y. Xue and B. Li, “A location-aided power-aware routing protocol in mobile ad hoc networks," in Proc. IEEE GLOBECOM’01, vol. 5, pp. 2837-2841, Nov. 2001.

[5]

R. Wattenhofer, L. Li, P. Bahl, and Y.-M. Wang, “Distributed topology control for power efficient operation in multihop wireless ad hoc networks," in Proc. IEEE INFOCOM 2001, vol. 3, pp. 1388-1397, Apr. 2001.

[6]

P. Gupta and P. R. Kumar, “The capacity of wireless networks," IEEE Trans. Inform. Theory, vol. 46, no. 2, pp. 388-404, Mar. 2000.

[7]

J. G. Proakis, Digital Communications, 4th ed. Publishing House of Electronics Industry, Beijing, 2001.

[8]

D. M. Blough, M. Leoncini, G. Resta, and P. Santi, “The kneighbors approach to interference bounded and symmetric topology control in ad hoc networks," IEEE Trans. Mobile Computing, vol. 5, no. 9, pp. 1267-1282, Sept. 2006,

[9]

J. Orriss and S. K. Barton, “Probability distributions for the number of radio transceivers which can communicate with one another," IEEE Trans. Commun., vol. 51, no. 4, pp. 676-681, Apr. 2003.

Scenario 2: Pause Time: 500 Sec; nodes: 100, Sources: 10-45 For a large value of pause time and for various traffic loads, Fig.4 shows, that in case of light traffic ( for 10-30 sources) more no. of nodes are exhausted in DSR as compare to AODV and PAR but for heavy traffic ( more than 30 sources) more deaths are reported in case of AODV as compare to DSR and PAR. But PAR is hardly poor for a few values of sources as compare to other strategies.

Figure 4. Exhausted nodes for 100 nodes,pause time 0 seconds

VI. CONCLUSION Simulation result shows that the proposed scheme PAR is outperforms in terms of different energy related parameters over AODV and DSR even in high mobility scenarios. At the time route selection PAR take care of crucial things like traffic level on the path, battery status of the path, and type of request from user side. With these factors in consideration PAR always select less congested and more stable route for data delivery.


359


Search Engines-An Insight V.Vinu Chakravarthi* & Dr.G.Veeramani** * Ph.D., Research Scholar-Manonmaniyam Sundaranar University ** Professor and Head, Dept. of Management Studies, Dhanalakshmi College of Engineering, Tambaram, Chennai – 601 301

2 SEARCH ENGINE HISTORY:

Abstract:: Search engines are the key to finding information on the vast expanse of the World Wide Web. With out sophisticated search engines, it would be virtually impossible to locate any thing on the web without knowing a specific URL. Users request information from search engines and in return they receive a list of possible URLs that match the request. Hence it is important to understand what is search engine, how search engines work? And the various types of search engines and developing an effective strategy to list or promote an URL in search engine.

The very first tool used for searching on the Internet was Archie. The name stands for "archive" without the "v". It was created in 1990 by Alan Emtage, Bill Heelan and J. Peter Deutsch, computer science students at McGill University in Montreal. The program downloaded the directory listings of all the files located on public anonymous FTP (File Transfer Protocol) sites, creating a searchable database of file names; however, Archie did not index the contents of these sites since the amount of data was so limited it could be readily searched manually.

Keywords: Search engines, Search engines in India, Search engine suggestions

The rise of Gopher (created in 1991 by Mark McCahill at the University of Minnesota) led to two new search programs, Veronica and Jughead. Like Archie, they searched the file names and titles stored in Gopher index systems. Veronica (Very Easy Rodent-Oriented Net-wide Index to Computerized Archives) provided a keyword search of most Gopher menu titles in the entire Gopher listings. Jughead (Jonzy's Universal Gopher Hierarchy Excavation and Display) was a tool for obtaining menu information from specific Gopher servers. While the name of the search engine "Archie" was not a reference to the Archie comic book series, "Veronica" and "Jughead" are characters in the series, thus referencing their predecessor.

1 WHAT ARE SEARCH ENGINES? A search engines is a program designed to help find information stored on a computer system such as the World Wide Web, or a personal computer. The search engine allows one to to ask for content meeting specific criteria (typically those containing a given word or a phrase) and retrieving a list of references that match those criteria. Search engines use regularly updated indexes to operate quickly and efficiently. A program that searches documents for specified keywords and returns a list of the documents where the keywords were found. Although search engines is really a general class of programs, the term is often used to specifically describe systems like Alta Vista and Excite that enable users to search for documents on the world wide web and USENET newsgroups. Typically, a search engine works by sending out a spider to fetch as many documents as possible. Another program, called an indexer, then reads these documents and creates an index based on the worlds contained in each document. Each search engine uses a proprietary algorithm to create its indices such that ideally only meaningful results are returned for each query.


One of the first "full text" crawler-based search engines was WebCrawler, which came out in 1994. Unlike its predecessors, it let users search for any word in any webpage, which has become the standard for all major search engines since. It was also the first one to be widely known by the public. Also in 1994, Lycos (which started at Carnegie Mellon University) was launched and became a major commercial endeavor. Soon after, many search engines appeared and vied for popularity. These included Magellan (search engine), Excite, Infoseek, Inktomi, Northern Light, and AltaVista. Yahoo! was among the most popular ways for people to find web pages of interest, but its search function operated on its web

360


collected and stored and subsequently searched. This explains why sometimes a search on a commercial search engine such as Yahoo or Google will return results that are in fact dead links. Since the search results are based on the index, if the index , if the index hasn’t been updated since a web page as still an active link even though it no longer s. Google ranks based on popularity, and yahoo is a directory or editorial site.

directory, rather than full-text copies of web pages. Information seekers could also browse the directory instead of doing a keyword-based search. In 1996, Netscape was looking to give a single search engine an exclusive deal to be the featured search engine on Netscape's web browser. There was so much interest that instead a deal was struck with Netscape by five of the major search engines, where for $5Million per year each search engine would be in a rotation on the Netscape search engine page. The five engines were Yahoo!, Magellan, Lycos, Info seek, and Excite.

Para sites collect and represent the search engine content eliminating the advertisements. Meta crawler is an example of this type. Askjeeves is a question based search and requires a personal email for submission to best place the sites to respond to a query.

Around 2000, Google's search engine rose to prominence. The Company achieved better results for many searches with an innovation called PageRank. This iterative algorithm ranks web pages based on the number and PageRank of other web sites and pages that link there, on the premise that good or desirable pages are linked to more than others. Google also maintained a minimalist interface to its search engine. In contrast, many of its competitors embedded a search engine in a web portal.

So why will the same search on different search engines produce different results? Part of the answer to that question is because not all indices are going to be exactly the same. It depends on what the spiders find or what the humans submitted. But more important not every search engines uses the same algorithm to search through the indices. The algorithm is what the search engines use top determine the relevance of the information in the index to what the user is searching for.

3 TYPES OF SEARCH ENGINES: The basic types of search engines: Those that are powered by robots (called crawler ants or spiders) and thoses that are powered by human submissions; and thoses that are a hybrid of the two.

One of the elements that a search engine algorithm scans for is the frequency and location of the keywords on a web page. Those with higher frequency are typically considered more relevant. But search engine technology is becoming sophisticated in its attempt to discourage what is known keyword stuffing or spamdexing.

3.1 Crawler-based search engines are those that use automated software agents called crawlers that visit a website, read the information on the actual site,read the site’s meta tags and also follow the links that the site connects to performing indexing on all linked web sites as well. The crawler returns all that information back to a central depository, where the data is indexed. The crawler will periodically return to the sites to check for any information that has changed. The frequency with which this happens is determined by the admininistrators of the search engine.

Another common element that algorithms analyze is the way that pages link to other pages in web. By analyzing how pages link to each other an engine can both determine what a page is about and Whether that page is considered important and deserving of a boost in ranking Just as the technology is becoming increasingly sophisticated to ignore keyword stuffing, it is also becoming more savvy to web masters who build artificial links in to their sites in order to build an artificial ranking.

3.2 Human powered search engines rely on humans to submit information that is subsequently indexed and catelogued. Only information that is submitted is put into the index.

4 PARTS OF SEARCH ENGINE: Search engines have three major elements. First is the spider also called the crawler. The spider visits a web page reads it and then follows links to other pages within the site. This is what it means when someone refers to a site being spidered or crawled. The spider returns to the site on a regular basis such as every month or two, to look for

In both cases, when you query a search engine to locate information, you’re actually searching through the index that the search engine has created you are not actually searching the web. These indices are giant database of information that is


361


7.2 Outsourced optimization

changes. Every thing the spider finds goes in to the second part of a search engine called the index. The index some times called the catalog is like a giant book containing a copy of every web page that the spider finds. If a web page changes then this book is updated with new information. Some times it can take a while with the new pages or changes that the spider finds to be added to the index. Thus a webpage may have been spidered but not yet indexed. Until it is indexed it is not available to those searching with the search engine. Search engine software is the third part of search engine. This is the program that sifts through the millions of pages recorded in the index to find matches to a search and rank them in order of what it believes is most relevant.

Companies will offer clients a range of services that will optimize their website with in search engines. Along with side optimization services these companies can offer a range of technical solutions to maximize the presence of a site with in the major search engines. 7.3 In house optimization: A lot of organization chooses to adopt their own search engine optimization by attempting to make their own websites search engine friendly. This route is limited but can be achieved by using following variables. Search engines work on the basis of themes when inserting metatags description and image names they must contain words that are relevant to the site. It the website address is www.recruitment.com then the tags, text, descriptions and image names must contain words such as recruitment and other words associated with it so that the search engine is to believe that the site is highly relevant to the subject.

5. WEBSITE LISTING WITH SEARCH ENGINE: Depending on the search engine there are two common ways they can discover your site, you submit a request for your web site to be included or reviewed in their index be referenced by another website that is already listed in a search engines index. Search engines discover your site when they crawl or spider the web. This means that they automatically ready web page text and then to follow most of the normal kinds of HTML link. They rely on web site text to record the nature and content of the site.

SUMMARY As the internet users in the India is increasing day by day it is essentially important to the companies to understand the importance of search engine, as today search engines are not only a directory but it becoming essentially important tool even in increasing brand image of any product or a company.

6. INCLUSION IN SEARCH ENGINE INDEX: Although there are many different types of search engines, there are ways to refine your website so that is picked up or ranked well by search engines in general. These key factors include:

REFERENCE: Choosing your keywords carefully 1.

http://en.wikipedia.org/wiki/Search_engine_optimizat ion 2. Brian Pinkerton. Finding What People Want: Experiences with the WebCrawler 3. http://www.webir.org/resources/phd/pinkerton_2000.p df 4. DannySullivan(June 14, 2004). Who Invented the Term "Search Engine Optimization"? 5. http://forums.searchenginewatch.com/showthread.php ?s=16bfbd0dd4626be835a5e803617a2a03&p=2119#p ost2119 6. Cory Doctorow (August 26, 2001).Metacrap:Putting the torch to seven straw-men of the meta-utopia 7. (http://www.elearningguru.com/articles/metacrap.htm). 8. Bing Liu (2007), Web Data Mining: Exploring Hyperlinks, Contents and Usage Data. Springer, ISBN 3540378812 9. Xie, M.; et al. (1998). "Quality dimensions of Internet search engines". Journal of Information Science 24 (5): 365–372. 10. Stefanie Olsen (May 30, 2003). "Judge dismisses suit against Google". CNET 11. Melissa Burdon (March 13, 2007). "The Battle Between Search Engine Optimization and Conversion: Who Wins?"

Thoroughly applying the keywords in the text of the site Testing the site’s rankings and updating the site often Never fools the search engine. Nothing replaces quality content Submitting your site to the proper engines. 7 FEW TIPS TO IMPROVE SITE: There are three clear distinctions in making site a search engine friendly 7.1 Site side optimization: Optimize your site by improving your actual site and amongst other techniques, insert strategically chosen words and text in to your code to provide search engine spiders with relevant content. These keywords and text are referred to as Meta tags and meta descriptions.


362


Improvement of Stability of a Generator Excitation Control System using Stabilizer with Time Delays G. Naveen kumar1, V. Usha Reddy2 1

PG Student, Dept. of E.E.E, S.V. U. College of Engineering, SV University, Tirupati, 2Assistant Professor of E.E.E, S. V. U. College of Engineering, SV University, Tirupati. 1 [email protected] [email protected]

Abstract: The evaluation of time delay is very important for power system stability of a generator excitation control system. Due to the use of measurement devices, communication links for data transfer, there exists significant time delays. This paper presents the effect of time delays on the stability of generator excitation control system. A delay margin is determined theoretically that the power system can sustain without losing its stability. For certain values of time delays, the excitation control system loses the stability. A stabilizer is used to damp out the power system oscillations for the time delays. A simulink model using stabilizer with time delays used for the better stable operation of the generator excitation control system is presented.

delay can typically in the range of 0.5 – 1.0 s. These time delays in power system control have a destabilizing impact on the system dynamics which leads to loss of synchronism and instability. Therefore there is a need to know the maximum amount of time delay known as delay margin of AVR system that the system can tolerate without losing its stability [6]. Conventional power system stabilizers are major local damping controllers acting through generator excitation systems. Here a stabilizer is used to damp out the power system oscillations for the time delays [1,2].

Keywords— Generator excitation control - Time delay Delay Margin - Stability - Stabilizer - Matlab/Simulink

II. MODEL OF EXCITATION CONTROL SYSTEM WITH TIME DELAYS

I. INTRODUCTION

A linear or linearized models are commonly used to analyze the system dynamics and to design a controller [3]. Figure 1 shows the block diagram of excitation control system with delays. Each component of the system is modeled by a first-order transfer function and their terms are having usual meanings. The transfer function of the PI controller is described as

In an interconnected power system, the control strategy is to generate and deliver power as economically and reliably as possible while maintaining the voltage and frequency within the permissible limits [1,2]. The excitation control system known as automatic voltage regulator (AVR) regulates the reactive power and voltage magnitude of a synchronous generator at a specified level [3]. The controllers are used to set for a particular operating condition and take care of small changes in load demand to maintain the frequency and voltage magnitude within the specified limits. This paper investigates the effect of the time delay on the stability of the generator excitation control system. Due to the use of measurement devices and communication networks for data transfer, there exist significant time delays. The measurement and communication delays involved between the instant of measurement and that of signal being available to the controller are the major problem in the power system control [4,5]. This


(1) where KP and KI are the proportional and integral gains. The measurement and communication delay

1

and processing delay 2 are the respective time delays which are expressed in exponential terms. The effect of PI controller will give the response of the generator excitation system for the desired performance [7]. The characteristic equation of the excitation control system can be obtained from

363

1+GC(s)GA(s)GE(s)GG(s)GR(s)

as

(2) (3)


Fig. 1 Block diagram of excitation control system with time delays

where = 1 + 2. The polynomials P(s) and Q(s) are having the coefficients in terms of time constants of the system are given in Appendix A. P(s) = s(1+TAs)(1+TEs)(1+TGs)(1+TRs) =T5s5 + T4s4 + T3s3 + T2s2 + s Q(s) = (KPs + KI) KAKEKGKR

(4)

The problem of stability can be stated as follows [11]: Given : A linear time-delay system of (3) Determine : if it is delay-independent stable or not; If not (the time-delay system is delaydependent stable); Find

In the following section, a formula is to compute delay margin for a stable operation of the excitation control system is presented. III. STABILITY CRITERIA

:

the delay-margin system stable.

that keeps the

In the following section, an analytical formula is derived to compute the delay margin for the delay-dependent case is presented.

A. Problem Identification

B. Solution Method

The stability criteria for two classes of time-delay systems are given in the following [9,10] :

For the system to be asymptotically stable, all the roots of (3) lie in the left half of the complex plane. If for some

(i) Delay-independent stability: Asymptotic stability holds for all positive values of the delay. (ii) Delay-dependent stability: Asymptotic stability holds for some values of the delay (

) and

the system is violated for other values of delay

,

has a root on the imaginary axis at s=j , so does for the same value of Hence, looking for roots on the imaginary axis reduces to finding values of for which common root. That is,

and

have a

then the system becomes unstable. Note that the delay-free system ( = 0) is assumed to be stable.


364

P(s) + Q(s)

=0

P(-s) + Q(-s)

=0

(5)


By eliminating exponential terms in (5), we get the following polynomial: P(s)P(-s) – Q(s)Q(-s) = 0 If we replace s by

+

; r = 0, 1, 2,. . . .

(6) -1

in (6), we have the following polynomial

2

in W( ):

(10)

It must be stated here that inverse tangent operation always gives angles in the right half-plane. For this reason, the angle

W(

= P(j

(7)

Substituting P(s) and Q(s) polynomials given in (4) into (7), we

s in the range [ /2 /2]. However, must be in the left halfplane when cos( k ) < 0. Therefore, must be added to or subtracted from the angle

when cos(

obtain margin results W(

=

+

+

+

+

+ (8)

by

, then

, so that for real value of , the value of must be positive. If there is no positive root of (8) with respect to 2, that is if all r k < 0, the system is stable for all 0, which indicates that the system is delay-independent stable. If there exist the positive roots of the polynomial (8) with respect to 2, then the system is delay-dependent stable. The value of be obtained as follows: ,

) + Q(j

)

) 0 and stream Q is not available in the Activelist of class J. Step 4 If both the conditions are satisfied, then append stream Q to the active list of class J; else end the process. 3) Algorithm for schedules class J traffic when one buffer is used for all streams in class J: Step 1 Schedule the class J traffic. Step 2 Check whether X Bits < the expected frame length and the stream Q is backlogged. Step 3 If both the conditions are satisfied then send a packet from that stream Q . Step 4 Update the grant parameters and X Bits. Step 5 Repeat the process until both the conditions fail. Let the expected frame length be 4400 bits at the start of the frame. Table 1 shows a two-class system buffer outline in which Pi,j denotes j from stream i with K bits and Gi is the available grant of stream i.

o In the DequeueInit process, the grant of each stream inside A class is incremented by some quantum computed based on the frame beginning time. If a nonbacklogged stream becomes backlogged, its reference is appended to ActiveList. o

In the DequeueProcess, packet scheduling is performed.

1) OCGRR scheduling algorithm within one frame: Step 1 Get the expected frame length that can be representing in . Step 2 The index of class is represented as J. The J range from 1 to (total no. of traffic classes) . Each class contains a number of streams and it is represented as Q. Step 3 Update the grant value for that stream. Step 4 Check whether the stream is backlogged or not. Step 5 If the stream is backlogged then append that stream to the active list of that class J and go to step 7. Step 6 If the stream is not backlogged then go to the next step. Step 7 Get the next stream, Q. This process is called dequeueinit process and repeat the process from steps 2 to 6 until there is no more stream in that class J. Step 8 Schedule the class J traffic. Step 9 check whether the total transmitted bits in a frame (X bit) is greater than 1. If so, end the process else get the next class and repeat the


Table 1 Example Status of the Stream

Table 2 Packet Transmission Order

Table 2 shows the grant values, the output sequences and the total transmitted bits at the end of 418


REFERENCES

each EF round. Stream-3 is removed from the ActiveList EF at the first round when G3 becomes negative. Streams 2,1, and 7 are removed from this list at the second, third, and fourth rounds, respectively. The transmission of the EF packets is finished at the fourth round with the total transmission of 3,000 bits. The transmitted EF sequence: P11;P21;P31;P71;P12;P22;P72;P13;P73;P74. This sequence shows that packets from different streams have fair access to the bandwidth via a smooth schedule. In comparison, the DRR output sequence with the same grants: P11; P12; P21; P71; P72; P73, not a smooth packet transmission. After finishing the EF packets, four packets from the BE buffer are sent until the BE grant becomes negative, and the frame ends before exceeding the frame length than .

[1] Miaoyan Li and Bo Song,” Design and Implementation of a New Queue Scheduling Scheme in DiffServ Networks”, Proc. IEEE ,pp.117-122,Oct. 2010. [2] Dong Lin and Mourir Hanmdi,”Two-Stage Fair Queuing using Budget Round-Robin”, Proc. IEEE Communication Society, ICC, Oct. 2010. [3] Jingnan Yao, Jlani Guo,” Ordered Round-Robin: An Efficient Sequ ence Preserving Packet Scheduler”, IEEE Trans. On Computers, vol. 57, no. 12,Dec. 2008. [4] Srikanth Jagabathula, Vishal Doshi and Devavrat Shah,” Fair scheduling through packet election”, Proc. IEEE Communication Society, INFOCOM, pp. 915-923, Nov. 2008. [5] Nan Jin and Scott Jordon, ”On the Feasibility of Dynamic Congestion –Based Pricing in Differentiated Services Networks”, IEEE/ACM Trans. Networking, vol. 16, no. 5, Oct. 2008. [6] John Musacchio and Shuang Wu, “The price of Anarchy in Competing Differentiated Services Networks”, Proc. IEEE Technology and Information Management Program, WeD3.1,pp. 615 -622, Sep. 2008. [7] Lijie Sheng, Haoyu Waoyu, BaoBao Wang,” Throughput Fairness Round Robin Scheduler for Non-continuous Flows”, IEEE Fourth International Conference on Networking and Services, pp. 128-133, Sep. 2008. [8] A.G.P. Rahbar and O. Yang, “The Output-Controlled Round Robin Scheduling in Differentiated Services Edge Switches”, Proc. IEEE BROADNETS ’05, Oct. 2005. [9] C. Guo, “SRR: An O(1) Time-Complexity Packet Scheduler for Flows in Multiservice Packet Networks”, IEEE/ACM Trans. Networking, vol. 12, no. 6, Dec. 2004. [10] R.R.F Liaos and A.T . Campbell, “Dynamic core provisioning for Quantitative Differentiated services”, IEEE/ACM Trans. Networking, vol. 12,no.3 pp.429-442, June 2004. [11] H.M. Chaskar and U. Madhow, ”Fair Scheduling with tunable Latency: A Round Robin apporach” , IEEE/ACM Trans. Networking, vol. 11, no.4 pp. 592-601, Aug 2003. [12] L. Ji, T.N. Arvanitis, and S.I. Woolley, “Fair Weighted Round Robin Scheduling Scheme for Diffserv Network”, IEE Electronic Letters, vol. 39, no. 3, Feb. 2003. [13] C. Zhang and M. MacGregor, “Scheduling Latency-Critical Traffic: A Measurement Study of DRR+ and DRR++”, Proc. IEEE High Performance Switching and Routing (HPSR), June 2002. [14] Y. Ito, S. Tasaka, and Y. Ishibashi, “Variably Weighted Round Robin Queueing for Core IP Routers”, Proc. IEEE Int’l Performance, Computing, and Comm. Conf. (IPCCC ’02), Apr. 2002. [15] S. Kanhere, A. Parekh, and H. Sethu, “Fair and Efficient Packet Scheduling Using Elastic Round Robin”, IEEE Trans. Parallel and Distributed Systems, vol. 13, no. 3, pp. 324-336, Mar. 2002.

4. CONCLUSION OCGRR has the capabilities of using smaller frame lengths and rounds; sending traffic packet by packet in smaller rounds; reducing the inter transmission time from the same stream; reducing queuing delay, jitter, and startup latency; controlling the starvation of lower priority classes; and beginning the transmission in each class from a delayed stream in the previous logical frame to ensure low latency and fairness. A desired QoS performance can be obtained by adjusting class indices. 5. FUTURE WORK There are number of avenues for future work; our future research is to automate the blocking and backlogging options according to the traffic flow and to study end-to-end QoS in a DiffServ domain by using OCGRR at the routers. This is a challenging and interesting issue as it would require the schedulers in routers along the path to cooperate with each other to provide a desired QoS.


419


Virtualized Network Infrastructure for Secure Cloud Computing in VLANs P.Sivaraman#1, G.Barath#2, C. Priyadharsini #3 , G.Appasami#4 PG Student, Department Of Computer Science and Engineering#1,2,3 Assistant Professor, Department Of Computer Science and Engineering#4 Dr.Pauls Engineering College, Villupuram. [email protected], [email protected], [email protected], [email protected]

Abstract The major hurdles in cloud computing is security. Due to the rapid development in the particular field. According to cost Most cloud services (e.g. Amazon EC2) are provided at low cost and no protection to users. At the other end of the spectrum, highly secured cloud services (e.g. Google “government cloud”) are offered at much higher cost by using isolated hardware, facility, and administrators with security clearance. This paper, deals with the “middle ground”, where users can still share physical hardware resource, but user networks are isolated and accesses are controlled in the way similar to that in enter-prise networks. We believe this covers the need for most enterprise and individual users. We propose an architecture that takes advantage of network virtualization and centralized controller. This architecture overcomes scalability limitations of prior solutions based on VLANs, and enables users to customize security policy settings the same way they control their onsite network.

I. INTRODUCTION Despite the rapid development in the field of cloud computing, security is still one of the major obstacles to cloud computing adoption [5, 4, 3]. To ease the concerns of IT managers, it is critical to ensure data privacy and integrity in the cloud at a level that is at least comparable to that in current enterprise networks. However, the current cloud computing services fall in short on isolating computing resources and networks between customers. This is not surprising because the success of cloud computing depends on economy of large scales. It is essential for cloud service providers to take advantage of resource sharing and multiplexing among customers. Virtual machines of different customers may reside on the same physical machine, and their data packets may share the same LAN. Such lack of isolation brings security risks to users. For example, [15] has shown that it is possible for a hacker to conduct attacks towards another Amazon EC2 user who shares hardware resources with the hacker in the cloud. On the other end of the spectrum, Google has pro-posed “government cloud”, which creates entirely separate hardware, software, and administrators (with appropriate background checks) for special customers. While such cloud service can be very secure, it is also very ex-pensive — almost like building a separate data centre for each customer.


In this paper, we explore the “middle ground”, where users can still share physical hardware resource, but user networks are isolated and accesses are controlled in the way similar to that in enterprise networks. We believe this covers the need for most enterprise and in-dividual users. More specifically, we propose a new data centre architecture with following • Isolation. The architecture provides effective isolation between different customer networks. This includes supporting their private IP address spaces, which may potentially be overlapping, and isolating their trac. Resource allocation should be managed so that customers cannot impact each other’s resource usage in an uncontrolled manner. • Transparency. The underlying data centre infrastructure and hardware should be transparent to the customers. Each customer should have a logical view of its own network, independent of the actual implementation. This simplifies the administration for the customer and improves security. • Location independence. The virtual machines (VM) and networks of customers should be “location in-dependent”, i.e., can be physically allocated any-where in the data centre. This can greatly improve resource utilization and simplify provisioning. • Easy policy control. Each customer may have its own policy and security requirements. The architecture should allow customers to configure their individual policy settings on the fly, and enforce such settings in the network. • Scalability. The number of customers that can be supported should be restricted only by the resources available in the data centre, not by design artifacts. • Low cost. The solution must mostly rely on o - the-shelf devices, so that new investment for cloud service providers can be reduced. In this paper, we exploit recent advances in technologies amenable to network virtualization [1, 7, 8]. Network virtualization techniques can logically separate different networks on the same hardware and partition resources accordingly [9, 1]. This feature is

420


useful for providing good isolation as well as networkresource sharing among different users. Furthermore, recently proposed mechanisms simplify packetforwarding elements and make control functions more flexible and manageable [7, 8] by using centralized control. Given the high density of physical resources and demand for high manageability of devices, the centralized control architecture suits data centre networks very well. However, unlike in typical network virtualization solutions, our design does not require deploying specialized routers or switches across the entire data centre network. Instead, conventional o -the-shelf Ethernet switches can be used in most parts of the network. Enhanced layer 2 switches, which we refer as Forwarding Elements (FE), are deployed only at certain aggregation points to provide the required virtualization functions. In this architecture, each customer has its own isolated virtual network in the data centre, to which access is tightly controlled. But physically, such virtual network may be distributed at anywhere in the data centre. This architecture intends to over a more secured elastic cloud computing (SEC2) service. But the design can also naturally support virtual private cloud (VPC) service, where each user’s private network in cloud is connected to its on-site network via VPN. We describe the architecture design in the next section. We then present the design details in Section 3, followed by further discussion of several design issues in Section 4. In Section 5 we explain why existing data centre designs are not sufficient for solving this problem. Concluding remarks are in Section 6.

limitation is that it is not easy to enable per-user security policy control. There are potentially two places where security policies can be enforced: at the middle boxes at aggregation layer, or in software setup at each hypervisor. Just relying on hypervisor setup may carry security risks given that hypervisor runs on the same physical host as the user VMs, and thus share same resources. The middle box solution is likely to be more secure since it is more difficult to be hacked by users. But to enforce packets always traverse through given middle boxes in a given sequence requires nontrivial planning, configuration, and even tweaking physical link connections, rendering the solution not ad-ministrative feasible. The latest Cisco fibre extender solution addresses this problem by forcing all trac to go through the aggregation switch. But this creates unnecessary trac load in aggregation switches since in many data centres majority internal trac is between nodes within the same rack. II.2 Secure Elastic Cloud Computing (SEC2) Based on the above observations, we propose a new design based on network virtualization. In this de-sign, network virtualization is accomplished by two entities: Forwarding Elements (FEs) and Central Controller (CC). FEs are basically Ethernet switches with enhanced APIs that allow them to be controlled from a remote CC. Packet handling actions such as address mapping, policy checking and enforcement, and forwarding are done in FEs. CC stores control information such as addresses, location, and policy. Such information is distributed to different FEs as needed.

II. ARCHITECTURE II.1 Conventional data centre architecture A conventional data centre network is typically partitioned into three layers: access, aggregation, and core. It is possible to support multiple isolated networks in this architecture by using VLANs combined with “virtual switches” in hypervisors such as VmWare and Xen. VMs on physical hosts can be configured to use different VLANs. Both virtual switches in hypervisors and physical layer 2 and 3 devices in data centre can be con-figured to support VLANs for each customer. To extend VLANs across layer 3 networks, “VLAN trunking” can be used to tunnel packets across routers. The main limitation of this solution is caused by the scalability issues of VLAN. The maximum number of VLANs is limited to 4K due to VLAN id size. Furthermore, overlaying thousands of VLANs on the same physical network may cause network management com-plications and increase control overhead. For example, One physical link failure may trigger spanning tree computation on all VLANs that run on it. In a more general sense, VLAN couples both access control and packet forwarding. We believe it is cleaner and more scalable design to separate the two. Another Sathyabama University

Figure 1 illustrates the basic SEC2 architecture. Unlike in the conventional three layer design, here the net-work is structured in two levels: one core domain and multiple edge domains surrounding it. The core domain consists of layer 2 Ethernet switches with high switching capacity. Its function is to transport packets between edge domains. An edge domain contains physical hosts connected by Ethernet switches. Each edge domain is connected to the core domain through one or more FEs. The edge

421


domain has three functions. First, it resolves packet addresses, and determines the edge domain and MAC address of an IP packet. Second, it ensures that packets of different customers are isolated, and are processed ac-cording to their configured policy settings. Packets can-not traverse across customer network boundary with-out being checked and enforced based on security rules. Third, depending on the destination, packets are either delivered locally, or tunnelled through the core domain to reach the destination edge domain. Note that FEs serve as gateways between core and edge domains, while CC provides control and configuration functions. Middle boxes such as firewalls and load balancers are attached to FEs, so that packets needing such treatment will be forwarded through them. In this design, each customer has its own isolated virtual network in the data centre, but such virtual network may be physically distributed across any edge domains. Each customer can set up security policies for their net-work through a web portal, which are in turn translated into policy settings in CC. II.3 Numbering and addressing To distinguish between different customers, we assign each customer a unique cnet id (customer network id). If a customer has multiple subnets or needs to set up several different security zones, then each subnet or se-curity zone is assigned a unique cnet id. A cnet is a logical entity where all nodes in it share the same pol-icy rules, independent of the physical domain structure described previously. Logically, an end point (i.e. VM) can be identified by the combination of (cnet id, IP). This is in turn mapped to a unique layer 2 MAC address 1. Most existing host virtualization platforms support assigning virtual MAC addresses to VMs. In platforms that do not support such configuration, VMs usually share the same MAC address as their host machine. In this case we generate a pseudo MAC address for the VM, and use such pseudo address for identification purpose but not for packet forwarding. Each edge domain is assigned a unique eid. We use VLANs to separate different customers within each edge domain. In the same edge domain, there is one to one mapping between VLAN id and cnet id. VLAN configuration is done at all networking devices in the edge domain, including FEs, Ethernet switches, and virtual switches on host hypervisors. Such configuration is transparent to VMs, so that applications that run on the VMs are not aware of VLAN configurations. The scope of a VLAN is limited within the same edge domain. Different edge domains can reuse VLAN ids. As a result, different customers in different edge do-mains may have the same VLAN id. Hence we eliminate the limit on the number of customers that can be accommodated due to VLAN id size. In addition, each subnet may be expanded to multiple edge domains using different VLAN ids, so


the scale of each customer subnet is not limited by edge domain size. This design implicitly poses a limit on the number of VMs that can be supported in the same edge domain. In the worst case, where each VM belongs to a different customer in the same edge domain, there can be a maximum of 4K VMs in each edge domain. However in general an edge domain can accommodate many more VMs since many customers are likely to have more than one VM. Note that this limit is only for the same edge domain but the proposed architecture does not impose limits on the number of edge domains in a data centre. II.4 Integration with customer’s on-site network This architecture can be naturally extended to accommodate the service where customers need to integrate cloud resources with their existing on-site net-work. One example of such service is Amazon Virtual Private Cloud (VPC) service, where customers can ex-tend their network to the data centre via IPSec tunnels. The customer’s on-premise network can be treated as a special edge domain for the data centre network. The customer site can be connected to the data centre network using existing VPN technologies such as VPLS, layer 3 MPLS VPN, or IP tunnels such as GRE or IPSec. In particular, if a customer has already been using the VPN services from a service provider, it will be easy to add data centre as an additional site of the same VPN instance. The FEs at the data centre edge can serve as Customer Edge (CE) routers since the data centre is a customer of the service provider’s VPN service. Hence such FEs are referred as CEacting FEs for convenience. III. DESIGN DETAILS In this section, we present further details on each component of our design. III.1 Central Controller (CC) CC controls the operation of FEs. It maintains both address mapping and policy databases. The following mappings are maintained by CC: • VM MAC (cnet id, IP). This resolves the IP address of each VM to its corresponding MAC address. • VM MAC edge domain id eid. This identifies the edge domain where the VM is located at the present time. • eid FE MAC list. FE MAC refers to the MAC addresses of the FEs to which the edge domain connects. Note that it is possible for an edge do-main to have multiple FEs for load sharing and reliability reasons. • (cnet id, eid) VLAN id. This identifies the VLAN used by each customer in each edge domain.

422


CC also maintains policy rules. An example of policy rule can be: packets from customer A are first forwarded to firewall F, and then to its destination. In such case, the first hop FE that enforces this policy needs to tunnel such packets to firewall F. Although CC is conceptually one entity in this architecture, it can be implemented in a distributed way. For example, different customers can be assigned to different CCs by using Distributed Hash Table (DHT). Since the management and policy control of different customer networks are relatively independent, such partition does not a ect the functionality of CC. III.2 Forwarding Elements (FE) FEs are gateways between the core domain and the edge domains. Each edge domain may have more than one FEs for redundancy purpose. Functions of FE include the following: • Address lookup and mapping. When a packet is originated from its edge domain to other domains, it looks up the FE MAC of the destination domain and VLAN ID in destination domain. This is done by first checking its local cache, and if there is no hit, it inquires the CC.

The ARP request reaches FE1 at e1’s edge. FE1 looks up its ARP table. If no entries are cached, it inquires CC. By using the mapping tables defined in Section 3.1, CC looks up B’s MAC M AC B , its edge domain e2, and corresponding VLAN id y and FE M ACF E2 in e2. Based on CC’s response, FE1 then installs entry “to M ACB : tunnel to M ACF E2, dest VLAN y” along with other policies in its forwarding table, and returns A with the ARP reply that contains M ACB . Packets sent in the same VLAN are directly delivered to the destination VM without policy checking 2. Otherwise they first reach FEs, which in turn enforce security policies and make forwarding decisions. This holds even for packets between different VLANs in the same edge domain. After A receives M ACB , it can send data packet to B with M ACB as its layer 2 destination. The packet is forwarded to FE1. FE1 enforces appropriate policies and performs QoS treatments, and then change the VLAN id to y and add an outer MAC header M ACF E2 . The packet is then forwarded across the core domain, reaching FE2. FE2 strips o outer MAC header, and delivers the packet to B.

• Policy enforcement. FEs enforce policy by applying filtering, QoS treatment, or redirecting to middle boxes that are attached to them. By de-fault, The customer can o er public access to node B by packets designated to a different customer are dropped. requesting a public IP address, which is in turn NATed to B’s private address. In this case, external • Tunnelling. FE tunnels the packet across the packet will be forced by FE to traverse through core domain to the destination FE via MAC-in-MAC. firewall and NAT middle boxes before reaching the The source FE adds another MAC header to the private network. original Ethernet frame. The destination FE strips o the outer layer header upon receiving the packet. Most modern Ethernet switches allow larger frame sizes III.4 Connectivity across sites (jumbo frames) to be used so the extra few bytes of The customer network can be connected to the data MAC header is not a problem. This is especially true for the core domain since we expect high-end Ethernet centre via layer 2 or 3 VPN tunnels, depending on switches to be deployed to meet capacity needs in the whether the customer requires the cloud resource to be part of the same LAN of its on-site network or not. core. Layer 2 VPLS tunnels can be used to attach the data III.3 Data forwarding path centre VMs into customer’s existing LAN. In the VPLS setup, at the customer end, customer edge (CE) router or switch is connected to a VPLS provider edge (PE) router of the service provider. Similarly at the data centre end, CE-acting FE is connected to another PE router of the Internet service provider. For CEs at both ends, the PE router appears the same as a local Ethernet switch. However, this setup may have a scalability issue since different ports need to be allocated at both CE and PE router interfaces for different customers, so that the total number of customers that can be supported by PEs and CE-acting FEs is limited by the number of ports. To scale it Figure 2 illustrates how VMs that belong to the beyond the port limit, Q in Q encapsulation can be same customer network but reside in different edge used between CE-acting FE and PE routers. By doing domains communicate. Suppose VMs A and B belong so, each port can support up to 4K customers, each to the same subnet of the same customer. A resides in customer with 4K VLANs. VLAN x of edge domain e1, and B resides in VLAN y For customers that allocate cloud resources as of edge domain e2. Before A can send packets to B, it different subnets, layer 3 MPLS connection can be first discovers B’s MAC address by an ARP request. Sathyabama University

423


used. In this case, the CE-acting FE can serve as a virtual router that connects customer VPN with its data centre sub-nets. For small businesses or individual users, L2TP/IPSec can be used between the sites to provide basic connectivity and security without assistance from network ser-vice providers. IV. FURTHER DISCUSSION IV.1 Security via isolation and access control In SEC2, each user has its own private network and IP address space. The only way different users can communicate with each other is through FEs, which in turn use firewall, NAT, and other middle boxes to ensure proper access. Such isolation significantly reduces the security risk caused by resource sharing in cloud computing. For example, the attack outlined in [15] relies on determining co-resident VMs (i.e. VMs on same physical host). This is done by jointly using three methods: (1) determine matching Dom0 IP address via traceroute; (2) test for small round-trip time; and (3) check for numerically close IP addresses. None of the three methods would work in SEC2. For (1), traceroute is disabled between different customer networks. For (2), all packets across networks are forced to go through FEs even if source and destination co-locate at the same physical host, so the round-trip time cannot reveal location. (3) also does not work because each customer has private IP addresses. Like in many other systems, the overall security of the architecture depends on the security of its components. In SEC2 architecture, isolation among customers can still be compromised if hypervisors, switches, or middle boxes are compromised. Fortunately, security issues in switches and middle boxes are well studied, and hypervisor security has also received attentions from both industry and research community [21, 22]. IV.2 Cached vs. Installed forwarding table The MAC forwarding and policy table each FE is a fraction of the global table maintained at CC. There are two options to set up the FE table: either to maintain a cached copy of entries that are actively in use, or to install a complete table that includes MAC addresses of all those within the same subnet of the local VMs in this edge domain, and hence can potentially communicate with the local VMs.


We have assumed using cached table in our discussion in Section 3.3. Cached table is more scalable since typically a node only talks to a small fraction of other nodes, especially at a given point of time. Cached en-try is created at FE triggered by an unknown MAC address contained in ARP or data packets. Cached en-try is removed after a certain idle time, and can also be replaced by newer entries. The drawback of caching is that it may not be easy to keep track of which entry is cached where. Which option is better largely depends on how subnets are split across edge domains. IV.3 VM migration There are two ways VM migration may come into play. First, within the data center, the cloud provider may need to move re-optimize the placement of VMs to balance load, save power, or avoid resource fragmen-tation. In this case, VMs in the data center can be moved across physical machines either within or across edge domains. This is done by transferring dynamic VM state from source to destination hosts. Once state transfer is complete, a gratuitous ARP message is sent from the destination host to announce the VM’s new location. Note that this announcement only reaches hosts in the same VLAN in the same edge domain. If both source and destination hosts are in the same do-main, FE and CC are not involved in the process. If the VM is migrated to a different domain, then CC updates the VM’s location in its table, including both eid and VLAN id. In the second case, the customer may need to migrate VMs between its on-site network and data centre. This can be easily achieved if the customer’s network and its data centre network are configured to be in the same LAN, connected through layer 2 VPN such as VPLS. In this case, FE and CC will register the VM’s new location. But from the customer’s point of view, the migration procedure is the same as migrating VMs within its on-site LAN. Migration between customer site and data centre across different subnets is more challenging. Here a solution proposed in [16] can be used. We can deploy an FE at the edge of customer site network, which can register the location of the VM, and tunnel packets to and from FEs in the data centre. IV.4 Implementation considerations Conventional Ethernet switches can be used in both the core domain and the edge domains. The switches in the core domain are purely for packet forwarding, and hence the topology design is not limited by any policy. In order to ensure better resource usage, shortest-path frame routing [20] or other schemes that allow multiple paths being used should be deployed. The switches in edge domains need to handle different VLANs. They can be configured in a conventional tree-based topology, rooted at FEs. Conventional L2 forwarding is used in such switches.

424


NIC teaming or Spanning Tree Protocol (STP) can be used to provider redundancy within the edge domain. The new elements are the CC and FEs. The CC is essentially a directory server for storing configuration information. Standard techniques can be employed for its scalability and reliability. FEs are essentially high capacity Ethernet switches with flow forwarding and tunnelling capabilities. They needs to expose an API to allow CC to install flow forwarding entries and policy rules. An OpenFlow or SoftRouter like device is suitable for these functions. V. RELATED WORK Much recent work has been focusing on improving data centre networks [19, 18, 17, 13, 12]. Some of our design goals are similar to what has been addressed in recent work. Examples are location independence of VMs [19, 18], scalability [19, 18, 17, 13, 12], and utilization of o -the-shelf devices [17]. However, they do not address the issue of creating isolated private net-works in data centres, which is the focus of our design. Since our design does not pose any restrictions on the data centre fabric, it is possible to apply existing fabric designs such as [12] to the edge or core domain of our design. Joseph at al. have proposed a policy-aware switching mechanism for the data center network [6]. This fits well into our design: FEs can function as policyaware switches, and middle boxes such as load balancers and firewalls can be attached to FEs. Depending on the requirement of each customer, trac can be treated differently and forwarded through different middle boxes. CloudNet was recently proposed to support virtual private cloud service for enterprises [14], where different customers are assigned different VLANs in the data center. The main limitation is caused by the scalability is-sues of VLAN, as we have discussed before. This makes CloudNet more suitable for a relatively small number of enterprises. It is possible to solve VLAN scalability problem by using VPLS. Instead of using FEs to do VLAN remapping, VPLS capable routers can be used to extend VLANs across edge domains. The main differences that we see are: (1) VPLS is a distributed solution, where it may be more challenging to maintain and update policy set-tings of each customer on-the-fly; and (2) the switches in the core domain need to support MPLS when using VPLS, which is not required in SEC2.

side the technology domain, such as laws, regulations, human resource management and so on, it is important to explore technical solutions to this problem. In this paper, we have proposed SEC2, a scalable data center network architecture that intends to support secure cloud computing for both enterprise and individual users. It o ers e ective isolation between different customers while allowing physical resource shar-ing. Users can specify and manage their individual se-curity and QoS policy settings the same way as they manage the conventional on-site networks. This archi-tecture can also enable users to combine cloud-based resources seamlessly with their existing network infras-tructure through VPN. Unlike prior solutions, SEC2 eliminates the scalability limitation caused by VLANs. The architecture takes advantage of network virtualization and centralized control, using enhanced FE switches at certain ag-gregation points (i.e., between edge and core domain). For most part of the network, it relies o -theshelf layer 2 devices for both within each domain and in the core domain.

REFERENCES

[1] [2]

[3]

[4]

[5]

[6] [7]

[8]

[9]

[10]

[11]

[12]

VI. CONCLUSION Security is one major hurdle that cloud computing services need to overcome before their mass adoption. While certain aspects of security rely on solutions out-


425

Dark Reading, Security is chief obstacle to cloud computing adoption, study says, http://www.darkreading.com. D. A. Joseph, A. Tavakoli, and I. Stoica, A Policy-aware Switching Layer for Data Centers. In ACM SIGCOMM, 2008. J. Rexford, A. Greenberg, G. Hjalmtysson, D. A. Maltz, A. Myers, G. Xie, J. Zhan, and H. Zhang, Network-Wide Decision Making: Toward a Wafer-Thin Control Plane. In ACM SIGCOMM HotNets Workshop, 2004. T. Ristenpart, E. Tromer, H. Shacham, and S. Savage, Hey, You, Get O of My Cloud: Exploring Information Leakage in Third-Party Compute Clouds. In CCS, 2009. J. Touch, and R. Perlman, RFC 5556: Transparent Interconnection of Lots of Links (TRILL): Problem and Applicability Statement. http://www.ietf.org, 2009. Z. Wang, and X. Jiang, HyperSafe, a Lightweight Approach to Provide Lifetime Hypervisor R. Mysore, A. Pamboris, N. Farrington, N. Huang, P. Miri, S. Radhakrishnan, V. Subramanya, and A. Vahdat, PortLand: A Scalable Fault-Tolerant Layer 2 Data Center Network Fabric. In ACM SIGCOMM, 2009. F. Hao, T.V. Lakshman, S. Mukherjee, and H. Song, Enhancing Dynamic Cloud-based Services using Network Virtualization, In VISA, 2009. N. Farrington, E. Rubow, and A. Vahdat, Scaling Data Center Switches Using Commodity Silicon and Optics. In ACM SIGCOMM, 2008. T. Lakshman, T. Nandagopal, R. Ramjee, K. Sabnani, and T. Woo, The SoftRouter Architecture. In ACM HOTNETS, 2004. C. Guo, H. Wu, K. Tan, L. Shi, Y. Zhang, and S. Lu, DCell: A Scalable and Fault-Tolerant Network Structure for Data Centers. In ACM SIGCOMM, 2008. The Invisible Things Lab’s blog, http://theinvisiblethings.blogspot.com.


Simulation of C4ISR System Architecture for Lower Echelon of Armed Forces Shambhuling R. Doddamani, Manish Singh CAIR, DRDO, Bangalore, India [email protected], [email protected]

organisational structures operating in a dynamic environment. The commander and his staff interact with each other, aided by the available tools to achieve specified goals. To meet the complex operational requirements,C4ISR systems must be able to interoperate with other weapons systems as part of a large complex system or System-ofSystems (SoS). The C4ISR system shall assist to commander for surveillance capability, planning,

Abstract: The Command, Control, Communications, Computers and Intelligence Surveillance Reconnaissance (C4ISR) system is the main primary system of the military electronic information and communication system. The simulation of C4ISR system is essential for performance improvement, power estimation and interoperability issues of different specifications and make of systems. It is impractical to carryout large-scale tests in the field due to constrained resources and reduced availability of support units and sensors. The system simulation consists data/videos/images from sensors & communication system, day night images from sensor, positioning information from GPS/DRM, keyboard, LCD display with Intel Atom Processor and it provides latency, power consumed, power usage and power percent and data simulation for better performance of the C4ISR system for Lower Echelon of Armed forces.

mission coordination and communication between solider, leader and commander. II. BACKGROUND: The C4ISR systems are used in a variety of departments such as police, rail, airports, oil and gas where command and control scenarios exist. The most important focus of these subsystems is in Lower Echelon of armed forces application. The command and control system is the arrangement of personnel, information management, procedures and equipment and facilities essential to commander to conduct operations. The C4ISR system supports by performing three functions:

Keyword- C4ISR, Situational Awareness, COP, C2 Software, GPS/DRM I. INTRODUCTION: The Command, Control, Communications, Computers and Intelligence Surveillance Reconnaissance (C4ISR) systems provide situational awareness and battlefield information for the commanders to make decisions the military forces to accomplish Centre for Artificial Intelligence (CAIR), Defence Research &

· Creating and maintaining the common operational picture(COP)

and to control missions. The and Robotics Development

·Supporting decision making by improving its speed and accuracy Supporting preparation and communication of execution of information

Organisation (DRDO) leverages state-of-the-art information, communication, and Intelligence Surveillance Reconnaissance technology to design C4ISR systems that bring better situational awareness to the commanders. The C4ISR systems would provide comprehensive information to the commanders in a timely fashion and enable the

The developing of evaluation environments and tools has laid solid theoretical and practical foundation for C4ISR effectiveness evaluation. In recent years, lots of works have been done on the evaluation of C4ISR system architecture effectiveness all over the world.

commanders to disseminate orders expeditiously to the soldiers on the ground. This will enable the soldiers to execute their missions effectively. They consist of people, tools, processes, procedures, Sathyabama University

Australian Defense Organization Architecture Framework (ADOAF) is based on DODAF. It has three main views such as architecture product view

426


process/methodology view and tool/environment view. It has some additional products in Data view that are not formally adopted . Department of National Defense Architecture Framework (DNDAF) Canada was developed by Department of National Defense, Canada and its first version was released in June 2007. It has six views. Further, this architecture has been extended by Aegis R&D Canada with human view in September 2008. French Military Architecture (AGATE) was developed by General Delegation for Ordnance (DGA), France. The first version of AGATE was released in November 2001. AGATE is abbreviation of French words like “Atelier de Gestion de l’ArchiTEcture des systems d’erudition et de communication". The current version of AGATE is V 3.0 released in December 2005. This framework is organizes into five views such as stakes, business architecture, service oriented architecture, logical architecture and physical architecture. It has similar features to DODAF and MODAF.

Fig.1 C4ISR Design Architecture

FEA (Federal Enterprise Architecture) was developed by US Office of Management and Budget (OMB). The focus of this framework is civil, aegis, and intelligence that was determined by 124-member Working Group comprised of 30 U.S. Federal agencies. It consists of reference models such as Performance Reference Model, Business Reference Model, Service Component Reference Model, Data Reference Model and Technical Reference Model. III. C4ISR SYSTEM ARCHITECTURE: The C4I-ELITE design architecture has divided in three parts: Operational Architecture, System Architecture and Technical Architecture as indicated in Figure1.

Fig.2. C4ISR System

Soldier command control software which provides better situational awareness, decision making capability, and mission planning faster. Operational Architecture: The descriptions of task are mission/war fighting activities, logical connectivity, information exchanges between the systems and organization structures. System Architecture: The descriptions of task Sathyabama University

427


are GIS functionalities, data, functions, platforms to one another system, defines system interconnection, system performance, environmental conditions of system. Technical Architecture: The descriptions of task are C4ISR system activities, modularity and interoperability etc. The C4ISR system is integrated with Command and Control (C2) Software, Intelligence Surveillance Reconnaissance (ISR), Communication, Computing with display input as shown in Figure-2. The C4ISR system architecture has divided in many sub systems and these subsystems like Knowledge management subsystem, communication subsystem and integrated C2 subsystem shall be integrated with

• Knowledge Management Subsystem: Combining information producers and consumers, computational functions, and system management, the Knowledge Management Subsystem (KMS) is the overarching system for data, information, and enabling knowledge. •Communication

Fig.3. Conceptual, functional classification of C4ISR System

Subsystem: Containing session for operational plans with soldiers, leaders and commanders with communication and networking subsystems. It is to use of software allowing the data fusion of all the sensors and it circulates the information as per the operational plans. The simulation of C4ISR system provides performance, power estimation of system and processor for Lower Echelon of armed forces and it realizes the system activities as given below:

Communication system single or double as per required situation and Personal Area Network (PAN).


428


PAN-based power grid for appliances. Soldiers will only carry power that is needed for mission-specific configuration. The C4ISR system simulation provides power consumed power usage and power percent by processor and bus in this paper.

• Integrated C2 Subsystem: A subset of KMS and integrated components, the Information Systems Services (ISS) subsystem may be viewed as the information grid that provides Future Soldier connectivity. It is comprised of the GIS, Communication systems, and the Publish/Subscribe (Pub/Sub) framework. It can be called as Information Systems Services (ISS) subsystem.

V. PERFORMANCE AND POWER ESTIMATION OF C4ISR SYSTEM: IV. USAGE OF C4ISR SYSTEM FOR LOWER ECHELON OF ARMED FORCES:

The simulation of C4ISR system parameters are given below: Simulation Time: 3 msec Sensors image characteristics: X_Pixel: 640, Y_Pixel:480 FPS: 25, Color bits: 4 Image_Bytes :( X_Pixel*Y_Pixel *Color_Bits)/8 VPSS_Speed_Mhz: 405 Bus_Speed_Mhz:533 RAM_Speed_Mhz:866 Atom Processor Speed: 1330 MHz NVD sensor: X_Distance: 100.0 Y_Distance: 75.0 Link Speed: 75 0.0

The C4ISR system is to provide reception, analysis, modification, storage, display and transfer of digital data e.g. video, images, digital maps and graphics to the Lower Echelon of armed forces. The system shall aid to perform interactive (a) Command and Control Capability: To identify what commanders do and how they execute the task of leading their units to accomplish missions. Command and Control (C2) is an essential element of the art and science of warfare. No single specialized function, either by itself or combined with others, has a purpose without it. Commanders are responsible for C2. However, C2 is also of great concern of staff officers and some staff specialists. (b) Surveillance Capability: The situational awareness tool shall assist to soldier to know what is happening around him and his location on the battlefield with respect to enemy and own troops. (c) Planning: Capability of sending and receiving orders in graphical form with the use of digitized cartography on the basis of what has been previously planned and also cater for contingency plans. (d) Mission Coordination: Soldier has capability to know how and when to respond in coordination with other members of his team in order to act as one group with Situational awareness tool, messaging and communication system. (e) Communications: The soldier shall be equipped with communication system for inter-group and intra-group communications. (f) Integration of Equipments: The C4ISR capability shall allow the logical connection between each component of the equipment thereby achieving integrated C4ISR solutions for the armed forces.

The simulation of C4ISR realizes the latency is 2.2 x 10-5 at 0.01, 0.8, 1.55 msec and 0.99x10-5 at 2.37 msec simulation time of sensor in Figure-3, power consumed is 0.32mW at 2.4 msec simulation time in integrated system by processor in Figure-4, power percent is 60% in Atom processor and bus is 20% in integrated system by processor in Figure-5, power usage is 5 x 10-7 W for Atom processor and 1.5 x 10-7 W for bus in integrated system by processor in Figure-6, simulation GPS data is X-coordinate , Ycoordinate, Z-coordinate (Altitude) position showed in Figure-7 and detailed processor activity showed in Figure-8.

The C4ISR system goal is to drastically reduce power consumption, which will reduce system weight by extension. The C4ISR system do this through power management (instant on/off), multifunctionality, modularity, by issuing subsystems on a mission-specific basis, by distributing processing and power (appliance-based approach), and by using a


Fig. 4 Latency in integrated system by Processor

429


Fig. 5 Power consumed in integrated system by Processor

Fig. 6 Power Percent in integrated system by Processor

Fig. 7 Power Usage in integrated system by Processor


430


REFERENCES [1] Michael R., Lt Con Donald H., Using Army Force-on-Force Simulations to Stimulate C4I Systems for Testing and Experimentation [2] Huang zhonghua, Li Yinlin, “Model Research of the Military Electronic Information System” on 2010 International Conference on Computer, Mechatronics, Control and Electronics Engineering (CMCE) [3] ZHANG Ying-chao, Ye Feng, YU Qin-zhang, Chen Xin, “Research on Multi-Agent based C4ISR Effectiveness Simulation and Evaluation” on 2010 Second International Conference on Computer Modeling and Simulation [4] Lean Weng YEOH, Ming Chun NG, “Architecting C4I Systems” on Second International Symposium on Engineering Systems, MIT, Cambridge, Massachusetts, June 15-17, 2009 [5] M. Harn, V. Berzins, Luqi, W. Kemple, “Evolution of C4I Systems” Computer Science Department, Naval Postgraduate School, Monterey, CA 93943 [6] Mario R. Barbacci, William G. Wood “Architecture Tradeoff Analyses of C4ISR Products” on June 1999 [7] Albert R. Zehetner, Manager, Information Security Group, Electronic Warfare Associates - Australia “INFORMATION OPERATIONS: THE IMPACTS ON C4I SYSTEMS” on AOC International Symposium and Exhibition, Adelaide, Australia2004 [8] Michael R. Hieb, AB Technologies, Inc., Lit. Col. DONALD H. Timian, Army Model and Simulation Office “Using Army Force-on-Force Simulations to Stimulate C4I Systems for Testing and Experimentation” [9] Hand Tin French and Amanda Hutchinson, Defence Science & Technology Organisation, Australia “Measurement of Situational Awareness in a C4ISR Experiment” [10] Dr. Andreas Tolk, Virinia Modeling Analysis & Simulation Centre, Suffolk VA and Susan Solick ,TRADOC Analysis Centre, Fort Leavenworth, KS “Using the C4ISR Architecture Framework as a Tool to Facilitate V&V for Simulation Systems within the Military Application Domain” on April 2003 Spring Simulation Interoperability Workshop, Orlando, Florida

Fig. 8 Simulated GPS data

VI. CONCLUSION: The C4ISR system’s effectiveness simulation and evaluation is a very important research domain which is full of challenges. Based on analyzing the current research works on C4ISR effectiveness evaluation, it realizes performance and power estimation of the system-of-system. The C4ISR system simulation of data/videos/images from sensors & communication system, day night images from sensor, positioning information from GPS/DRM, keyboard, LCD display with Intel Atom Processor and it indicated latency, power consumed power usage and power percent.


431


Modeling Massive RFID Data Sets: A Gateway-Based Movement Graph Approach Mr. N.Krishnaiah # & G.Manikyalarao #

Assoc. Prof, CSE PG Student, Department of CSE, BVC Engg. College: Odalarevu, EG, AP, INDIA-533210 Email:* [email protected] & [email protected], Abstract— In this model Radio Frequency Identification (RFID) data sets are expected to become commonplace in supply chain management systems. It tells about movement graph model as a compact representation of RFID data sets. Since spatiotemporal as well as item information can be associated with the objects in such a model, the movement graph can be huge, complex, and multidimensional in nature.RFID holds the promise of real-time identifying, locating, tracking and monitoring physical objects without line of sight, and can be used for a wide range of pervasive computing applications. To achieve these goals, RFID data have to be collected, filtered, and transformed into semantic application data. Such data cannot be used directly by applications unless they are filtered and cleaned. To compensate for the inherent unreliability of RFID data streams, most RFID middleware systems employ a “smoothing filtering” and an efficient cubing algorithm that performs simultaneous aggregation of both spatiotemporal and item dimensions on a partitioned movement graph.Main goal of this research project is to design and develop an efficient, robust RFID stream processing system that addresses the challenges posed by the data-information mismatch, incomplete and noisy data, and high data volume, and enables real-time tracking and monitoring .

RFID-based pervasive computing environment. The filtered RFID data often need to preserve the original order, i.e., the first observed tagged object will be output first after filtering. Such order can be critical for many RFID applications. II. RADIO FREQUENCY IDENTIFICATION (RFID) Technology that allows a sensor (reader) to read, from a distance, and without line of sight, a unique electronic product code (EPC) associated with a tag.

Fig. 1 Radio Frequency Identification.

III. RFID FRAMEWORK ARCHITECTURE: Keywords— RFID, data warehousing, data models

I. INTRODUCTION RFID (Radio Frequency Identification) is a noncontact automatic identification technology, which aims at identifying and tracking items by using radio frequency electromagnetic wave to let readers capture the data on RF tags attached to them. However, unreliable data (original data) captured by readers is a major factor hindering the development of RFID technology. Under normal circumstances, it is quite often that the loss and error rate is between 30-40 %. For effectively and efficiently supporting high-level RFID business logic processing, it is necessary to provide high-quality RFID data. For that case, it is critical to clean the original data. One major problem to be solved in pervasive computing is to identify and track physical objects, and RFID technology is a perfect fit to solve this. By tagging objects with EPC 1 tags that virtually represent these objects, the identifications and behaviors of objects can be precisely observed and tracked. RFID readers can be deployed at different locations and networked together, which provides an


Fig. 2 RFID Framework Architecture

It is one of the first RFID integration platforms focusing on large-scale deployments. Its serviceoriented architecture provides network services to applications through several standard protocols and interfaces. Java System RFID Software consists of two major components: The event manager processes (filters and aggregates) RFID data, while the information server provides access to the business events generated by the event manager and serves as an integration layer that offers options for integrating with enterprise applications.

435


IV. PROBLEM DEFINITION In order to realize the full benefits of detailed object tracking information, we need to develop a compact and efficient RFID cube model that provides OLAPstyle operators useful to navigate through the movement data at different levels of abstraction of both spatiotemporal and item information dimensions. This is a challenging problem that cannot be efficiently solved by traditional data cube operators, as RFID data sets require the aggregation of highdimensional graphs representing object movements, not just that of entries in a flat fact table.

b.

Second Procedure

Find the edge, if that edge is removed from the graph then the graph splits into two disconnected components. c.

Third Procedure

Based on the large traffic flow through nodes and they can also be identified by using a concept of betweeness and centrality in social network analysis. Algorithm been used to find the Gateway node is Depth First Search (Backtracking is possible). ALGORTHIM TO CREATE A GRAPH

V. PROPOSAL STATEMENT

Step: 1 begin

In paper will be developed on effective and efficient RFID data filtering techniques to generate clean RFID data, which can be further interpreted and integrated into RFID-based applications.

Step :2 Read no of locations Step :3 Read the adjacency among locations (0 or 1) in an array x[i][j] .

In this paper, two types of filtering is proposed: noise is removed from RFID data (de-noising or smoothing), and duplicates are merged into one distinct reading (duplicate elimination, or merging).

Step :4 Initially all locations are un-visited i.e visited[I]$99) for storage. We’ve read of users getting "I think you are a robot" messages, like sometimes pop up on other Google properties. We haven't seen it, but it would alarm me. GAE's payment is also very fine-grained - only for the resources you use. Azure (and AWS) is "blocky" - you pay something for each server instance you will run (plus resources), irrespective of whether it gets any use at all. Azure seems to be better designed if we have a SOA-type approach.[5] Their architectures seem to benefit from experience in the enterprise world. GAE seems more focused on simply serving web pages. GAE has the lightest admin load. Once we’ve setup, deploying and re-deploying is quick and they will autoeverything. For example, you don't worry about how many servers our app is using, how to shard the data, how to load-balance. Mail just works. At the time of writing, Azure doesn't offer SMTP out so you need a 3rd party server. You can run the app under debug, put in breakpoints, etc. Azure has a "staging" environment where you can deploy to the cloud, but not make it live until we are happy it works. Great integration with many of the Google offerings was calendars, mail or whatever. You can delegate user management to Google if you don't want control over our user base. We are using .Net for other things, and integrating them with .Net on the backend is much easier than with GAE. (Update - using Java on GAE works fine and the 10-second timeout is now 30 seconds). With GAE you know any features


they add to the store, we will get. With Azure, we get the feeling Sql Azure Database will get most of the love but it will be more expensive. Azure Storage is likely to have the most got Chas.[5] No relational integrity, no order-by, we will fiddle with the in-memory context more. GAE's store has far fewer restrictions and more features than Azure Tables. Azure has two approaches to storage, offering more choice. They are SQL Azure Database (SAD) which is a relational DB, and Azure Storage, which consists of non-relational tables, blobs and queues. If we have an investment in SQL Server then SAD will be easy to move to, but is quite costly and might be less scalable. Good choice if we are using Python or JVM-based languages already. Many languages compile to Java byte code nowadays. Updating the app is very fast. For Python, We had a shortcut key setup and it took no time at all. We now use the Eclipse Plug-in for Java and it works very well. Integration with many MS "Live" offerings .A locally tested app will probably run on the cloud without (much or any) changes. With Azure, the config is different and we spent some time stopping-deleting-buildinguploading-starting before we got it right. GAE has a great UI that includes a log viewer a data editor. With Azure, you currently have to find external viewers/editors for this. GAE lets you have multiple versions of our application running on the same data store. We can deploy, test a version and then set the current 'live' version when you are ready. You can change back if something goes wrong. VII. WINDOWS AZURE FABRICS To use the Compute service, a developer creates a Windows application. This application might be written using C# and the .NET Framework, using C++ and the Win32 APIs, or in some other way. However it’s built, the application must be implemented as Web roles, Worker roles, or both. [8]As the name suggests, a Web role instance accepts Web requests. It can be created using ASP.NET or another technology that works with Internet Information Services (IIS). Whatever technology is used, Windows Azure provides built-in hardware load balancing across all Web role instances in a particular application. For functions that aren’t intended to respond directly to Web requests, a Windows Azure application can also contain Worker role instances. A Worker role instance is just a Windows application with a main(), and it can run indefinitely. Among other things, this model allows creating scalable applications where Web role instances accept requests, then pass them to Worker role instances to be processed. And while both Web role instances and

470


Worker role instances are ordinary Windows applications, a few things, such as logging, require direct access to Windows Azure. To allow this, applications can call directly into a Windows Azure agent. Notice that each instance—Web role or Worker role—runs inside its own virtual machine (VM). This provides isolation, letting Windows Azure applications run with full trust, and it also allows a clear view into application performance, since there’s a defined mapping between VMs and processor cores. But a developer doesn’t explicitly create VMs. Instead, she uploads an application to Windows Azure, together with an XML configuration file that specifies how many Web role instances and Worker role instances should run. Once this is done, Windows Azure creates the required number of VMs, then monitors their execution. If an instance fails, Windows Azure will start a new one, making sure that the specified number of Web role and Worker role instances are always running. (This work is done by the Fabric Controller, software that’s in charge of all machines in a particular instance of the Fabric.) To increase or decrease the number of running instances, the owner of an application can change the value for either instance type in the application’s configuration. Windows Azure automatically creates or shuts down VMs to match this new setting. Given that Windows Azure applications are essentially Windows apps, it shouldn’t be surprising that developers can create them with Visual Studio. This tool provides templates for creating cloud applications as Web roles, Worker roles, or both. Windows Azure also provides a Development Fabric, which is a facsimile of Windows Azure that runs on a local machine. Developers can use this to create their code and do initial testing, then upload the app to Windows Azure when it’s ready. The three kinds of Windows Azure storage are: Blobs: allow storing large binary objects, such as videos and images. Tables: provide highly scalable entity-based storage (not relational tables).


Queues: allow sending and receiving messages, such as between an application’s Web role instances and Worker role instances. It’s important to point out that all three of these can also be accessed by applications that aren’t running on the Windows Azure Compute service. For example, an onpremises or hosted application might choose to store large video files as Windows Azure blobs. The Windows Azure platform also includes SQL Azure Database (previously known as SQL Data Services). SQL Azure Database offers standard relational storage based on SQL Server, complete with stored procedures and more. While a single database in SQL Azure Database can’t hold as much information as a single Windows Azure Storage table, these databases offer a familiar storage model accessible through ADO.NET and other widely used data access mechanisms. VIII. THREE NEW CLOUD RESEARCH ENGAGEMENT PROGRAMS IN EUROPE Microsoft has initiated a large project, called VenusC, as part of the European Commission FP7 funding program. Venus-C will demonstrate how Windows Azure can interoperate with other cloud platforms and the existing European science grid. [7]In early 2011, the Venus-C project will request up to 20 submissions of cloud platforms to use in this interoperability study. In addition, seven initial cloud platforms are currently being ported to Windows Azure. This work will provide useful data to help facilitate interoperability among cloud platforms. The U.K. Engineering and Physical Sciences Research Council (EPSRC) is funding “digital hubs” to examine how new technologies can be used to enhance the quality of life for everyone. The emphasis is on developing innovative, inclusive products and services. Microsoft is involved through an agreement with the Horizon project, which is the "digital economy hub" based at the University of Nottingham. INRIA is France’s premier information technology research laboratory. The Microsoft Research-INRIA Joint Centre was founded by INRIA, Microsoft Corporation, and the Microsoft Research Laboratory Cambridge to pursue fundamental, long-term research in

471


formal methods, software security, and application of computer science research to the sciences. This agreement creates a new INRIA-Microsoft collaboration area on the topic of cloud computing. IX. CONCLUSION

[1]

[2]

For the developers who are addicted to Microsoft Technologies, Windows Azure surely provides a good web application cloud platform to develop their application. Moreover easy integration of .Net, PHP, python and Ruby ensures the naïve users about the uncomplicated possibility of development of web applications.


REFERENCES

[3] [4] [5] [6]

[7] [8]

472

M. Armbrust, A. Fox, R. Griffith, A. Joseph, R. Katz, A.Konwinski, G. Lee, D. Patterson, A. Rabkin, I. Stoica,M. Zaharia. Above the Clouds: A Berkeley View of Cloud computing. Technical Report No. UCB/EECS-2009-28, University of California at Berkley, USA, Feb. 10, 2009.. D. Chappell. Introducing the Azure services platform. White paper, Oct. 2010. Microsoft, 'Transforming IT with the Windows Azure Platform',White Paper, Nov. 2010. http://msdn.microsoft.com/en-us/library/gg432976.aspx . http://stackoverflow.com/questions/791447/windows-azure-vsamazon-ec2-vs-google-app-engine http://www.microsoft.com/enus/cloud/cloudpowersolutions/development-andhosting.aspx#tab3-tabs http://research.microsoft.com/en-us/projects/azure/ http://www.microsoft.com/windowsazure/ Whitepapers/AzureAndISV