Javaid's energy and enthusiasm opened my eyes to the importance of research. ...... consumption in the coronas near and farther from the sink is not balanced.
Balanced Load Distribution and Energy Efficient Routing to Avoid Energy Holes for Underwater Wireless Sensor Networks
By
Irfan Azam CIIT/SP14-RCS-001/ISB MS Thesis In Computer Science COMSATS Institute of Information Technology Islamabad - Pakistan
Spring, 2016
COMSATS Institute of Information Technology
Balanced Load Distribution and Energy Efficient Routing to Avoid Energy Holes for Underwater Wireless Sensor Networks A Thesis Presented to
COMSATS Institute of Information Technology, Islamabad In partial fulfillment of the requirement for the degree of
MS (Computer Science) By
Irfan Azam CIIT/SP14-RCS-001/ISB
Spring, 2016 ii
Balanced Load Distribution and Energy Efficient Routing to Avoid Energy Holes for Underwater Wireless Sensor Networks A Post Graduate Thesis submitted to the Department of Computer Science as partial fulfilment of the requirement for the award of Degree of MS (Computer Science).
Name Registration Number Irfan Azam CIIT/SP14-RCS-001/ISB
Supervisor:
Dr. Nadeem Javaid Associate Professor, Department of Computer Science, COMSATS Institute of Information Technology (CIIT), Islamabad Campus.
iii
Final Approval This thesis titled
Balanced Load Distribution and Energy Efficient Routing to Avoid Energy Holes for Underwater Wireless Sensor Networks By
Irfan Azam CIIT/SP14-RCS-001/ISB has been approved For the COMSATS Institute of Information Technology, Islamabad External Examiner:
Dr. Aamer Nadeem, Associate Professor, Head Department of Bioinformatics and Biosciences Head of Center for Software Dependability Capital University of Science & Information Technology (CUST), Islamabad Supervisor:
Dr. Nadeem Javaid Associate Professor, Department of Computer Science, COMSATS Institute of Information Technology, Islamabad HoD: Dr. M. Manzoor Illahi Tamimy Associate Professor, Department of Computer Science, COMSATS Institute of Information Technology, Islamabad
iv
Declaration I IRFAN AZAM (Registration No. CIIT/SP14-RCS-001/ISB) hereby declare that I have produced the work presented in this thesis, during the scheduled period of study. I also declare that I have not taken any material from any source except referred to wherever due that amount of plagiarism is within acceptable range.
If a
violation of HEC rules on research has occurred in this thesis, I shall be liable to punishable action under the plagiarism rules of the HEC.
Date:
June 19, 2016 Irfan Azam CIIT/SP14-RCS-001/ISB
v
Certificate It is certified that IRFAN AZAM (Registration No. CIIT/SP14-RCS-001/ISB) has carried out all the work related to this thesis under my supervision at the Department of Computer Science, COMSATS Institute of Information Technology, Islamabad and the work fulfills the requirement for award of MS degree.
Date: June 19, 2016
Supervisor:
Dr. Nadeem Javaid Associate Professor Department of Computer Science Head of Department:
Dr. M. Manzoor Illahi Tamimy Associate Professor Department of Computer Science
vi
DEDICATION
π
edicated
to my father
vii
ACKNOWLEDGEMENTS Alhamdulillah, all praises to Allah Almighty, the most Merciful and the most Gracious, for the strengths and His blessing in completing this thesis.
I would like to thank, Almighty Allah, who give me strength to complete this work under the kind supervision of Dr. Nadeem Javaid. He has his own style of supervision. Working on thesis with him was indeed a challenging task that demanded regularity and consistent hardwork. I feel proud to express my deepest sense of gratitude and appreciation to my supervisor Dr. Nadeem Javaid for his kind help, advice, inspired guidance, unlimited support, sympathetic attitude and sincere personal involvement throughout the study. He has been an incredibly supportive, inspirational, and helpful teacher, and mentor during my entire exciting journey at CIIT. Dr.
Nadeem
Javaidβs energy and enthusiasm opened my eyes to the importance of research. Every conversation with him is exciting, and I am thankful for his high-level perspective.
I would never have been able to reach this stage but for the prayers and great support of my family. I am also thankful to my parents who always give me lots of encouragement. Thanks and best wishes for all those who have made this learning experience so wonderful for me.
viii
ABSTRACT
Balanced Load Distribution and Energy Efficient Routing to Avoid Energy Holes for Underwater Wireless Sensor Networks Underwater Wireless Sensor Networks (UWSNs) consist of sensor nodes deployed to sense underwater environment.
Sensor nodes gather the required information and
report it to sink through a predefined routing path. Research community is getting interest in UWSNs due to its emerging applications such as costal surveillance for defense strategies, disaster monitoring, oil and mineral extractions, pollution monitoring etc. UWSNs are used for monitoring environments where humans access is quite impossible. However, UWSNs work for a limited time because battery power provided to sensor nodes are difficult to replace or recharge due to harsh underwater environment. Therefore, an energy efficient routing protocol is required to make these networks able to work for longtime. This research work aims to provide energy efficient routing schemes to reduce energy consumption of sensor nodes. Energy consumption is reduced by dividing network field into subregions and using clustering technique in dense regions of network.
Cluster head node in each dense region is
selected in each transmission round to locally gather data from sensor nodes. Each cluster head forwards compressed data to sink using multi-hop routing technique to save energy.
Mobile sinks are used to gather data in sparse regions where nodes
are not able to forward data because of limited transmission range. In addition, an energy minimization technique is proposed to prolong network lifetime for UWSNs designed for continuous monitoring applications. Apart from this, we propose a routing scheme to avoid energy holes in UWSNs. Energy holes are formed due to unbalanced load mostly on sensor nodes near to sink.
A balanced load distribution scheme is
proposed to overcome the energy hole problem in UWSNs.
Extensive simulations
are performed by considering different node deployment like uniform and random in homogeneous and heterogeneous environments. The results show that our proposed protocols perform better in terms of stability, energy efficiency and network lifetime.
ix
Contents Dedication
vii
Acknowledgements
viii
Abstract
ix
List of Publications
xiii
List of Figures
xv
List of Tables
xvii
List of Algorithms
xviii
1 Introduction 1.1
1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 SEEC Protocol for UWSNs
2
8
2.1
Summary of the Chapter . . . . . . . . . . . . . . . . . . . . . . . . .
9
2.2
Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
2.3
Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
2.4
SEEC: The Proposed Scheme
. . . . . . . . . . . . . . . . . . . . . .
16
. . . . . . . . . . . . . . . . . . . . . . . . . .
17
2.4.1
Network Model
2.4.2
Mathematical Model
. . . . . . . . . . . . . . . . . . . . . . .
17
2.4.3
Network Configuration . . . . . . . . . . . . . . . . . . . . . .
20
2.4.4
Searching Sparse and Dense Regions
20
x
. . . . . . . . . . . . . .
2.5
2.4.5
Clustering in Dense Region
. . . . . . . . . . . . . . . . . . .
2.4.6
Sink Mobility in Sparse Region
2.4.7
Data Transmission
21
. . . . . . . . . . . . . . . . .
23
. . . . . . . . . . . . . . . . . . . . . . . .
25
. . . . . . . . . . . . . . . . . . . . . . . . .
25
2.5.1
Network Lifetime . . . . . . . . . . . . . . . . . . . . . . . . .
26
2.5.2
Stability Period and Instability Period
. . . . . . . . . . . . .
27
2.5.3
Network Residual Energy
. . . . . . . . . . . . . . . . . . . .
27
2.5.4
Packets Sent and Packets Received
Performance Evaluation
. . . . . . . . . . . . . . .
28
. . . . . . . . . . . . . . . . . . . . . . . . . .
28
2.6
Performance Tradeoffs
2.7
Conclusion of the Chapter
. . . . . . . . . . . . . . . . . . . . . . . .
30
3 EMT: Energy Efficient Routing Protocol to Maximize Network Lifetime in UWSNs 34 3.1
Summary of the Chapter . . . . . . . . . . . . . . . . . . . . . . . . .
35
3.2
Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
35
3.3
Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
3.4
EMT: The Proposed Scheme . . . . . . . . . . . . . . . . . . . . . . .
41
3.4.1
Network Model
. . . . . . . . . . . . . . . . . . . . . . . . . .
41
3.4.2
Network Configuration . . . . . . . . . . . . . . . . . . . . . .
41
3.4.3
Forwarder Node Selection
42
3.4.4
Data Distribution and Data Transmission
3.5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
43
Performance Evaluation
. . . . . . . . . . . . . . . . . . . . . . . . .
43
3.5.1
Simulation Setup
. . . . . . . . . . . . . . . . . . . . . . . . .
43
3.5.2
Metrics
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
3.5.3
Energy Consumption . . . . . . . . . . . . . . . . . . . . . . .
45
3.5.4
Stability and Instability Period
. . . . . . . . . . . . . . . . .
45
3.5.5
Network Lifetime . . . . . . . . . . . . . . . . . . . . . . . . .
46
3.5.6
Implementation of EMT in Homogeneous and Heterogeneous Environments . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6
Performance Tradeoffs
. . . . . . . . . . . . . . . . . . . . . . . . . .
xi
47 48
3.7
Conclusion of the Chapter
. . . . . . . . . . . . . . . . . . . . . . . .
48
4 BLOAD: Energy Holes Avoidance with Balanced Energy Consumption for UWSNs 52 4.1
Summary of the Chapter . . . . . . . . . . . . . . . . . . . . . . . . .
53
4.2
Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
4.3
Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
4.4
System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
58
4.4.1
The Underwater Channel Model . . . . . . . . . . . . . . . . .
59
4.4.2
Energy Consumption Model . . . . . . . . . . . . . . . . . . .
60
The BLOAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
61
4.5.1
Network Configuration . . . . . . . . . . . . . . . . . . . . . .
61
4.5.2
Load Distribution Phase . . . . . . . . . . . . . . . . . . . . .
61
4.5.3
Data Transmission Phase . . . . . . . . . . . . . . . . . . . . .
63
4.5
4.6
Performance Evaluation
. . . . . . . . . . . . . . . . . . . . . . . . .
65
4.7
Performance Tradeoffs
. . . . . . . . . . . . . . . . . . . . . . . . . .
79
4.8
Conclusion of the Chapter
. . . . . . . . . . . . . . . . . . . . . . . .
5 Conclusion 5.1
80
86
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xii
87
List of Publications Journal Publication 1.
I. Azam,
N. Javaid,.., βBLOAD: Energy holes avoidance with balanced energy
consumption for underwater wireless sensor networks,β Journal of Network and Computer Applications, 2016. (Submitted)
Conference Proceedings 2.
I. Azam, A. Majid, I. Ahmad, U. Shakeel, H. Maqsood, Z.A. Khan, U. Qasim, and
N. Javaid. 2016, March. βSEEC: Sparsity-aware energy efficient clustering protocol for underwater wireless sensor networksβ. In 2016 IEEE 30th International Conference on Advanced Information Networking and Applications (AINA) (pp. 352-361). IEEE.
3. A. Majid,
I. Azam,
A. Waheed, Zain-ul-Abidin, M., T. Hafeez, Z.A. Khan, U.
Qasim, and N. Javaid, 2016, March. βAn energy efficient and balanced energy consumption cluster based routing protocol for underwater wireless sensor networksβ. In 2016 IEEE 30th International Conference on Advanced Information Networking and Applications (AINA) (pp. 324-333). IEEE.
4. T. Khan, I. Ahmad, W. Aman,
I. Azam, Z.A. Khan, U. Qasim, S. Avais, and N.
Javaid, 2016, March. βClustering depth based routing for underwater wireless sensor networksβ.
In 2016 IEEE 30th International Conference on Advanced Information
xiii
Networking and Applications (AINA) (pp. 506-515). IEEE.
5.
I. Azam, A. Majid, T. Khan, Sajjad, Z.A. Khan, U. Qasim, and N. Javaid, 2016,
βAvoiding Energy Holes in Underwater Wireless Sensor Networks with Balanced Load Distributionβ. In 2016 IEEE 10th International Conference on Complex, Intelligent and Software Intensive Systems (CISIS), IEEE. (Accepted)
6.
A. Majid,
I. Azam,
T. Khan, Sangeen, Z.A. Khan, U. Qasim, and N. Javaid,
2016, βA reliable and interference-aware routing protocol for underwater wireless sensor networksβ. In 2016 IEEE 10th International Conference on Complex, Intelligent and Software Intensive Systems (CISIS), IEEE. (Accepted)
xiv
List of Figures 1-1
Real Life Applications of WSNs . . . . . . . . . . . . . . . . . . . . .
2
2-1
Network Model of SEEC . . . . . . . . . . . . . . . . . . . . . . . . .
17
2-2
Division of Network Field into Left and Right
Regions. . . . . . . .
20
2-3
Working Flow of SEEC . . . . . . . . . . . . . . . . . . . . . . . . . .
22
2-4
Scenario of CH Selection in SEEC . . . . . . . . . . . . . . . . . . . .
23
2-5
Lifetime of Network . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
2-6
Stability and Instability Period of Network . . . . . . . . . . . . . . .
27
2-7
Total Network Residual Energy
. . . . . . . . . . . . . . . . . . . . .
28
2-8
Total Packets Sent to Sink . . . . . . . . . . . . . . . . . . . . . . . .
29
2-9
Total Packets Sent or Received at Sink Per Round . . . . . . . . . . .
29
3-1
Network Model of EMT Scheme . . . . . . . . . . . . . . . . . . . . .
42
3-2
Energy Consumption of EMT Scheme . . . . . . . . . . . . . . . . . .
45
3-3
Number of Dead Nodes in Network
. . . . . . . . . . . . . . . . . . .
46
3-4
Network Lifetime of EMT Scheme . . . . . . . . . . . . . . . . . . . .
47
4-1
Network Model of BLOAD Scheme
59
4-2
Network Linear Model of BLOAD Scheme
. . . . . . . . . . . . . . .
65
4-3
Working Flow of BLOAD Scheme . . . . . . . . . . . . . . . . . . . .
66
4-4
Comb:01: {π
1
π, π 2 2π, π 3 ππ‘π₯}
. . . . . . . . . . . . . . . . . . . . .
70
4-5
Comb:02: {π
1
2π, π 2 π, π 3 ππ‘π₯}
. . . . . . . . . . . . . . . . . . . . .
71
4-6
Comb:03: {π
1
ππ‘π₯, π 2 π, π 3 2π}
. . . . . . . . . . . . . . . . . . . . .
72
4-7
Comb:04: {π
1
π, π 2 ππ‘π₯, π 3 2π}
. . . . . . . . . . . . . . . . . . . . .
73
xv
n
. . . . . . . . . . . . . . . . . . .
4-8
Comb:05: {π
1
2π, π 2 ππ‘π₯, π 3 π}
. . . . . . . . . . . . . . . . . . . . .
74
4-9
Comb:06: {π
1
ππ‘π₯, π 2 2π, π 3 π}
. . . . . . . . . . . . . . . . . . . . .
75
4-10 FNDT and ANDT of Homo-BLOAD Scheme . . . . . . . . . . . . . .
78
4-11 Residual Energy of Homo-BLOAD, Hetero-BLOAD and Hetero-BR Schemes
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
81
4-12 Energy Consumption of Homo-BLOAD, Hetero-BLOAD and HeteroBR Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
82
4-13 Residual Energy for All Possible Combinations of Transmission Ranges and Load Weights of Both Homo-BLOAD and Hetero-BLOAD Schemes 83
xvi
List of Tables 2.1
Comparison of UWSNs Routing Protocols
. . . . . . . . . . . . . . .
15
2.2
Comparison of WSNs Clustering Protocols . . . . . . . . . . . . . . .
16
2.3
Performance Tradeoffs
30
3.1
State-of-the-art Related Work
. . . . . . . . . . . . . . . . . . . . . .
40
3.1
State-of-the-art Related Work
. . . . . . . . . . . . . . . . . . . . . .
41
3.2
Parameters Setting for Simulation . . . . . . . . . . . . . . . . . . . .
44
3.3
Performance Tradeoffs
48
4.1
State-of-the-art Related Work
4.2
Possible Combinations of Two Sets W and R.
. . . . . . . . . . . . .
62
4.3
Packet Load Distribution of BLOAD Scheme . . . . . . . . . . . . . .
63
4.4
Performance Tradeoffs
76
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
xvii
58
List of Algorithms 1
SSA
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
2
DSA
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
xviii
Chapter 1 Introduction
1
Chapter 1
1.1 Introduction Involvement and importance of Wireless Sensor Networks (WSNs) in our daily life is increasing day by day with the need and use of automated systems. We categorize WSNs into the following main categories on the basis of the implementation environment: Terrestrial WSNs (TWSNs), Mobile WSNs (MWSNs), Wireless Body Area Networks (WBANs), Wireless Underground Sensor Networks (WUSNs), Underwater WSNs (UWSNs), etc. Applications of WSNs are shown in fig. 1-1. Research community is getting interest in UWSNs from the last 15 years [1]. UWSNs are getting importance due to demanding oceanic applications i.e. ocean surveillance for defense strategies, underwater explorations, tsunami and earthquake monitoring, pollution monitoring, etc. Designing, manufacturing, and deploying the UWSNs are difficult tasks and underwater sensor nodes are costly [2]. Moreover, it is challenging to design a protocol for UWSNs by considering the limited parameters of underwater communication. The protocols designed for TWSNs cannot work well for UWSNs because of different implementation environment.
Radio and optical signals are mostly affected by
absorption loss in underwater environment and cannot be used for underwater communication. Acoustic signals are used as transmission media in UWSNs instead of radio signals, which is quite challenging due to the following reasons [1β3].
WSNs
Healthcare Informaon (WBANs)
War Mine Detectors (TWSNs)
Wild Life Monitoring (MWSNs)
Mineral Extracons (UWSNs)
Agricultural Works (WUSNs)
Figure 1-1: Real Life Applications of WSNs
β
Bandwidth of acoustic signals is low than radio signals.
β
Speed of acoustic signals is 1500 m/s while radio signals move with the speed of
3 Γ 108
m/s.
2
1.1. Introduction β
Chapter 1
Multipath fading, noise, path loss, transmission loss effects the underwater acoustic channel and leads to high Bit Error Rate (BER).
Generally, UWSN model consists of static sink(s) at the top of the water surface and sensor nodes are deployed randomly. Sensor nodes send information to sink(s) through multi-hoping which results in data corruption due to high link impairments and noise. To reduce this problem mobile sink(s) are used in [4]. Static sinks also create hotspot problem in which sensor nodes near the sink die earlier leaving some areas of a network field unmonitored.
Mobile sink(s) are useful to overcome such
problems [5]. In UWSNs, lowest depth sensor nodes are frequently selected as data forwarders, due to this, they die earlier than other nodes in the network which creates a routing hole in the network.
The network splits due to these routing holes
which affects the network efficiency and lifetime [6]. The phenomena of quick energy consumption of a sensor node due to unbalanced load of data forwarding is called energy hole.
The energy hole problem causes death of a sensor node earlier than
other nodes which creates a coverage hole due to which the area of this sensor node remains unobserved or un-sensed [7]. In addition, underwater sensor nodes are large in size so, sensor nodes are deployed with the help of ships, which is costly.
The
sensor nodes are sparsely deployed due to high design and manufacturing cost and also sensor nodes monitor large area under water. Therefore, sparse deployment is main factor in implementation and design process of sensor nodes [8]. Many researchers have addressed the issues in UWSNs and more research work has been done on physical and MAC layers issues than network layer. Routing of data gathered by sensor nodes through a predefined path to the sink with minimum delay and less energy consumption in UWSNs is one of the main challenges on network layer. Energy efficient routing in UWSNs is very important to save energy consumption of underwater sensor nodes with limited battery power. Therefore, an energy efficient routing protocol needs to be designed for a prolonged network lifetime. Manual node deployment is one of the important characteristics of UWSNs.
It
should be exploited for less energy consumption and minimum transmission delay. Sensor nodes can acquire the knowledge of their location as well as of their neighbor
3
Chapter 1 nodes if sensor nodes are manually bottom anchored in a static UWSNs. We can take advantage of this characteristic which satisfies our application requirement in order to save energy. In UWSNs, sensor nodes are deployed to sense the underwater environment and gather information. The sensor nodes forward the gathered data to the sink through a predefined routing path. An easy way to send data to the sink is direct transmission if transmission range of each sensor node is large enough to reach the sink. However, in direct transmission the nodes farther from the sink quickly drain out of energy than other nodes due to transmitting data at long distance. In order to save energy consumption multihop routing is used, in which each sensor node transmits data hop-by-hop. However, in multihop routing the nodes near the sink quickly run out of energy due to unbalanced data load. Thus, sensor nodes near the sink die quickly within no time and leads to the energy hole near the sink. The authors in [9] analyzed that when the sensor nodes near the sink drain out energy the nodes away from the sink still have 93% of the residual energy. Load balancing is an important technique for balanced energy consumption of all sensor nodes specially, for smooth energy consumption of the nodes near to the sink. Mixed routing scheme [10], [11], is mostly used to deal with unbalanced data load problem.
Each node in the mixed routing
scheme alternates between hop-by-hop and direct transmission to save its energy. This thesis is organized as follows.
A Sparsity-aware Energy Efficient Cluster-
ing (SEEC) protocol is presented in chapter 2 for energy efficiency in UWSNs. The network field is divided into subregions in order to find sparse and dense regions. Clustering technique is used in dense regions to reduce energy consumption of nodes. Each dense region act as a static cluster. A node is selected as a cluster head (CH) in each dense region, which locally gathers data from rest of sensor nodes of that region. CHs forward compressed data to sink in multi-hop fashion. In sparse regions of network, two mobile sinks are used to gather data to improve energy efficiency. In chapter 3, an Energy Minimization Technique (EMT) is discussed. EMT is proposed for UWSNs which are designed for continuous monitoring applications. Energy efficiency and network lifetime is achieved by considering circular network field with
4
1.1. Introduction
Chapter 1
random deployment of nodes. Network field is divided into sectors and a forwarder node is selected in each sector to gather data and report it to sink. In addition, nodes also forward data directly to sink to minimize load on forwarder node. Balanced Load Distribution (BLOAD) routing protocol is presented in chapter 4 to equally distribute load among nodes for balanced energy consumption. Each node forwards data to sink using three transmissions; one-hop (r), two-hop (2r) and directly transmission (dtx) to avoid energy holes through balanced distribution of data. Finally, summary of this thesis and future work is given in conclusion section of chapter 5.
5
References SoftWater: Software-defined networking for nextgeneration underwater communication systems. Ad Hoc Networks. 2016.
[1] Akyildiz IF, Wang P, Lin SC.
WDFAD-DBR: Weighting depth and forwarding area division DBR routing protocol for UASNs. Ad Hoc Networks.
[2] Yu H, Yao N, Wang T, Li G, Gao Z, Tan G. 2016. [3] Muhammad ayaz, abdullah azween.
issues and future challenges.
Underwater wireless sensor networks: routing
Proceedings of the 7th international conference on
advances in mobile computing and multimedia, acm, 2009.
DEADS: Depth and Energy Aware Dominating Set Based Algorithm for Cooperative Routing along with Sink Mobility in Underwater WSNs. Sensors, 2015.
[4] Umar A, Javaid N, Ahmad A, Khan ZA, Qasim U, Alrajeh N, Hayat A.
An adaptive surface sink redeployment strategy for Underwater Sensor Networks. In Proceedings of the
[5] Al-Bzoor M, Zhu Y, Liu J, Ammar R, Cui JH, Rajasekaran S. IEEE ISCC, Split, Croatia, 2013.
An energy efficient localization-free routing protocol for underwater wireless sensor networks. International journal of distributed sensor net-
[6] Wahid A, Kim D. works, 2012.
Energy Consumption Model for Density Controlled Divide-and-Rule Scheme for Energy Efficient Routing in Wireless Sensor Networks. International Journal of Adhoc and Ubiquitous Com-
[7] Latif K, Javaid N, Najmus M, Saqib ZA, Alrajeh N.
puting, 2015.
Greedy routing in underwater acoustic sensor networks: a survey. International Journal of Distributed
[8] Mohammad Taghi Kheirabadi, Murtadha Mohamad Mohd. Sensor Networks, 2013.
Avoiding energy holes in wireless sensor networks with nonuniform node distribution. IEEE Transactions on Parallel and Distributed Sys-
[9] Wu X, Chen G, Das SK. tems, 2008. [10] Guo W, Liu Z, Wu G.
works.
An energy-balanced transmission scheme for sensor net-
In Proceedings of the 1st International Conference on Embedded Net-
worked Sensor Systems, ACM, 2003.
6
REFERENCES
Chapter 1
[11] C. Efthymiou, S. Nikoletseas, and J. Rolim.
in Wireless Sensor Networks
Energy Balanced Data Propagation
Proc. 18th Intβl Parallel and Distributed Processing
Symp. (IPDPS β04), 2004.
7
Chapter 2 SEEC Protocol for UWSNs
8
2.1. Summary of the Chapter
Chapter 2
2.1 Summary of the Chapter Many routing protocols are proposed regarding energy efficiency in UWSNs.
We
propose SEEC protocol for UWSNs. SEEC specially search sparse regions of the network. We divide the network region into subregions of equal size and search sparse and dense regions of the network field with the help of Sparsity Search Algorithm (SSA) and Density Search Algorithm (DSA). SEEC improves network lifetime through sink mobility in sparse regions and clustering in dense regions of the network.
SEEC
also achieves network stability with optimal number of clusters formation in dense regions of the network where each dense region logically represents a static cluster. The division of the network region into subregions control routing hole problem in the UWSNs.
SEEC minimizes network energy consumption with balanced scheme
operations. Effectiveness of our proposed protocol is verified by simulation results.
2.2 Motivation Depth Base Routing (DBR) [1] and Energy Efficient DBR (EEDBR) [2] are localization free routing protocols for UWSNs. In DBR, forwarder sensor nodes are selected on the basis of depth while in EEDBR, selection of forwarders is based on residual energy and depth. Therefore, low depth and high residual energy sensor nodes die earlier than other sensor nodes in the network due to unbalanced load. The unbalanced load of data forwarding causes quick energy depletion of sensor nodes which is called as energy hole phenomena. Due to energy hole problem some of the sensor nodes die earlier than other sensor nodes of the network which creates a coverage hole. The energy and coverage hole problems maximize energy consumption and network lifetime decreases due to these consequences. We proposed SEEC routing protocol for UWSNs. SEEC divides network region into ten subregions and searches for sparse and dense regions of the network. We use SSA and DSA algorithms to search sparse regions and dense regions of the network. SEEC achieves network stability and energy efficiency with static clustering technique in dense regions of the network. Two
9
Chapter 2 mobile sinks are used in SEEC to collect data from sparse regions of the network. Thus SEEC achieves better network lifetime than DBR and EEDBR.
2.3 Related Work Most of researchers have worked on energy efficiency in UWSNs. Some of the related papers are discussed and summary of papers are given in table 2.1 and table 2.2.
et al., proposed Weighting Depth and Forwarding Area Division DBR
Yu, Haitao,
routing protocol for UWSNs, called WDFAD-DBR [3].
WDFAD-DBR reduces the
probability of encountering void holes problem. WDFAD-DBR improves the communication reliability in sparse regions of the network by advance sensing of void holes. It also reduces energy consumption in dense regions of the network through dividing forwarding regions according to the nodes density and channel condition.
et al.,
Latif, Kamran,
presented a clustering technique at routing layer for terres-
trial WSNs called Density controlled Divide-and-Rule (DDR) [4]. DDR have constant number of CHs in each round and it is based on static clustering. In DDR, the network field is divided into logical segments in order to reduce communication distance between sensor node to CH and between CH to sink(s). DDR resolves energy and coverage hole problems by controlling the network density and it keeps number of CHs constant throughout network operations. DDR improves energy utilization and also helps in uniform load distribution in terrestrial WSNs.
Many techniques are
proposed for detection of coverage hole problems in terrestrial WSNs.
In [5], Jing
et al, proposed Boundary Detection Method (BDM) for large-scale coverage holes in terrestrial WSNs which is based on minimum critical threshold constraint. BDM has low computational complexity when detecting large-scale coverage holes in terrestrial WSNs.
et al.,
Latif, Kamran,
discussed the energy hole problem also in DR [6] and pro-
posed a new hybrid approach of static clustering and dynamic selection of CHs for terrestrial WSNs.
The proposed technique is useful for energy hole minimization
and maximum network lifetime is achieved with DR. The network area is divided
10
2.3. Related Work
Chapter 2
into rectangular and square-shaped sub-regions with the help of which the network coverage and cluster load balancing is achieved.
et al.,
Heinzelman, Wendi Rabiner,
proposed a first clustering base protocol for
terrestrial WSNs (LEACH) [7]. In LEACH, local clusters of nodes are formed on the basis of minimum distance. In each cluster, CH collects data from nodes and then forward it to a Base Station (BS) after data aggregation and fusion. Due to only selected CHs transmission to a BS, LEACH achieves energy efficiency. In LEACH, CHs are randomly rotated through a defined mechanism, in order to balance energy consumption of nodes. In LEACH, CH selection is not optimum in number which results in clusters formation of different sizes.
Thus leads to unbalance energy consump-
tion. TEEN [8] overcome deficiencies of LEACH by introducing clustering scheme in reactive WSNs.
SEP [9] and DEEC [10] are routing protocols for WSNs and used
clustering for heterogeneous WSNs. In SEP and DEEC, two different energy levels are defined.
Nodes are categorized on the basis of these energy levels i.e., normal
nodes and advanced nodes. Ahmad, Ashfaq,
et al.
presented away CHs with Adaptive Clustering Habit in [11]
for terrestrial WSNs. The free association mechanism; in which nodes associate with CHs, reduces communication distance, avoids back transmission. energy consumption and increases network lifetime.
Thus minimizes
First, CHs are elected on the
basis of threshold value. Then, through natural selection phase optimal number of CHs are selected on the basis of optimal distance between them due to which balanced load on each CH is achieved.
π΄πΆπ» 2
achieves maximum throughput and prolonged
network lifetime in homogeneous, heterogenous, reactive and proactive protocols for WSNs. Liao, Yifan,
et al.
proposed a load-balanced clustering algorithm for terrestrial
WSNs [12]. The load-balanced clustering algorithm creates clusters on the basis of residual energy of sensor nodes and density distribution of sensor nodes. The balanced clustering mechanism in the network overcome the unbalancing and non-uniform load problem.
Energy consumption is reduced and network lifetime is enhanced with
load balancing problem.In [13], Genetic Algorithm based Energy Efficient Clustering
11
Chapter 2 Hierarchy (GAECH) technique is proposed for load balancing between sensor nodes of WSN. In order to increase stability period and network lifetime, GAECH uses fitness function to form well balanced clusters in the network. Tarhani, Mehdi,
et al.,
proposed Scalable Energy Efficient Clustering Hierarchy (SEECH) [14]. In SEECH, nodes are divided into three types; member nodes, CHs and relays. Generally, selected CHs collect data from sensor nodes as well as relay data to sink(s) which reduces energy consumption of CHs and minimizes network lifetime. Clustering schemes are proposed in [15β19] for routing in UWSNs. The working principle is same in UWSNs and terrestrial WSNs. Two phases are most common in clustering which are: CH selection phase and data communication phase. In phase-I, CH(s) are selected for a cluster on the basis of residual energy and location. In most of the clustering schemes residual energy and location metrics are used for the selection of a CH. Phase-II is data communication phase, the sensor nodes gather data and send the data to their respective CH which is in their transmission range. The CH receives and aggregates the data and sends it to sink(s) through multi-hop routing. To avoid collision of packets from sensor nodes TDMA is used for data communication in clustering. For collecting information from the network field algorithm like Artificial Bee Colony (ABC) is proposed in a clustering technique which increases lifespan of a network. Also, Ant Colony Optimization (ACO) technique is used to achieve better network lifetime. In [20], hybrid swarm intelligence energy efficient clustered routing algorithm for WSNs is proposed. ABC and ACO are combined and a new algorithm ABCACO is designed. ABCACO works in three phases. In first phase, optimal number of subregions and subregion parts are selected. In second phase, CHs are selected using ABC algorithm in subregions. Final phase, is the efficient data transmission phase.
In this phase, data is transmitted by CHs using ACO algorithm.
ACO al-
gorithm chooses best route to BS for minimum energy consumption of sensor nodes. ABCACO achieves better stability period than LEACH clustering protocol. Umar, Amara,
et al.
proposed Depth and Energy Aware Dominating Set (DEADS)
based algorithm for cooperative routing along with sink mobility [21] in UWSNs. It is
12
2.3. Related Work
Chapter 2
a routing protocol for reactive UWSNs, in which data transmission occurs only when a predefined depth threshold is met.
DEADS is not suitable for continuous data
monitoring. DEADS achieves better network lifetime with sink mobility and achieves maximum throughput due to cooperative routing. In DEADS, energy of sensor nodes is compromised due to cooperative routing. Hence, maximum throughput is achieved at the cost of increased energy consumption of the sensor nodes. Stability period is compromised for reliable data transmission. In [22], presented Energy-Aware Sink Relocation (EASR) strategy for mobile sink in terrestrial WSNs. Sensor nodes near the sink die earlier due to unbalanced load on these nodes which reduces network lifetime. EASR, uses information related to residual energy of sensor nodes to adjust transmission range of the sensor nodes adaptively and sink is relocated accordingly for uniform load on all of the sensor nodes in the network. Thus EASR achieves maximum network lifetime.
et al.
Javaid, Nadeem,
proposed Improved Adaptive Mobility of Courier nodes in
Threshold-optimized DBR (iAMCTD) [23] to increase network lifetime of reactive UWSNs.
iAMCTD is a forwarding-function
πΉπΉ
based routing protocol.
iAMCTD
is suitable for delay-sensitive applications due to variation in depth-threshold which increases the forwarders. Throughput in iAMCTD is decreased as it avoids unnecessary data transmissions.
In iAMCTD, optimized sink mobility is introduced to
minimize end-to-end delay for delay sensitive applications.
Optimized mobility of
sink in iAMCTD also minimizes delay in sparse conditions. Javaid, Nadeem,
et al.
proposed an Autonomous Underwater Vehicle (AUV)-Aided
Efficient Data-Gathering (AEDG) routing protocol [24].
AEDG makes reliable de-
livery of data in UWSNs. In AEDG, AUV is used for data gathering from gateway nodes and Shortest Path Tree (SPT) algorithm is used for associating sensor nodes with the gateway nodes to achieve maximum network lifetime.
In order to mini-
mize the network energy consumption, there are only limited number of sensor nodes associated with each gateway node. In [25], a Distributed Data Gathering (DDG) protocol is presented using AUV in UWSNs. In DDG, clustering technique is used where sensor nodes form clusters
13
Chapter 2 and a single node is selected as a CH to collect data from sensor nodes. Generally, AUV visits every sensor nodes or CHs, where in DDG, AUV collects data from only selected nodes. In this way, transmission power of sensor nodes is reduced and the lifetime of a network is increased. In [26], Jiang, Peng,
et al.
presented Node Non-uniform Deployment Based on
Clustering (NNDBC) algorithm. NNDBC improves network connectivity by determining the heterogeneous communication range instead of homogeneous communication range of sensor nodes.
Moreover the aggregate contribution degree concept is
defined and lower contribution degree sensor nodes are used to substitute the dying sensor nodes. In this way, the total movement distance of sensor nodes is decreased and energy consumption of sensor nodes is reduced and maximum network lifetime is achieved. Cao, Jiabao, [27].
et al.
proposed Balance Transmission Mechanism (BTM) in UWSNs
Generally, sensors are deployed at high cost with limited energy.
Due to the
path loss, transmission loss and link impairments in UWSNs, balanced transmission is not an easy task.
In BTM, a balance transmission mechanism is presented by
dividing the data communication phase into two phases. In phase-I, a routing path is setup using an efficient routing algorithm which is based on optimum transmission distance for energy optimization in UWSNs. While, in phase-II a data balance transmission algorithm is presented for stable data transmission. BTM increases the energy efficiency with balance data transmission and increases network lifetime.
14
2.3. Related Work
Chapter 2 Table 2.1: Comparison of UWSNs Routing Protocols
Protocols DBR
EEDBR
Technique
Multi-hop routing
Forwarders
Mobile Sink(s)
Parameters
Parameters
metrics
AUV(s)
achieved
compromised
Depth
5
Low energy consumption End-to-end
Multi-hop routing
Maximum throughput
5
Residual energy
delay minimizes Maximum throughput
Stability period
High energy consumption Stability period
Depth Multi-hop WDFAD-DBR
iAMCTD
routing
Depth
5
Multi-hop routing
4
Linkβs SNR
High energy efficiency Stability period increases
High energy efficiency Stability period increases
Throughput decreases
Minimum throughput
Residual energy Depth
DEADS
Data reliability
Cooperative routing
4
Residual energy
Maximum throughput Less packet drop
High energy consumption Stability period decreases
Depth
BTM
NNDBC
Multi-hop routing
Clustering
4
Residual energy
Depth
5
Balanced energy consumption
High connectivity rate High coverage rate
Minimum throughput
High energy consumption
High energy efficiency AEDG
DDG
Multi-hop routing
4
Residual energy
High network throughput Low transmission loss
Maximum end-to-end delay
Improves delay
Multi-hop routing
High stability period
4
Residual energy
Maximum throughput Low energy consumption
15
Larger overhead
Chapter 2 Table 2.2: Comparison of WSNs Clustering Protocols
Protocols
DR
DDR
CH selection
Parameters
Parameters
Technique
parameters
achieved
compromised
Energy hole minimization
Dynamic
with field division
Centralized
Maximum stability period Maximum throughput
Reducing coverage
High energy efficiency
hole problem
Improved connectivity
with field division
Sequential
High packet drop
Low energy consumption
Maximum throughput
Data reliability High packet drop
Minimum distance based
LEACH
ACH2
GAECH
Reducing global
Probabilistic
communication in WSNs
Sequential
Maximize throughput and lifetime of WSNs Balanced clustering in terrestrial WSNs
in smart space and
Residual energy
Solving Nondeterministic Polynomial (NP) hard and finite problem.
consumption Short stability period
Maximum throughput
Transmission delay
Stability period Fitness function
Node degree
extreme environments
ABCACO
Maximum throughput
Energy efficiency
Energy-aware clustering SEECH
Maximum energy
Residual energy and Distance from BS
Energy efficiency
Maximum delay
Stability extends
Minimum throughput
Stability extends High energy efficiency Less energy consumption Maximum throughput
Minimum throughput
Maximum end-to-end
Stability period
2.4 SEEC: The Proposed Scheme We describe energy efficiency as maximum work done by consuming minimum energy. Specifically, in UWSNs energy efficiency means maximum network lifetime with minimum energy consumption of the UWSN sensor nodes.
16
delay
2.4.1. Network Model
Chapter 2
2.4.1 Network Model Initially, we divide the network field into 10 regions of same size. Then, sensor nodes are deployed randomly in the underwater network field. The position of sensor nodes is dynamic, due to underwater environment.
A static sink is deployed at the top
center of the network field and 2 mobile sinks are moving inside the network field. Mobile Sink 1 (MS1) changes its position after every round and move within the sparse regions of the network. While, Mobile Sink 2 (MS2) remains in a single sparse region, until the death of all sensor nodes of that region. The network model of SEEC is shown in fig. 2-1.
R5
R 10
R4
R9
100m R 3
R8
R2
R7
R1
MS2
R6
100m
Static Sink Normal Node
Dead Node
Cluster Head
Mobile Sink
Figure 2-1: Network Model of SEEC
2.4.2 Mathematical Model In SEEC, we divide the network field into 10 equal size subregions to find sparse and dense regions of the network. We represent regions of the left side with regions of the right side with
π
π
.
π
πΏ
and
The coordinates of origin point are taken as ref-
erence point for the formation of regions. Coordinates of origin are represented with
π(π₯0 , π¦0 ).
The following equations divide the network field into the left side 5 regions:
17
Chapter 2
π(π₯0 , π¦0 ) = (0, 0) π
πΏ1 (π₯1 , π¦1 ) = (π₯0 + π, π¦0 + π΅),
(2.1)
π
πΏ2 (π₯2 , π¦2 ) = (π₯0 + π, π¦0 + 2π΅),
(2.2)
π
πΏ3 (π₯3 , π¦3 ) = (π₯0 + π, π¦0 + 3π΅),
(2.3)
π
πΏ4 (π₯4 , π¦4 ) = (π₯0 + π, π¦0 + 4π΅),
(2.4)
π
πΏ5 (π₯5 , π¦5 ) = (π₯0 + π, π¦0 + 5π΅),
(2.5)
The following equations divide the network field into the right side 5 regions:
Where,
π
side regions from total
π
π
1 (π₯1 , π¦1 ) = (π₯0 + 2π, π¦0 + π΅),
(2.6)
π
π
2 (π₯2 , π¦2 ) = (π₯0 + 2π, π¦0 + 2π΅),
(2.7)
π
π
3 (π₯3 , π¦3 ) = (π₯0 + 2π, π¦0 + 3π΅),
(2.8)
π
π
4 (π₯4 , π¦4 ) = (π₯0 + 2π, π¦0 + 4π΅),
(2.9)
π
π
5 (π₯5 , π¦5 ) = (π₯0 + 2π, π¦0 + 5π΅),
(2.10)
is the x-dimension point of a region.
π
πΏ
and
π€πππ‘β
π
for the left side regions
π
π
.
of the network field. The value of
18
π
π
is multiple of 2 for the right
The value of
π
is calculated
is calculated in eq. (2.11).
2.4.2. Mathematical Model
Chapter 2
π = π€πππ‘β/2, In eq. (2.12), we calculate the value of
π΅
π΅ from breadth of the network field.
is the y-dimension point of a region. For
of 2 for
π
πΏ2
and
π
π
2 ,
and multiple of 5 for
multiple of 3 for
π
πΏ5
and
(2.11)
π
πΏ1
π
πΏ3
and
and
π
π
1
π
π
3 ,
its value is
π΅. π΅
multiple of 4 for
Where,
is multiple
π
πΏ4
and
π
π
4
π
π
5 .
π΅ = ππππππ‘β/5,
(2.12)
π
πΏ1π‘ππΏπ = {π΅, 2π΅, 3π΅, 4π΅, 5π΅, ...ππ΅},
(2.13)
π
π
1π‘ππ
π = {π΅, 2π΅, 3π΅, 4π΅, 5π΅, ...ππ΅},
(2.14)
From eq. (2.12), eq. (2.13) and eq. (2.14), we establish a relation to find point for
π
πΏπ
and
π
π
π
as:
π΅ = ππ΅, Where,
ππ‘β
π΅
(2.15)
π is the total number of left side or right side regions.
number of the top left region
π
πΏπ
and top right region
π
π
π
The coordinates for
can be calculated as:
π
πΏπ (π₯π , π¦π ) = (π₯0 + π, π¦0 + ππ΅),
(2.16)
π
π
π (π₯π , π¦π ) = (π₯0 + 2π, π¦0 + ππ΅),
(2.17)
The network field division into the left side and right side regions is shown diagrammatically in fig. 2-2:
19
Chapter 2 W
W RRn (xn ,yn) = (2W,nB)
RLn(xn ,yn) = (W,nB) B
R Ln
R Rn (W,3B)
B
R L3
B
R L2
B
R L1
O (x0 ,y0) = (0,0)
(W,2B) (W,B)
(2W,3B)
R R3 R R2
(2W,2B) (2W,B)
R R1
R L- Regions
R R - Regions
Figure 2-2: Division of Network Field into Left and Right
n
Regions.
We divide the SEEC routing scheme in 5 phases. In phase-I, we discuss network configuration in underwater environment then, phase-II is sparse and dense regions searching phase. In phase-III, we discuss clustering mechanism in dense regions of the network field and we describe CH selection mechanism in subsection. Sinks mobility in sparse regions is discussed in phase-IV. In the last phase-V, we describe data transmission mechanism in SEEC routing protocol.
2.4.3 Network Configuration We divide the SEEC operation into rounds. Initially, all the sensor nodes are deployed randomly in the network field. All the sensor nodes are equipped in such a way that depth of each sensor node is known with the depth finding module installed in it. A static sink is deployed at the top of the water surface. The other two mobile sinks: MS1 and MS2, are deployed in the sparse regions of the network after sparse and dense regions are searched. All the sensor nodes are supplied with same energy.
2.4.4 Searching Sparse and Dense Regions We divide regions of the network field into subregions of two categories: sparse regions; the regions with minimum number of sensor nodes and dense regions; the regions with maximum number of sensor nodes. In SEEC, first we find coordinates of a sensor node, in order to find region of that sensor node.
Then, number of nodes is searched in
each region of the network field to find sparsity or density of that region as shown in
20
2.4.5. Clustering in Dense Region
Chapter 2
algorithm 1 and algorithm 2. If the number of nodes in that region is minimum, then the region is sparse and if the number of nodes in that region is maximum, then the region is dense. Sparse and dense regions search mechanism is shown in fig. 2-3.
2.4.5 Clustering in Dense Region Once sparse and dense regions are searched, the next phase is clustering of nodes in dense regions.
We used, clustering technique for top 4 most dense regions, to
increase energy efficiency and network lifetime.
In SEEC, nodes of a dense region
collaboratively select a node as a CH and send data to the selected CH. The CH performs data aggregation and sends the compressed data to one of the nearest sink. In SEEC, the CH selection is on the basis of depth and residual energy.
2.4.5.1 CH Selection CH selection in SEEC is different from other protocols. In SEEC, a node with low depth and high residual energy is selected as a CH. The selection criterion of selecting a node as a CH in a region is based on the following conditions:
πΆπ»π < π
π
(2.18)
πΈπΆπ» > πΈππ£π
(2.19)
π€βπππ, πΈππ£π =
π ππ‘πππ
ππ πππ’πππΈπππππ¦ π π’ππππππ π΄πππ£ππ ππππ
ππππ β€ π β π€βπππ, π β(π) =
Where
π
π
(2.20)
π 1 β π(πππ(π, π1 ))
is depth of nodes of a region. First of all, for the selection of CH, eq. (2.18)
is checked. If depth of the node is higher than the other nodes of that region, then the node cannot be a CH in the current round and if depth of the node is lower than the other nodes of that region then eq. (2.19) is checked, where
21
πΈπΆπ»
is the residual
Chapter 2
Divide network field into regions
Start
Find sparse and dense regions with SSA and DSA algorithms No
if region is sparse
if region is dense
No
Yes
Yes
for r = 1 to Rsparse
for r = 1 to Rdense
for n = 1 to rsparse
for n = 1 to rdense
Is Node n alive?
Is Node n alive?
No
Yes
Yes
Is sink in range?
Yes
Check EPOCH No Yes
No
for n = 1 to rsparse with lowest depth highest residual energy
No
Lowest depth Highest Re
if found
No
Yes
Node n is selected as CH
Yes No
Nodes finshed
No
Yes
Marked as normal node N
if sink in range
Send data to selected CH
No
Yes
for c = 1 to CHs with lowest depth highest residual energy
Send data to sink
Yes
End
If CH found
No
Figure 2-3: Working Flow of SEEC
22
2.4.6. Sink Mobility in Sparse Region energy of a CH and
πΈππ£π
Chapter 2
is the average residual energy of an individual node. The
node is selected as a CH, if residual energy of the node is greater than the average residual energy calculated for an individual node. In eq. (2.20), p is the probability of a node to be a CH and it is calculated at the start of each round and r represents the current round. The node is selected as CH if it has not been CH for the last 1/p rounds. with
The node generates a random number rand, the compares the rand value
π β(π) given in [7].
The node is considered as a normal node, if rand of a node is
greater than the threshold value. If rand of the node is less than
πβ
value, then the
node is finally a CH node. In SEEC, only one node is selected as a CH in each dense region per round. We illustrate, an example of sensor node selection as a CH. In fig. 2-4(A), we have two nodes N1 and N2 with different energy levels and at the same depth. The node with high residual energy is selected as a CH i.e. N2 because . While in fig. 2-4(B), the node N1 is selected as a CH because both nodes have same energy level but they are at different depth. Thus low depth node is selected as a CH.
Energy = 3 Depth = 25
Energy = 5 Depth = 25
Energy = 2 Depth = 40
Energy = 2 Depth = 80
N1
N2 CH
N1 CH
N2 B
A
Figure 2-4: Scenario of CH Selection in SEEC
2.4.6 Sink Mobility in Sparse Region The regions with minimum number of nodes are called sparse regions.
In SEEC,
sparse regions are searched and two mobile sinks are sent in those sparse regions for data collection. The sinks mobility is in such a way that, MS1 changes its position
23
Chapter 2 in each round from topmost sparse to least sparse region except the region of MS2. While, MS2 remains in the topmost sparest region until the death of all the sensor nodes of that region. If all sensor nodes died in the region of MS2, then MS2 changes its position to the topmost sparest region of the remaining sparse regions. While, MS1 changes position accordingly. Mobile sink always takes midpoint position in a region because midpoint of a region is within the transmission range of maximum sensor nodes of that region.
Sparsity awareness is the most suitable approach to collect
data from maximum sensor nodes of the network, to achieve maximum throughput with minimum energy consumption of the entire network. Also, sparsity awareness approach is also helpful to find dense regions of the network field, where the sensor nodes form clusters.
Instead all the sensor nodes send data to sink through multi
hopping, nodes send data to the respective CH of their region. Thus only CH send data to the sink. This makes SEEC an energy efficient routing protocol.
Algorithm 1 SSA 1: procedure Sparsity Search 2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20:
π β Number of nodes π
β Number of regions for each node π β π do
Find coordinates of node n
for each region π β π
do if π β π then Nodes(π ) =
end if
π
π
(π) β π
end for end for
π
= Descending sort(π
) π ππππ π={π |π β π
β§ π > π
/2} for each region π β π
do if π = π ππππ π then sparse regions= π
end if end for end procedure
24
2.4.7. Data Transmission
Chapter 2
Algorithm 2 DSA 1: procedure Density Search 2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20:
π β Number of nodes π
β Number of regions for each node π β π do
Find coordinates of node n
for each region π β π
do if π β π then Nodes(π ) =
end if
π
π
(π) β π
end for end for
π
= Descending sort(π
) ππππ π={π |π β π
β§ π β€ π
/2} for each region π β π
do if π = ππππ π then dense regions= π
end if end for end procedure
2.4.7 Data Transmission In SEEC, the sensor nodes in sparse regions of the network send data directly to mobile sink deployed at the centre of that region. Each of the mobile sink changes its position from one sparse region to other periodically in order to collect data from all of the sparse regions of the network. Data transmission in dense regions employs in 3-tiers. The sensor nodes in the dense regions gather data and send this data to CH in tier-I. In tier-II, the CH sends the collected compressed data to the respective CH having low depth and maximum residual energy. In tier-III, the respective CH sends data to static sink at the top or any of the MS within its transmission range.
2.5 Performance Evaluation For performance evaluation, we compare our routing protocol with two depth base routing protocols of UWSNs, DBR and EEDBR. In order to get fair results, we run
25
Chapter 2 2000 rounds simulation with same parameters.
All the sensor nodes are randomly
deployed in the network field of 100mx100m, the number of nodes was 100, the initial residual energy of each sensor nodes was set 5-joules with the transmission range of 30 meters and the pre-defined depth threshold of each sensor node was set 25 meters. The results of our protocol are compared with other protocol on the basis of metrics i.e.
network lifetime, network residual energy (j), packets received/packets
sent, packets received at sink, stability period and instability period.
2.5.1 Network Lifetime The network lifetime of our proposed routing protocol is better than the DBR and EEDBR routing protocols, because in DBR low depth while, in EEDBR higher residual energy nodes are selected for data forwarding and all the sensor nodes die before sensor nodes death in SEEC, as shown in fig. 2-5. In our proposed routing protocol network lifetime increases because in most dense regions we introduced cluster head (CH) nodes and in sparest regions 2 mobile sinks are collecting data. In our proposed routing protocol, instead of the nodes with higher residual energy are selected for data forwarding to sink, the CH in each dense region collects data from all the sensor nodes in that region and forwards aggregated data to sink.
100 SEEC DBR EEDBR
90 80
Alive Nodes
70 60 50 40 30 20 10 0
0
500
1000 Rounds
1500
Figure 2-5: Lifetime of Network
26
2000
2.5.2. Stability Period and Instability Period
Chapter 2
2.5.2 Stability Period and Instability Period In our proposed protocol, better stability period is achieved. In EEDBR, due to high load on high residual energy and low depth nodes, the stability period quickly ends as shown in fig. 2-6. Better stability period is achieved by SEEC, due to clustering. In clustering technique, instead of all the node communicate with the sink directly, only cluster heads send aggregated data to the sink. The number of dead nodes in our proposed routing protocol is less than the number of dead nodes in DBR and EEDBR as low depth and high residual energy nodes die frequently. The instability period of EEDBR is sharp, due to quick energy consumption. Instability period of our proposed routing protocol is shown in fig. 2-6.
100 90 80
Dead Nodes
70 60 50 40 30 20 SEEC DBR EEDBR
10 0
0
500
1000 Rounds
1500
2000
Figure 2-6: Stability and Instability Period of Network
2.5.3 Network Residual Energy In EEDBR, the nodes with high residual energy and low depth are selected for forwarding data to sink, which causes much energy consumption of the network while, in DBR low depth nodes are selected as data forwarders. We introduced 2 mobile sinks in sparse regions, while in dense regions clustering is performed which causes less consumption of residual energy of all the sensor nodes thus average network residual
27
Chapter 2 energy is not as much consumed in our routing protocol as in EEDBR. Total network residual energy consumption plot by SEEC, DBR and EEDBR is shown in fig. 2-7.
2.5.4 Packets Sent and Packets Received In EEDBR, ratio of packets sent in each round is high than DBR and SEEC, due to high residual energy and low depth sensor nodes are selected for data forwarders at the cost of higher residual energy consumption as shown in fig. 2-8.
While, in
our proposed routing protocol, the ratio of packets sent in each round is less because cluster heads in each dense region send composite packet to sink(s) instead of multiple packets in the network. In DBR and EEDBR, the total number of packets sent to sink decreases as network residual energy gets lower as shown in fig. 2-9. While, in our proposed routing protocol total packets sent to sink increases uniformly with the passage of time because high network residual energy.
500 SEEC DBR EEDBR
450
Residual energy(joules)
400 350 300 250 200 150 100 50 0
0
500
1000 Rounds
1500
2000
Figure 2-7: Total Network Residual Energy
2.6 Performance Tradeoffs Trade-offs made by SEEC are given in table 2.3. SEEC achieves stability period and better energy consumption than DBR and EEDBR at the cost of low throughput.
28
2.6. Performance Tradeoffs
Chapter 2
4
5
x 10
4.5 SEEC DBR EEDBR
Packets recieved at sink
4 3.5 3 2.5 2 1.5 1 0.5 0
0
500
1000 Rounds
1500
2000
Figure 2-8: Total Packets Sent to Sink
50 SEEC DBR EEDBR
Packets Recieved/Packets Sent
45 40 35 30 25 20 15 10 5 0
0
200
400
600
800 Rounds
1000
1200
1400
1600
Figure 2-9: Total Packets Sent or Received at Sink Per Round
While, throughput of DBR and EEDBR is high as compare to SEEC because of multi-hop routing mechanism. Stability period of DBR and EEDBR is compromised due to the unbalanced load on low depth and high residual energy nodes. Balanced energy consumption is achieved in SEEC through static clustering and division of the network field in subregions.
29
Chapter 2 Table 2.3: Performance Tradeoffs Protocols
Technique
Parameters
Cost to pay
achieved DBR
Multi-hop
rout-
ing
Low energy con-
Maximum
sumption, Max-
to-end delay
imum
end-
through-
put EEDBR
Multi-hop
rout-
ing
High
energy
efficiency, imum
Stability period
Max-
through-
put SEEC(proposed)
Clustering
with
sink mobility
Low
energy
Low throughput
consumption, stability period
2.7 Conclusion of the Chapter Underwater wireless sensor nodes change position dynamically with the water current due to which some areas of the network field become sparse.
Most commonly due
to the high manufacturing cost, high design cost and high deployment cost sensor nodes are sparsely deployed. In order to achieve energy efficiency, load balancing and to control routing hole problem in sparse regions, we presented SEEC. Results show that SEEC achieves better network lifetime and uniform load distribution through static clustering in dense regions of the network. The division of the network field into subregions also control routing hole problem. The simulation results show that network stability, lifetime and energy consumption of SEEC is better than other UWSNs routing protocols.
30
References [1] Yan H, Shi ZJ, Cui JH.
DBR: depth-based routing for underwater sensor networks.
In NETWORKING 2008 Ad Hoc and Sensor Networks, Wireless Networks, Next Generation Internet 2008.
An energy efficient localization-free routing protocol for underwater wireless sensor networks. International journal of distributed sensor net-
[2] Wahid A, Kim D. works, 2012.
WDFAD-DBR: Weighting depth and forwarding area division DBR routing protocol for UASNs. Ad Hoc Networks.
[3] Yu H, Yao N, Wang T, Li G, Gao Z, Tan G. 2016.
Energy Consumption Model for Density Controlled Divide-and-Rule Scheme for Energy Efficient Routing in Wireless Sensor Networks. International Journal of Adhoc and Ubiquitous Com-
[4] Latif K, Javaid N, Najmus M, Saqib ZA, Alrajeh N.
puting, 2015.
Boundary Detection Method for Large-Scale Coverage Holes in Wireless Sensor Network Based on Minimum Critical Threshold Constraint. Journal of Sensors, 2014.
[5] Jing R, Kong L, Kong L.
Energy Hole Minimization with Field Division for Energy Efficient Routing in WSNs.
[6] Latif K, Javaid N, Saqib MN, Khan ZA, Qasim U, Mahmood B, Ilahi M. International Journal of Distributed Sensor Networks, 2015. [7] Heinzelman, Balakrishnan.
sor networks.
Wendi
Rabiner,
Anantha
Chandrakasan,
and
Hari
Energy-efficient communication protocol for wireless microsen-
System sciences, 2000. Proceedings of the 33rd annual Hawaii
international conference on. IEEE, 2000.
TEEN: A routing protocol for enhanced efficiency in wireless sensor networks. In Proc. 15th IPDPS, 2001.
[8] A. Manjeshwar and D. P. Agrawal.
SEP: A stable election protocol for clustered heterogeneous wireless sensor networks Dept.Comput. Sci., Boston
[9] G. Smaragdakis, I. Matta, and A. Bestavros.
Univ., Boston, MA, USA, Tech. Rep. BUCS-TR-2004-022, 2004.
Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor networks In Comput.Commun., 2006.
[10] L. Qing, Q. Zhu, and M. Wang.
31
Chapter 2 ACH2: Routing Scheme to Maximize Lifetime and Throughput of Wireless Sensor Networks. Sensors Jour-
[11] Ahmad A, Javaid N, Khan ZA, Qasim U, Alghamdi TA. nal, IEEE, 2014.
Load-balanced clustering algorithm with distributed selforganization for wireless sensor networks. Sensors Journal, IEEE, 2013.
[12] Liao Y, Qi H, Li W.
GAECH: Genetic Algorithm Based Energy Efficient Clustering Hierarchy in Wireless Sensor Networks. Journal of Sensors. 2015.
[13] Baranidharan B, Santhi B.
SEECH: Scalable energy efficient clustering hierarchy protocol in wireless sensor networks. Sensors Journal, IEEE. 2014.
[14] Tarhani M, Kavian YS, Siavoshi S.
A distributed energy-aware routing protocol for underwater wireless sensor networks. Wireless Pers. Commun, 2011.
[15] Mari Carmen Domingo,
Distributed minimum-cost clustering protocol for underwater sensor networks (UWSNs) IEEE International Conference on Com-
[16] Pu Wang, Cheng Li, Jun Zheng.
munications,ICCβ07, IEEE, 2007.
A locationbased clustering algorithm for data gathering in 3D underwater wireless sensor networks
[17] K.R. Anupama, Aparna Sasidharan, Supriya Vadlamani,
International Symposium on Telecommunications, IST 2008, IEEE, 2008.
LUM-HEED: a location unaware, multi-hop routing protocol for underwater acoustic sensor networks International Conference on
[18] Ce Wang, Gongliang Liu.
Computer Science and Network Technology (ICCSNT), vol. 4, IEEE, 2011.
A new multi-path routing protocol based on cluster for underwater acoustic sensor networks International Conference on Multi-
[19] Guangzhong Liu, Changye Wei.
media Technology (ICMT), IEEE, 2011.
Hybrid Swarm Intelligence Energy Efficient Clustered Routing Algorithm for Wireless Sensor Networks. Journal of Sensors,
[20] Rajeev Kumar and Dilip Kumar. 2016.
DEADS: Depth and Energy Aware Dominating Set Based Algorithm for Cooperative Routing along with Sink Mobility in Underwater WSNs. Sensors, 2015.
[21] Umar A, Javaid N, Ahmad A, Khan ZA, Qasim U, Alrajeh N, Hayat A.
A network lifetime enhancement method for sink relocation and its analysis in wireless sensor networks Sensors Journal,
[22] Wang CF, Shih JD, Pan BH, Wu TY. IEEE. 2014.
[23] Javaid, N., Jafri, M.R., Khan, Z.A., Qasim, U., Alghamdi, T.A. and Ali,
iAMCTD: Improved adaptive mobility of courier nodes in threshold-optimized dbr protocol for underwater wireless sensor networks. International Journal of DisM.
tributed Sensor Networks, 2014.
32
REFERENCES
Chapter 2
[24] Javaid N, Ilyas N, Ahmad A, Alrajeh N, Qasim U, Khan ZA, Liaqat T, Khan MI.
An Efficient Data-Gathering Routing Protocol for Underwater Wireless Sensor Networks. Sensors, 2015. A Distributed Data-Gathering Protocol Using AUV in Underwater Sensor Networks. Sensors, 2015.
[25] Khan JU, Cho HS.
Node Non-Uniform Deployment Based on Clustering Algorithm for Underwater Sensor Networks. Sensors, 2015.
[26] Jiang P, Liu J, Wu F.
Balance Transmission Mechanism in Underwater Acoustic Sensor Networks. International Journal of Distributed Sensor Networks. 2015.
[27] Cao J, Dou J, Dong S.
33
Chapter 3 EMT: Energy Efficient Routing Protocol to Maximize Network Lifetime in UWSNs
34
3.1. Summary of the Chapter
Chapter 3
3.1 Summary of the Chapter Energy efficiency is one of the key requirement in UWSNs to prolong the lifetime. In this chapter, we present an EMT to improve network lifetime of UWSNs, designed for continuous monitoring applications.
EMT, minimizes load on randomly
deployed sensor nodes in a network by selecting a forwarder node on the basis of minimum distance from sink which locally gather data from all nodes of sector and sends received plus generated data to the sink. In EMT, each sensor node reports some fraction of data to selected forwarder node and some fraction of data directly to sink in order to prolong network lifetime. EMT is implemented in homogeneous and heterogeneous environments and compared with existing techniques for performance evaluation. Simulation results show that it performs better in terms of network lifetime and energy consumption.
3.2 Motivation In Balanced Routing (BR) scheme [1], nodes forward data using two transmission ranges; one-hop and two-hop. Sensor nodes which are one-hop neighbors to sink are relieved of relaying data because two-hop neighbor nodes of sink directly transmit data to sink. However, due to maximum data received at two-hop neighbor to sink, these nodes consume maximum energy in transmitting data at two-hop and an energy hole is created. In addition, each node in BR scheme, forwards maximum data using two-hop transmission range and consumes maximum energy which minimizes network lifetime. We propose EMT to minimize energy consumption of nodes during data transmission and to prolong network lifetime. In EMT, nodes with adjustable transmission power level are used. Sensor nodes adjust transmission power level according to the distance of forwarder node and sink from the sender. First, distance between sink and nodes of a sector is calculated to find an eligible forwarder node. Then, a node with minimum calculated distance is selected as a forwarder in a sector. In each sector there is only one sensor node selected as a forwarder. Rest of the nodes in each sector forwards
35
Chapter 3 some fraction of data to forwarder node and some fraction of data directly to sink for maximum network lifetime. The forwarder node send all data received plus generated directly to sink. After each transmission round an eligible forwarder is selected in a sector. EMT minimizes energy consumption of nodes by locally forwarding data to a forwarder node.
3.3 Related Work Energy efficiency is one of the important key requirement in UWSNs.
Many re-
searchers work to minimize energy conservation in UWSNs. Summary of some literature is given in table 3.1. Some related work is presented in this section. Cao
et al.
in [2], present a Balanced Transmission Mechanism (BTM) technique.
In BTM, energy of sensor nodes is divided into levels and each node transmits data according to that energy level. If energy level of a node decreases excessively during multi-hop transmission, sensor node forwards data directly to sink.
In BTM, due
to long distance transmission most of the nodes drain-out energy quickly.
Energy
consumption in network is unbalanced in BTM due to long distance transmission. In [3], an Enhanced Efficient and Balanced Energy consumption Technique (EEBET) is presented to overcome the flaws in BTM. A depth threshold is defined to limit the nodes in order to minimize nodes towards sink and to avoid backward transmission. EEBET calculates optimum energy levels and restrict direct transmission at long distance by presenting Efficient and Balanced Energy consumption Technique (EBET) to save energy and prolong network lifetime. Kamran
et al.
present, Spherical Hole Repair Technique (SHORT) in [4] to avoid
coverage hole problem in UWSNs. SHORT is a coverage hole repair technique which first, locates coverage hole in a network and repair it. Authors, use sensor nodes with adjustable transmission power level. SHORT is not suitable for time-critical applications because of coverage hole repair time. Coverage hole problem in UWSNs is also addressed by Peng
et al.
in [5]. They propose two algorithms; Guaranteed Full
Connectivity Node Deployment (GFCND) and Location Dispatch Based on Com-
36
3.3. Related Work
Chapter 3
mand Nodes (LDBCN) algorithm. In GFCND, nodes are logically considered of two types. Command nodes are used to adjust location of common nodes in network using LDBCN algorithm. While, connectivity nodes are used for maximum connectivity. In [6], authors propose a Differential Initial Battery (DIB) assignment technique and Energy Balanced Hybrid (EBH) algorithm for UWSNs. EBH algorithm is presented for sparse linear networks to balance the energy consumption and to prolong network lifetime. These technique are not efficient in dense and non-linear networks. A Relative Distance Based Forwarding (RDBF) protocol for energy efficient routing is presented in [7] for UWSNs. A forwarder node is selected on the basis of fitness function and only these forwarder nodes take part in routing of data. On the basis of two factors, forwarders are selected; minimum distance from sink and minimum hop counts of routing path. RDBF is designed specially for time-critical applications and it achieves low end-to-end delay and energy efficiency. In [8], authors present a Hop-by-Hop Dynamic Addressing Based (H2-DAB) routing protocol to control the problem of node mobility in UWSNs.
Each node in a
network is assigned a routing address without need of any information of dimensional location and new nodes are added and configured easily. In H2-DAB, delivery ratios are independent of density or sparsity of network. Authors in [9], present a Power Efficient Routing (PER) protocol, which works in two phases; a forwarder node selection phase and a forwarding tree trimming phase. In forwarder node selection phase, two nodes are selected as forwarders. Selection is based on three parameters; residual energy of node, distance and angle between two neighbors. In order to save extra power consumption due to unnecessary forwarding of data packets, the tree trimming mechanism is used. PER is designed for UWSNs implemented for time-critical and monitoring applications in which power efficiency is one of the key requirement. In [10], authors propose Energy-efficient Routing Protocol (ERP2R) which is based on physical distance and residual energy.
A group of forwarder nodes is se-
lected which are at minimum distance from sink to minimize the packet forwarding of node for energy conservation. Residual energy of nodes is checked to forward data
37
Chapter 3 to prolong network lifetime.
Wahid
et al.
improve ERP2R and present a reliable
and energy-efficient routing protocol (R-ERP2R) in [11].
R-ERP2R selects a for-
warder node on physical distance and residual energy and check also link quality of forwarder node to ensure reliability of data. R-ERP2R improves energy consumption and network lifetime of UWSNs. In [12], authors present a Ring-Based Correlation Data Routing (RBCDR) scheme, which performs data correlation to aggregate data and minimizes the transmitting data amount. The generated data is routed to the next hop ring which has abundant energy then, data aggregation is performed along the ring and the aggregated data is sent to the sink through a shortest path. RBCDR scheme performs data aggregation in regions having abundant energy and forwards less data to the sink to achieve energy conservation near sink for maximum network lifetime. In [13], Layer-by-Layer Angle-Based Flooding (L2-ABF) routing protocol is presented for UWSNs. L2-ABF is localization-free routing protocol in which nodes are deployed in layers.
Each sensor node stores its calculated depth and a forwarding
angle. Data is forwarded by each node according to the forwarding angle. A node which lies in angle-based zone of sender only receives data due to which less energy is consumed and packet delivery ratio is increased. Zhang
et al.
present a localization technique for UWSNs which is based on Mo-
bility Prediction and Particle Swarm Optimization (MP-PSO) algorithm [14].
In
proposed technique, location of unknown nodes is found using beacon nodes and velocity information of each beacon node is calculated.
Velocities of unknown nodes
are estimated from the velocity of beacon nodes. The location of an unknown node is predicted with the help of mobility prediction technique. A Hierarchical Multi-path Routing-LEACH (HMR-LEACH) algorithm for UWSNs is presented in [15] which is based on LEACH algorithm. In HMR-LEACH, a multipath routing algorithm is used instead of one hop routing algorithm. First, energy and physical distance is checked for selecting the transmission path. Then, a weight to each transmission path is assigned during data transmission to the sink. The selection probability of each transmission path is different from other. Network lifetime
38
3.3. Related Work
Chapter 3
and energy consumption is balanced using HMR-LEACH technique. In [16], Yan
et al.
present a localization-free DBR protocol. In proposed protocol,
each node forwards data using greedy approach. involved in data forwarding.
Nodes with minimum depth are
During data forwarding each sender node sends its
depth along with data. In dense networks DBR works well while, DBR performance is not better in sparse networks in terms of packet delivery ratio. Deficiency in DBR is that due to greedy approach and selection of low depth, nodes near sink quickly drain-out energy. Therefore, stability of DBR is minimum. In EEDBR [17], along with depth of a node its residual energy is also used for its selection as an eligible forwarder. Therefore, energy consumption in EEDBR is better than DBR. Redundant data packets are restricted in EEDBR, as nodes with low depth and maximum residual energy are selected to forward data. Throughput of DBR is maximum than EEDBR because nodes find multiple paths to reach sink in DBR using greedy approach. A Weighting Depth and Forwarding Area Division DBR (WDFAD-DBR) scheme is discussed in [18]. In WDFAD-DBR, void holes are detected in advance and sender node does not forward data to node without neighbor nodes to improve communication reliability.
In dense regions of the network energy consumption is reduced
through dividing forwarding regions. The division of forwarding regions is according to channel condition and density of nodes. In [19], authors present an Improved Adaptive Mobility of Courier nodes in Thresholdoptimized DBR (iAMCTD). It is designed for reactive UWSNs for prolonged network lifetime. Due to variable depth-threshold and optimized sink mobility it reduces endto-end delay. Therefore, it is suitable for time-critical applications. iAMCTD avoids unnecessary data transmissions due to which throughput is reduced. Amara scheme.
et al.
discuss Depth and Energy Aware Dominating Set (DEADS) [20]
DEADS is designed for reactive UWSNs and it is not suitable for delay
sensitive applications. Due to cooperation and sink mobility, it achieves maximum lifespan and packet delivery ratio. Energy consumption is high in DEADS because each node forwards data through multiple paths. Stability of network is compromised to achieve reliability of data.
39
Chapter 3 Table 3.1: State-of-the-art Related Work
Protocols
Features
Deployment
Achievements
BTM [2]
Mixed transmission and energy balancing
Uniform
Achieves energy efficiency and network lifetime due to balanced transmission in network
EEBET [3]
Balanced transmission
Random uniform
SHORT [4]
Coverage hole repairing Random technique
GFCND [5]
Guaranteed full connectiv- Random ity node deployment technique
EBH [6]
RDBF [7]
Energy balancing with hybrid data transmission
Uniform
Relative distance based Uniform forwarding technique
H2-DAB [8]
Hop-by-hop dynamic addressing technique
Uniform
PER [9]
Power efficient routing
Random
R-ERP2R [11]
RBCDR [12] DBR [16]
EEDBR [17]
Localization free routing Random protocol with route selection on the basis of physical distance, link quality and residual energy Correlation data Uniform aggregation routing Multi-hop routing using Random depth metric for forwarder node selection
Consumes high energy when transmitting at long distance, Formation of transmission loops Due to balanced transmis- Performance decreases sion it achieves minimum with energy consumption, max- increase of network radius imum network lifetime and throughput Network lifetime and High end-to-end delay due throughput increased due to hole repairing to maximum network coverage and connectivity High network coverage and With random node scatconnectivity tering their is no relationship of network topology Minimum energy con- Not efficient for dense as sumption and maximum well as non-linear networks network lifetime due to hybrid transmission Low end-to-end delay Nodes with less distance and energy efficiency is and minimum hop counts achieved by involving from sink have unbalanced less number of nodes in load forwarding Minimum routing over- High end-to-end delay and head due to assigning energy consumption due to dynamic addresses to request and replay inquiry nodes and high throughput Energy efficiency, High Maximum energy conpacket delivery ratio sumption when number of nodes increases Increased network lifetime Control packet overhead and packet delivery ration, for calculating physical End-to-end delay reduced distance Less energy consumption and maximum lifetime Maximum throughput due to forwarding data through multiple available paths End-to-end delay minimizes and maximum throughput
Multi-hop routing using Random residual energy and depth parameters for forwarder node selection
40
Deficiencies
Delay due to data aggregation along rings Energy consumption and stability are compromised due to redundant data forwarding High energy consumption and stability period in dense networks and maximum end-to-end delay due to holding time
3.4. EMT: The Proposed Scheme
Chapter 3
Table 3.1: State-of-the-art Related Work
Protocols
Features
Deployment
WDFAD [18]
Multi-hop routing with Random depth as routing metric
iAMCTD [19]
Multi-hop routing
DEADS [20]
Cooperative routing with Random residual energy and depth as forwarder selection metrics
Achievements
Deficiencies
Low energy consumption and stability is achieved due to avoiding void holes Low energy consumption and maximum stability
Packet delivery ration decreases as selected nodes only forward data Packet delivery ratio is decreased due to restricted data transmission Data reliability, maximum High energy consumption throughput and less packet and stability period is drop is achieved compromised
Random
3.4 EMT: The Proposed Scheme We propose EMT scheme to minimize the energy consumption of sensor nodes in UWSNs. Lifetime of a UWSN increases when sensor nodes in that network conserve energy. Energy conservation is required specially, in UWSNs which are designed for time critical applications.
3.4.1 Network Model We deploy 50 sensor nodes with adjustable transmission power level in a circular monitoring area
π΄
of radius
π
as shown in fig. 3-1. Network area is divided into 8
sectors of equal size. Sensor nodes are randomly distributed because in UWSN, nodes change their position with water currents. Nodes sense environment and report data using two transmissions; hop-by-hop and direct.
Sensor nodes of each sector send
a fraction of data to a forwarder node of that sector which forwards received plus generated data to sink.
3.4.2 Network Configuration Initially, after node deployment sensor nodes in the network are unaware of relative coordinates. We initiate operation of EMT scheme with network configuration phase in which all sensor nodes exchange a HELLO message to inform neighbor nodes about
41
Chapter 3 their relative coordinates. Sensor nodes broadcast a HELLO message having information of sender node ID, neighbors and distance from sink. All sensor nodes broadcast a HELLO message at the start of each round because in underwater environment nodes change their position due to water currents. Also when a sensor node die in the network after data transmission and reception in a round then a network update configuration in next round to change the routing path.
Sink
Mul-hop Transmission
Sensor Node
Direct Transmission
Figure 3-1: Network Model of EMT Scheme
3.4.3 Forwarder Node Selection In EMT scheme, after network configuration a node is selected as forwarder which collects some fraction of data from previous hop nodes and forwards received plus generated data to sink. The forwarder is selected on the basis of distance from sink. First, distance of all nodes is checked from sink.
A node is selected as an eligible
forwarder if its distance from sink is minimum from all nodes in the sector. In each round energy of selected forwarder is checked, if it drain-outs all its energy then a new forwarder with minimum distance from sink is selected. There is a single forwarder selected in each sector and each forwarder node sends data to sink directly.
42
3.4.4. Data Distribution and Data Transmission
Chapter 3
3.4.4 Data Distribution and Data Transmission In EMT scheme, data is distributed in two fractions; small fraction of data and large fraction of data.
Nodes of each sector transmit small fraction of data directly to
sink and large fraction of data to the forwarder node in that sector. Forwarder node locally collects data from all sensor nodes of its sector and directly sends all data received plus generated to sink. We consider power adjustable sensor nodes, which transmit data directly to sink if no forwarder is found in sector. Also, due to random movement of nodes, the number of nodes in a sector can be increased or decreased. However, if number of nodes in a sector is maximum then load on forwarder will be maximum and if number of nodes in a sector is minimum than load on forwarder will be minimum.
3.5 Performance Evaluation The performance of EMT is evaluated by comparing its simulation results with existing BR scheme. Simulations are performed in both homogeneous and heterogeneous simulation environments to analyse the affects of the EMT protocol in both environments.
3.5.1 Simulation Setup We consider a circular monitoring area 8 sectors.
π΄
of radius
π
which is logically divided into
We randomly deploy 50 sensor nodes in the monitoring area.
Initially,
energy level of each sensor node is same. We also implement EMT in homogeneous and heterogeneous environments. In Homo-EMT, energy level of each node is same while, in Hetero-EMT nodes are provided different energy levels to prolong network lifetime. In EMT, we consider sensor nodes which adjust their transmission power level according to the distance of transmission. Sensor nodes transmit data directly as well as hop-by-hop to sink. Each node consumes energy during transmission and reception of data. Nodes near to sink directly forward the data to sink. We summarize
43
Chapter 3 simulation parameters in the table 3.2. Table 3.2: Parameters Setting for Simulation
Parameter
Value
Network Area Radius (R)
1000m
Nodes distribution
random
Corona Width (r)
100m
Sink Number
1 , Fixed at the Center
Data Rate
10 packet/s
Data packet size
1024 bits
Energy
1 Joule
3.5.2 Metrics The following metrics are considered for simulations.
3.5.2.1 Network Lifetime The duration from initialization of a network till death of the last node.
3.5.2.2 Residual Energy Remaining energy of a node after transmission and reception of data. It is measured in joules
3.5.2.3 Stability Period Duration between network establishment and death of first node.
3.5.2.4 Instability Period Instability period starts from the death of first node till the death of last node in the network.
44
3.5.3. Energy Consumption
Chapter 3
3.5.3 Energy Consumption In EMT scheme, sensor nodes transmit some fraction of data to the forwarder node near to sink and some fraction of data directly to sink while nodes which are one-hop neighbors to sink send all data directly.
Overall, energy consumption of nodes in
the network is minimized because maximum data is transmitted using hop-by-hop transmission. All sensor nodes forward data to sink in two transmissions; direct and hop-by-hop. In direct transmission sensor nodes send small fraction of data to sink to minimize energy consumption while in hop-by-hop transmission nodes near the sink receive large fraction of data. In addition, we consider random node deployment and also sensor nodes, which adjust their transmission power level according to distance from sink or selected forwarder nodes. In BR-scheme, all nodes transmit large portion of data using two-hop transmission range due to which maximum energy is consumed by each node as shown in fig. 3-2. Therefore, energy consumption of EMT scheme is better than BR scheme.
50 BR HomoβEMT HeteroβEMT
45 40
Residual energy
35 30 25 20 15 10 5 0
0
50
100
150
200
250
Rounds
Figure 3-2: Energy Consumption of EMT Scheme
3.5.4 Stability and Instability Period Stability period of BR scheme is better than EMT scheme because BR scheme balances the overall load on each node in network.
45
In EMT scheme, nodes near to
Chapter 3 sink forward maximum data and consume maximum energy. Therefore, a node near the sink which is selected as forwarder consumes maximum energy which leads to minimum stability of the network.
In EMT scheme, our target is to minimize en-
ergy consumption in a circular network.
Therefore, we select node near to sink as
a forwarder. Instability of BR scheme is minimum because all nodes transmit large fraction of data using two-hop transmission range due to which all sensor nodes die very soon as shown in fig. 3-3. In EMT scheme, sensor nodes in network consume minimum energy by forwarding minimum data directly to sink and maximum data using one-hop transmission range.
Therefore, instability of EMT scheme is better
than BR scheme.
50 45
Number of dead nodes
40 35 30 25 20 15 10
BR HomoβEMT HeteroβEMT
5 0
0
50
100
150
200
250
Rounds
Figure 3-3: Number of Dead Nodes in Network
3.5.5 Network Lifetime Overall network lifetime of EMT scheme is better than BR scheme because of minimum energy consumption of sensor nodes in network. All nodes near the sink forward maximum data due to which nodes far from the sink are relieved of data load which prolongs the lifetime of network in EMT scheme. In BR scheme, due to maximum data transmission at long distance sensor nodes consume maximum energy which minimizes lifetime of the network. Lifetime of EMT scheme shows a sudden decrease
46
3.5.6. Implementation of EMT in Homogeneous and Heterogeneous Environments Chapter 3 in fig. 3-4 because sensor nodes near the sink die quickly because of forwarding maximum data to the sink. However, at a certain point lifetime of EMT scheme improves gradually from BR scheme because in BR scheme nodes after a long distance transmissions drain-out energy and die. Therefore, EMT scheme beats BR scheme in terms of network lifetime.
50 BR HomoβEMT HeteroβEMT
45 40
Network lifetime
35 30 25 20 15 10 5 0
0
50
100
150
200
250
Rounds
Figure 3-4: Network Lifetime of EMT Scheme
3.5.6 Implementation of EMT in Homogeneous and Heterogeneous Environments First, we implement EMT scheme in homogeneous simulation environment and energy levels of all sensor nodes are set same. As discussed above, in Homo-EMT sensor nodes with same energy levels perform better than BR scheme due to minimum data transmission at long distance and maximum data transmission at near distance. However, some sensor nodes in Homo-EMT consume maximum energy while, some nodes consume minimum energy.
Therefore, to improve network lifetime we set different
energy levels of nodes. Energy levels of nodes which consume maximum energy are set maximum while, energy levels of sensor nodes which consume minimum energy are set minimum due to which Hetero-EMT scheme prolongs network lifetime as shown in fig. 3-4.
47
Chapter 3 Table 3.3: Performance Tradeoffs
Protocols
Technique
BR
Balanced
Parameters achieved
routing
technique
Load
balancing
and
network
Cost to pay Stability due to transmission at two-hop
lifetime
Homo-EMT Hetero-EMT
Mixed
Energy efficiency and
Unbalanced
same
network
consumption
energy levels
lifetime
Mixed
routing
routing
with
with
different energy levels
Stability
period
of
Instability period and
network, lifetime and
unbalanced
energy efficiency
consumption
3.6 Performance Tradeoffs In this section, we discuss performance tradeoffs made by proposed and existing schemes. In BR scheme, load is distributed to minimize energy consumption of nodes near sink and maximum data is forwarded at long distance using two-hop transmission range. Therefore, nodes transmitting data at two-hop near to sink consume maximum energy as compare to nodes far from sink. Therefore, these nodes die quickly which minimizes stability period of network.
In Homo-EMT and Hetero-EMT, nodes in
each sector send small fraction of data directly to sink and large fraction of data to nodes near sink.
Therefore, nodes receiving and forwarding large fraction of data
quickly depletes energy and die. In Hetero-EMT nodes are assigned different energy levels due to which nodes with minimum energy quickly die and minimizes instability period. Performance tradeoffs are shown in table 3.3.
3.7 Conclusion of the Chapter In UWSNs, energy efficiency is one of the most important factor to maximize network lifetime because sensor nodes are provided limited battery power and recharging or replacement of battery is difficult in harsh underwater environment. Many researchers have worked on energy efficiency in UWSNs considering unique characteristics of
48
energy
energy
3.7. Conclusion of the Chapter
Chapter 3
underwater environment. EMT is an energy minimization routing protocol, designed for continuous monitoring application in UWSNs.
In EMT, a forwarder node is
selected in each sector on the basis of physical distance. In forwarder selection phase distance between sink and sensor nodes of a sector is calculated then a node with minimum distance from sink is selected as an eligible forwarder. Each node in a sector forwards some fraction of data to a selected forwarder node and some fraction of data directly to sink to minimize load. Simulation results show that EMT performs better than existing BR scheme, designed for continuous monitoring applications in UWSNs. Implementation of EMT in homogeneous and heterogeneous environments show that Hetero-EMT performs better than Homo-EMT in terms of energy consumption and network lifetime.
49
References Routing design avoiding energy holes in underwater acoustic sensor networks. Wireless Communications and
[1] Zidi, C., Bouabdallah, F. and Boutaba, R., 2016. Mobile Computing. [2] Cao, J., Dou, J. and Dong, S., 2015.
derwater acoustic sensor networks.
Balance transmission mechanism in un-
International Journal of Distributed Sensor
Networks, 2015, p.2. [3] Javaid, N., Shah, M., Ahmad, A., Imran, M., Khan, M.I. and Vasilakos, A.V.,
An Enhanced Energy Balanced Data Transmission Protocol for Underwater Acoustic Sensor Networks. Sensors, 16(4), p.487. 2016.
[4] Latif, K., Javaid, N., Ahmad, A., Khan, Z.A., Alrajeh, N. and Khan, M.I., 2016.
On energy hole and coverage hole avoidance in underwater wireless sensor networks. IEEE Sensors Journal, 16(11), pp.4431-4442.
A New Node Deployment and Location Dispatch Algorithm for Underwater Sensor Networks. Sensors, 16(1),
[5] Jiang, P., Liu, J., Ruan, B., Jiang, L. and Wu, F., 2016. p.82.
Energy balanced strategies for maximizing the lifetime of sparsely deployed underwater acoustic sensor networks. Sensors, 9(9), pp.6626-6651.
[6] Luo, H., Guo, Z., Wu, K., Hong, F. and Feng, Y., 2009.
[7] Li, Z., Yao, N. and Gao, Q., 2014.
underwater wireless networks.
Relative distance based forwarding protocol for
International Journal of Distributed Sensor Net-
works, 2014.
An efficient dynamic addressing based routing protocol for underwater wireless sensor networks. Computer
[8] Ayaz, M., Abdullah, A., Faye, I. and Batira, Y., 2012. Communications, 35(4), pp.475-486.
[9] Huang, C.J., Wang, Y.W., Liao, H.H., Lin, C.F., Hu, K.W. and Chang, T.Y., 2011.
A power-efficient routing protocol for underwater wireless sensor networks.
Applied Soft Computing, 11(2), pp.2348-2355.
An energy-efficient routing protocol for UWSNs using physical distance and residual energy. In OCEANS, 2011 IEEE-
[10] Wahid, A., Lee, S. and Kim, D., 2011, June. Spain (pp. 1-6). IEEE.
50
REFERENCES
Chapter 3
A reliable and energyΓ’ΔΕefficient routing protocol for underwater wireless sensor networks. International Journal of Com-
[11] Wahid, A., Lee, S. and Kim, D., 2014.
munication Systems, 27(10), pp.2048-2062.
Lifetime maximization through dynamic ring-based routing scheme for correlated data collecting in WSNs. Com-
[12] Jiang, L., Liu, A., Hu, Y. and Chen, Z., 2015. puters & Electrical Engineering, 41, pp.191-215.
End-to-end delay and energy efficient routing protocol for underwater wireless sensor networks. Wireless Personal Com-
[13] Ali, T., Jung, L.T. and Faye, I., 2014. munications, 79(1), pp.339-361.
A Localization Method for Underwater Wireless Sensor Networks Based on Mobility Prediction and Particle Swarm Optimization Algorithms. Sensors, 16(2), p.212.
[14] Zhang, Y., Liang, J., Jiang, S. and Chen, W., 2016.
A new multi-path routing protocol based on cluster for underwater acoustic sensor networks. In Multimedia Technology (ICMT),
[15] Liu, G. and Wei, C., 2011, July.
2011 International Conference on (pp. 91-94). IEEE.
DBR: depth-based routing for underwater sensor networks. In NETWORKING 2008 Ad Hoc and Sensor Networks, Wireless
[16] Yan, H., Shi, Z.J. and Cui, J.H., 2008.
Networks, Next Generation Internet (pp. 72-86). Springer Berlin Heidelberg.
An energy efficient localization-free routing protocol for underwater wireless sensor networks. International journal of distributed
[17] Wahid, A. and Kim, D., 2012. sensor networks, 2012.
WDFAD-DBR: Weighting depth and forwarding area division DBR routing protocol for UASNs.
[18] Yu, H., Yao, N., Wang, T., Li, G., Gao, Z. and Tan, G., 2016. Ad Hoc Networks, 37, pp.256-282.
[19] Javaid, N., Jafri, M.R., Khan, Z.A., Qasim, U., Alghamdi, T.A. and Ali, M.,
Iamctd: Improved adaptive mobility of courier nodes in threshold-optimized dbr protocol for underwater wireless sensor networks. International Journal of
2014.
Distributed Sensor Networks, 2014, p.1. [20] Umar, A., Javaid, N., Ahmad, A., Khan, Z.A., Qasim, U., Alrajeh, N. and Hayat,
DEADS: Depth and Energy Aware Dominating Set Based Algorithm for Cooperative Routing along with Sink Mobility in Underwater WSNs. Sensors, A., 2015.
15(6), pp.14458-14486.
51
Chapter 4 BLOAD: Energy Holes Avoidance with Balanced Energy Consumption for UWSNs
52
4.1. Summary of the Chapter
Chapter 4
4.1 Summary of the Chapter Energy balancing is one of the key requirement of UWSNs because of limited energy resources. In this chapter, we present BLOAD scheme to avoid energy holes created due to unbalanced energy consumption in UWSNs. Our proposed scheme prolongs the stability and lifetime of UWSNs. In BLOAD scheme, the data (generated plus received) of each sensor node is distributed among its next hop neighbor nodes to balance the energy consumption for avoidance of energy hole in the network.
The
distinct feature of BLOAD scheme is that each sensor node in the network continuously report data to the sink till its death even an energy hole is created in its next hop region. We implement BLOAD scheme, by varying the data load weights of sensor nodes having variable transmission range in homogeneous and heterogeneous simulation environments.
The results prove that BLOAD scheme outperforms the
existing selected schemes in terms of network lifetime and stability.
4.2 Motivation Sensor nodes in UWSNs report generated data as well as relay received data of previous hop sensor nodes to the sink. Therefore, nodes near the sink forward more data as compare to the nodes farther from the sink and the energy consumption of the nodes near the sink is higher than other nodes of the network.
Direct data trans-
mission is an easy way for reporting data to the sink if transmission range of each sensor node is large enough to reach the sink. However, if each sensor node only uses power-adjusted direct transmission, then the nodes farther from the sink quickly run out of energy and die within no time. Hop-by-hop transmission is used to save energy for long-distance transmission. However, hop-by-hop transmission causes maximum energy consumption of sensor nodes near the sink which leads to energy hole. In Nominal Range Forwarding (NRF) scheme [1], sensor nodes send data (generated plus received) to the sink using one-hop transmission range i.e.
{π}
which
causes maximum data traffic load on sensor nodes of corona 1 near the sink.
53
Due
Chapter 4 to maximum energy consumption, nodes of corona 1 die within no time and leads to energy hole near the sink. Homogeneous-BR (Homo-BR) scheme [2] minimizes the total data traffic load at corona 1 by distributing data traffic load among all coronas of the network using transmission range of
{π, 2π}.
However, data load at corona 2,
which is 2-hop away neighbor of the sink increases because sensor nodes of corona 2 become one-hop neighbors of the sink using transmission range
2π,
which leads to
unbalanced data load and high energy consumption at corona 2. We observed that if sensor nodes at corona 2 die, the total data traffic at corona 3 and corona 4 increases, due to which data load received by corona 1 from corona 3 using transmission range
2π
also increases.
Thus, balanced data load distribution is needed to achieve bal-
anced energy consumption to avoid the energy hole problem. In our study, we target both 1-hop and 2-hop away neighbors of the sink because in continuous monitoring applications, maximum data is forwarded by these neighbor nodes of the sink. We summarize the problems discussed above as follows. (i)Balancing data load among all sensor nodes to balance energy consumption of all nodes to increase stability and network lifetime, (ii)Minimizing energy consumption of both 1-hop and 2-hop away neighbors of the sink to overcome the energy hole problem in manually deployed UWSNs. Our goal is to evenly distribute data among all sensor nodes of the network in which each sensor node is continuously monitoring the area and reports data (generated plus received) to the sink periodically.
4.3 Related Work In the past decade, many researchers have proposed routing protocols for energy balancing in WSNs, which are mostly unusable for UWSNs because of unique characteristics of underwater channel. The energy hole, which is formed due to unbalanced energy consumption is one of the key issue which has attracted the attention of the researchers. In this section, we discuss some existing protocols related to routing in UWSNs and state-of-the-art related work is given in table 4.1. In Balanced Transmission Mechanism (BTM) [3], hybrid routing mechanism is
54
4.3. Related Work
Chapter 4
adopted and a two-dimensional network model is proposed. BTM balances data load among all sensor nodes by dividing the energy of each sensor node into energy levels. When a node excessively consumes energy in multi-hop transmission, it communicates directly with the sink.
The disadvantage of BTM is high energy consumption at
long distance transmission when the network radius is increased.
Also, the energy
consumption in the coronas near and farther from the sink is not balanced. Enhanced Efficient and Balanced Energy consumption Technique (EEBET) [4], overcomes the deficiencies in BTM. EEBET presents Efficient and Balanced Energy consumption Technique (EBET) to avoid direct transmission at long distance for saving energy and calculates optimum number of energy levels to enhance network lifetime. The authors in EEBET protocol consider the problem of varying network radii for energy balancing.
The issue of the energy sink hole at 1-hop and 2-hop
neighbor coronas of the sink is not addressed by EEBET protocol. In [5], authors propose Spherical Hole Repair Technique (SHORT) to repair coverage holes which are created due to energy holes. The technique has three phases: Knowledge Sharing Phase (KSP), Network Operation Phase (NOP) and Hole Repair Phase (HRP). In SHORT, nodes adjust their transmission power levels according to the location of next hop node. SHORT takes the advantage of redundant overlapping of sensing ranges of nodes in dense UWSNs. However, due to coverage hole repairing time, SHORT is not suitable for delay sensitive applications. Peng
et al.
also address coverage problem in UWSNs and propose a Guaranteed
Full Connectivity Node Deployment (GFCND) and a Location Dispatch Based on Command Nodes (LDBCN) algorithms [6]. GFCND algorithm, logically divides sensor nodes into two types: command nodes and connectivity nodes. Greedy iterative strategy is used to deploy coverage nodes for large coverage of the network and connectivity nodes are used for maximum network connectivity. The location adjustment of the common nodes with the help of the command nodes and sink nodes is accomplished by the LDBCN algorithm, to obtain the required network coverage rate and fully connected UWSNs. Hanjiang
et al.
in [7], present Energy Balanced Hybrid (EBH) algorithm and Dif-
55
Chapter 4 ferential Initial Battery (DIB) assignment technique for balanced energy consumption of sensor nodes deployed in sparse linear networks. Maximum lifetime is achieved with EBH and DIB strategies in sparsely deployed UWSNs. However, the proposed strategies are inefficient for densely deployed UWSNs. These techniques are inefficient for non-linear networks. The authors in [8], analyze the impact of node deployment in three-dimensional underwater environment. They propose three deployment schemes; random deployment, cube deployment and regular tetrahedron deployment. The tetrahedron deployment scheme performs better than the other selected schemes in terms of localization ratio, localization error, average number of neighboring anchor nodes and network connectivity. In [9], authors propose a Relative Distance Based Forwarding (RDBF) scheme for energy efficient and minimum delay routing in UWSNs.
In RDBF, an appropriate
sensor node is selected as a forwarder on the basis of a fitness function. The sensor nodes selected as forwarders are only involved in routing. The forwarders nodes are selected for routing because of minimum distance from the sink and minimum hop counts of routing path. RDBF achieves low end-to-end delay and energy efficiency in the network. Ali
et al.
in [10], present Layer-by-Layer Angle-Based Flooding (L2-ABF) routing
protocol for UWSNs. Sensor nodes are deployed in the form of layers and each sensor node calculates its depth.
L2-ABF is a localization-free routing protocol in which
each sensor node forwards data to the sink according to the calculated forwarding angle.
Sensor nodes forward data to the node which lies in its angle-based zone.
L2-ABF achieves less energy consumption and high packet delivery ratio. Authors in [11], propose a localization technique based on Mobility Prediction and a Particle Swarm Optimization algorithm (MP-PSO) for UWSNs. Beacon nodes are deployed to find the location information of unknown nodes in the network. Velocity information of each beacon node is acquired with the help of which unknown nodes velocities are estimated and with the help of mobility prediction technique, the location of an unknown node is predicted.
56
4.3. Related Work
Chapter 4
In [12], authors present a Hierarchical Multi-path Routing-LEACH (HMR-LEACH) algorithm for UWSNs, which is based on LEACH algorithm. They uses multipath routing algorithm instead of one hop routing algorithm. HMR-LEACH, checks the energy and distance for selecting the transmission path while transmitting data to the sink and assigns a weight to each transmission path.
Each transmission path
has its own selection probability. The HMR-LEACH technique improves the network lifetime and balance the energy consumption.
57
Chapter 4 Table 4.1: State-of-the-art Related Work
Deployment Pattern
Protocol
Feature(s)
BTM [3]
Direct Transmission, Multihop Transmission, Balanced Energy Consumption
Uniform
EEBET [4]
Balanced Data Transmission
Random Uniform
Coverage Hole Repair Technique Guaranteed Full Connectivity Node Deployment Technique
Random
Energy Efficiency, Network Lifetime and Throughput Network Lifetime and Throughput
Random
Network Coverage and Connectivity rate
SHORT [5] GFCND [6]
Parameter(s) Achieved
Energy Efficiency, Network Lifetime
Reducing localization error and Increasing Regular Tetrahedron, localization accuracy Random and Cube
Deployments [8]
Node deployment technique
EBH [7]
Hybrid Data Transmission for Balanced Energy
Uniform
Balanced Energy Consumption, Maximum Network Lifetime
RDBF [9]
Relative Distance Based Forwarding
Uniform
Low end-to-end delay, Energy efficiency
L2-ABF [10]
Angle Based Flooding, Localization-free Routing
Random Uniform
Energy Efficiency, Packet Delivery Ratio
Localization-aware Routing Hierarchical Multipath Routing Technique
Random
Energy Efficiency, Localization accuracy Network Lifetime, Energy Efficiency
MP-PSO [11]
HMR-LEACH [12]
Random
Parameter(s) Compromised
High energy consumption for long distance transmission, Formation of transmission loops Performance decreases with increase of network radius End-to-end Delay No Relationship of network topology with random node scattering High deployment cost of sensor nodes Inefficient for Non-Linear and Dense Networks Unbalanced load on nodes with minimum hop counts and less distance from the sink End-to-end delay increases by increasing the number of layers High Computational Complexity High End-to-end delay
4.4 System Model The system under consideration is a circular monitoring area the density of deployed nodes is
π [1], [2].
π΄ of radius π
such that
We consider continuous-monitoring applica-
tions in which underwater sensor nodes are deterministically anchored at the bottom in shallow UWSN. The network field is logically divided into concentric coronas of
58
4.4.1. The Underwater Channel Model equal width
π
as shown in fig. 4-1.
Chapter 4
Each corona contains same number of nodes.
Sensor nodes periodically report data to the sink by forwarding data to the next corona node located in the way to sink. Traffic is distributed for balancing energy consumption. Each sensor node adjusts its transmission power level according to the transmission distance [13].
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 Sensor Node
Sink
Figure 4-1: Network Model of BLOAD Scheme
4.4.1 The Underwater Channel Model Transmission loss of an acoustic signal in underwater is given as follows [14]:
π πΏ = π Β· 10πππ(π) + π Β· 10πππ(π(π )) Where,
π
is distance in km,
for spherical spreading, spreading [15].
π
is signal frequency,
π = 1
In eq. (4.1),
π
is the spreading factor.
for cylindrical spreading and
10ππππ(π )
(4.1)
π = 1.5
for practical
is the absorption coefficient (in
which can be calculated using Thorpβs formula [16]:
59
π =2
ππ΅/πΎπ),
Chapter 4
πππ10 π(π ) = 0.11
π2 π2 + 44 + 1 + π2 4100 + π 2
(4.2)
2.75 Γ 10β4 π 2 + 0.003 Equation (4.2) is for frequencies above few hertz, for lower frequencies the following equation is used.
10πππ10 π(π ) = 0.11
π2 + 0.11π 2 + 0.002 1 + π2
Attenuation in underwater acoustic channel is given in
ππ΅
(4.3)
as follows:
π΄(π, π ) = ππ π(π )π The SNR of an emitted signal of power
π
ππ π
(π, π ) = where,
βπ
(4.4)
is given in eq. (4.5).
π/π΄(π, π ) π (π )βπ
(4.5)
is the receiver noise bandwidth.
Acoustic signals are effected by four types of noise in underwater communication: turbulence noise
ππ‘β (π ).
ππ‘ (π ),
waves noise
ππ€ (π ),
shipping noise
ππ (π )
and thermal noise
The power spectral density of ambient noise is given in eq. (4.6) as
π (π ) = ππ‘ (π ) + ππ (π ) + ππ€ (π ) + ππ‘β (π ).
(4.6)
4.4.2 Energy Consumption Model The energy consumption of sensor nodes is due to transmission and reception and the total energy consumed by transmitting a packet of
ππ
bits over a distance
π
is given
as follows:
πΈπ‘π₯ (π) = ππ (π) Γ
60
ππ πΆ3ππ΅(π)
.
(4.7)
4.5. The BLOAD
Chapter 4
The energy spent in receiving a packet of
πΈππ₯ (π) = πππ₯ Γ where,
ππ
is the transmitting power,
can be calculated similar to [2]. over bandwidth
π΅3ππ΅ (π)
πππ₯
πΆ3ππ΅ (π)
ππ
bits is given in eq. (4.8) as
ππ πΆ3ππ΅(π)
,
(4.8)
is the electronics power and their values
is the capacity which is maximum allowed
as [15] and it can be calculated as follows:
(οΈ
β«οΈ πΆ3ππ΅ (π) =
πππ2 π΅3ππ΅ (π)
ππ‘π₯ (π)/π΅3ππ΅ (π) 1+ π΄(π, π )π (π )
)οΈ ππ
(4.9)
4.5 The BLOAD BLOAD scheme operates according to the following phases and data is generated per unit time by each sensor node which is periodically reported to sink. In phaseI network is configured and all sensor nodes are informed about their location and distance from sink.
Then, load is calculated for each corona nodes in phase-II. In
phase-III, data is transmitted according to the calculated load weight.
4.5.1 Network Configuration Initially, when sensor nodes are deployed, they are unaware of their relative coordinates.
In order to start a successful communication, each sensor node must know
its location with respect to neighbor nodes and sink.
Network setup is configured
each time before all sensor nodes start data transmission. When network configuration starts, each sensor node is informed about the coordinates of sink and all sensor nodes in the network with HELLO message exchange mechanism. Now, the sensor nodes are ready to begin the second phase of load distribution.
4.5.2 Load Distribution Phase Before sensor nodes start transmission, load weight each transmission
{π, 2π, ππ‘π₯}.
{π 1 , π 2 , π 3 }
is calculated for
We found that different combinations of the data load
61
Chapter 4 Table 4.2: Possible Combinations of Two Sets W and R. Comb:01
Comb:02
R W 1
π π2 π3
Comb:03
R
Comb:04
R
Comb:05
R
Comb:06
R
R
r
2r
dtx
r
2r
dtx
r
2r
dtx
r
2r
dtx
r
2r
dtx
r
2r
dtx
1
0
0
0
1
0
0
0
1
1
0
0
0
1
0
0
0
1
0
1
0
1
0
0
1
0
0
0
0
1
0
0
1
0
1
0
0
0
1
0
0
1
0
1
0
0
1
0
1
0
0
1
0
0
r 2r dtx
2r r dtx
dtx r 2r
r dtx 2r
2r dtx r
dtx 2r r
weights and transmission ranges are formed using eq. (4.10) below which are shown in table 4.2. In order to find an optimum solution, we find all the possible combinations of the data load weights and transmission ranges.
We have two sets
π
and
π
as
follows: where,
π = {π 1 , π 2 , π 3 }
and
π
= {π, 2π, ππ‘π₯}. The total number of possible combinations can be calculated as follows:
πΆπππ(π, π
) =
π! . (π β π
)!
(4.10)
In BLOAD scheme, the data (generated plus received) by each node is distributed in such a way that energy consumption at each corona nodes is balanced and stability of the network is prolonged. Data distribution is shown in fig. 4-2. We suppose that initially each sensor node of the first corona total data traffic of each sensor of thus, their data load weight corona
πΆ1
πΆ1
generates
π·
number of packets. The
is composed of only locally generated data
π11 (π) is equal to 1.
π·
By receiving the traffic of the second
πΆ2 , the data forwarding load on πΆ1 nodes becomes relatively high.
We compute
π21 (π) and π22 (2π) for corona πΆ2 such that π21 (π)+π22 (2π) = 1 and π22 (2π) > π21 (π). When the data is received from third corona and
πΆ3
is balanced using load weight
π31 , π32
πΆ3 , and
the energy consumption of
π33 .
πΆ1 , πΆ2
The total data load weight is
given in eq. (4.11) as
π 1 + π 2 + π 3 = 1, where,
π 3 > π 2 > π 1.
62
(4.11)
4.5.3. Data Transmission Phase
Chapter 4
Table 4.3: Packet Load Distribution of BLOAD Scheme
Corona Corona 1 Corona 2 Corona 3 Corona 4 Corona 5 Corona 6 Corona 7 Corona 8 Corona 9 Corona 10
π1
π2
π3
1
0
0
0.02
0.98
0
0.2275
0.35
0.4225
0.1204
0.14
0.7396
0.1875
0.25
0.5625
0.1411
0.17
0.6889
0.1476
0.18
0.6724
0.1275
0.15
0.7225
0.1131
0.13
0.7569
0.09
0.1
0.81
4.5.3 Data Transmission Phase πΆπ
transmits total data (generated plus received)
with load weight
{ππ1 , ππ2 , ππ3 }
π·πΆπ ,
which is given in eq. (4.12)
using transmission range
{π, 2π, ππ‘π₯}:
1 2 Β· π·π+1 + ππ+2 Β· π·π+2 , π·πΆπ = π· + ππ+1
(4.12)
β 1 β€ π β€ πΆ β 2, where,
π·
is the average number of packets generated by each node in
πΆπ
per unit
of time.
Each node in a corona receives data from one-hop and two-hop neighbor nodes. The number of data packets received by
πΆπ
from one-hop and two-hop neighbor nodes
is given as
ππΆπ =
π βοΈ
1 ππ+π Β· π·πΆπ+π
π=1
63
(4.13)
Chapter 4
ππΆπ =
π βοΈ
2 Β· π·πΆπ+π ππ+π
(4.14)
π=1
π·πΆπ,π = ππΆπ + ππΆπ
By putting the values of respectively, and setting
ππΆπ
and
ππΆπ
in eq. (4.15) from eq. (4.13) and eq. (4.14),
1 2 ππ+π + ππ+π = 1,
π·πΆπ,π =
π βοΈ
(4.15)
we get the following equation.
β 1β€πβ€πΆ β2
π·πΆπ+π
(4.16)
π=1
Total energy consumption of nodes in each corona is analyzed by calculating the energy consumption for receiving data by corona
π
nodes. The energy consumed by
each node during data reception from single hop and multihop neighbor nodes is given as:
πππΆπ
π βοΈ )οΈ (οΈ 1 π·πΆπ+π Β· ππ+π = πΈππ₯(π)
(4.17)
π=1
πππΆπ = πΈππ₯(2π)
π βοΈ (οΈ )οΈ 2 π·πΆπ+π Β· ππ+π
(4.18)
π=1
By adding eq. (4.17) and eq. (4.18), we get:
(οΈ ππ,π =
πΈππ₯(π)
π βοΈ
)οΈ 1 π·πΆπ+π Β· ππ+π
+
π=1
(οΈ πΈππ₯(2π)
π βοΈ π=1
64
)οΈ 2 π·πΆπ+π Β· ππ+π
(4.19)
4.6. Performance Evaluation
Chapter 4
(οΈ ππ,π‘ =
πΈπ‘π₯(π)
π βοΈ
)οΈ 1 π·πΆπ+π Β· ππ+π
+
π=1
(οΈ πΈπ‘π₯(2π)
π βοΈ
)οΈ 2 π·πΆπ+π Β· ππ+π
+
(4.20)
π=1
(οΈ πΈπ‘π₯(ππ‘π₯)
π βοΈ
)οΈ 3 π·πΆπ+π Β· ππ+π
π=1 Thus, total energy consumption at corona
πΆπ
is calculated as
ππ = ππ,π + ππ,π‘
(4.21)
3
W 3
W
2
2
W -
1
K-2
1
W
Sink
K-1
1
W
2
W -
W K
1K
W
+1
1
W
K+ 2
W
3
W
3
r
W
2r
dtx
Figure 4-2: Network Linear Model of BLOAD Scheme
The working of BLOAD scheme is shown in fig. 4-3. Working of Hetero-BR scheme is same as Homo-BR scheme [2].
The only difference is that the energy levels are
kept different in Hetero-BR scheme on the basis of maximum and minimum energy consumption of nodes. The sum of the total energy of the network is same of both schemes. We set different energy levels of nodes along with dedicated deployment to achieve maximum stability and to avoid energy holes in the network.
4.6 Performance Evaluation In this section, we evaluate performance of BLOAD scheme by comparing its simulation results with two selected existing schemes: NRF scheme [1] and BR scheme [2]. There are 50 senosr nodes deployed uniformly in a circular network field of radius
π
= 1000π.
The width of each corona is
65
100π.
Each sensor node uses variable
Chapter 4 Uniform Node Deployment in a Circular Field
Network Conguraon
START
Compute Load Weight (W1,W2,W3)
if Coronan=1
for Corona n = 1 to n
Yes
Forwards Data to Sink using dtx
Yes
Forwards Data at 1-hop using r
Yes
Forwards Data at 2-hop using 2r
No
for Noden = 1 to n
No
if Coronan=2
No
No
if Noden Alive?
if Coronan>3
Yes
Generate Data END
Figure 4-3: Working Flow of BLOAD Scheme
transmission range and generates data 10 packet/s. We found that using nominal communication range
{π}
each corona except last
corona receives cumulative data of previous coronas and energy consumption of each corona increases accordingly which leads to 100 packet/s traffic load and maximum energy consumption at corona 1. Data traffic load is minimized by less than 40 packet/s and minimum energy consumption is achieved at corona 1 in Homo-BR scheme using variable transmission range i.e.
{π, 2π}
but energy consumption of corona 2 is
increased due to which all sensor nodes of corona 2 die within no time as shown in fig. 4-4. In Homo-BR scheme, a packet load distribution matrix is derived for transmission range
{π, 2π} to evenly distribute the energy consumption among all coronas.
We derive the packet load distribution matrix for mixed routing i.e.
{π, 2π, ππ‘π₯}
to
balance energy consumption of all coronas for maximum network lifetime as shown in table 4.3. In Homo-BLOAD scheme, the total traffic load at each corona is balanced because
66
4.6. Performance Evaluation
Chapter 4
some fraction of the data is directly sent to sink and energy consumption among all coronas is balanced using hop-by-hop and direct transmission. In figures below, it can be easily observed that load distribution of corona 10 is same in both Homo-BLOAD and Homo-BR schemes because corona 10 nodes do not receive data and only generate 10 packets/sec. It is found from the simulation results that energy consumption of corona 10 nodes in Homo-BLOAD scheme is greater than Homo-BR scheme because data packets are sent directly as well as hop-by-hop to the sink using transmission range
{π, 2π, ππ‘π₯} while, the total data in Homo-BR scheme is sent using transmission
range
{π, 2π}.
Overall packet load distribution of all coronas is balanced in Homo-BLOAD scheme because all coronas send a data fraction of the total data traffic direct to the sink using transmission range
ππ‘π₯ and the remaining data fraction of the total data traffic is sent
to next 1-hop and 2-hop coronas using transmission range
π
and
2π.
The minimum
data traffic load at 1-hop and 2-hop away neighbors of the sink leads to minimum energy consumption of corona 1 nodes and corona 2 nodes in Homo-BLOAD scheme as discussed in the problem statement section and it is shown in fig. 4-8a and fig. 4-8b. In Homo-BLOAD scheme, the total data traffic is forwarded with three data load weights to the sink using three transmission ranges.
We get six different scenarios
from the combination of data load weights and transmission ranges. First we discuss the scenario of the Comb:01.
In this scenario, the small data fraction of the total
data traffic is sent to the adjacent 1-hop neighbor of each node with data load weight
π1
using transmission range
π.
The medium data fraction of the total data is sent
to next 2-hop neighbor of a node with data load weight
2π
π2
using transmission range
and the large portion of the total data traffic is directly sent to the sink with load
weight
π3
using transmission range
ππ‘π₯.
The results of the packet load distribution
and energy consumption of scenario Comb:01 are shown in the fig. 4-4a and fig. 4-4b. Overall packet load distribution is balanced in this scenario because of the direct transmission of the large data portion to the sink. The energy consumption of the last corona is higher than all coronas because distance of the last corona is greater than the rest of the coronas.
The energy consumption of a corona is decreased as
67
Chapter 4 much as its distance from the sink because high energy is required to send large data portion of the total data traffic directly to the sink using transmission range
ππ‘π₯.
In
Comb:01, the energy consumption of the neighbours near sink is decreased but overall energy consumption of the network is not balanced. Medium data fraction of the total traffic is sent with data load 1-hop neighbor using transmission range
π
π2
to the next
and small data fraction is forwarded to
2-hop neighbor of a node with data load weight
π3
using transmission range
2π
in
Comb:02. The third transmission remains the same as Comb:01. fig. 4-5a and fig. 45b show almost same behaviour as Comb:01 because we observe that the transmission of large data fraction mainly affects the packet load distribution and the total energy consumption of the network as well. We found that the traffic load and energy consumption at corona 2 is relatively high because of sending the medium data fraction with
π2
using transmission range
π.
We consider the sensor nodes which adjust
their transmission power according to the transmission distance.
The total energy
consumption of the network is not balanced in Comb:02 because energy is consumed in data transmission at long distance by the coronas which are far from the sink. In Comb:03, small fraction of data is sent directly to the sink with data load weight
π1
using transmission range
weight
π2
ππ‘π₯
and medium fraction of data is sent with data load
using transmission range
π.
In Comb:04, the only difference from Comb:03
is that medium fraction of data is directly sent to the sink with load weight transmission range
ππ‘π₯
π2
using
and small fraction of data is forwarded with data load weight
π 1 using transmission range π.
The large fraction of data is forwarded to the adjacent
2-hop neighbor of a node with data load weight
π3
using transmission range
2π
in
both Comb:03 and Comb:04. The discussed scenario of three different transmissions of Comb:03 and Comb:04 shows the same behaviour as Homo-BR scheme because large fraction of data is sent using transmission range
2π
as in Homo-BR scheme.
The
fig. 4-6a and fig. 4-7a show the packet load distribution of Comb:03 and Comb:04, respectively. We observe that the fraction of data sent using two transmission ranges
π and 2π affect the packet load distribution while the rest of the data is sent directly to the sink in third transmission
ππ‘π₯.
The energy consumption of Comb:03 and Comb:04
68
4.6. Performance Evaluation
Chapter 4
is greater than Homo-BR scheme as shown in fig. 4-6b and fig. 4-7b because in HomoBLOAD scheme the total data traffic is forwarded using three transmission ranges {π, 2π, ππ‘π₯}.
Corona 2 nodes which are 2-hop neighbors of the sink consume more
energy than other corona nodes of the network. Thus, sensor nodes in corona 2 die within no time. Overall packet load distribution and the energy consumption of the network is not balanced as shown in fig. 4-6a, fig. 4-6b, fig. 4-7a and fig. 4-7b. We observe that in Comb:05 and Comb:06, the packet load distribution is balanced as shown in fig. 4-8a and fig. 4-9a. It is notable that the energy consumption of 1-hop and 2-hop neighbors of the sink is minimized as discussed in the problem statement section. Overall energy consumption in both Comb:05 and Comb:06 is balanced as shown in fig. 4-8b and fig. 4-9b because small and medium fractions of data of the total data traffic are forwarded to the 2-hop neighbor of a node and directly to the sink using transmission ranges
2π
and
ππ‘π₯.
The remaining large fraction of data is
forwarded to the adjacent neighbor node of each node with data load weight using transmission range
π,
π3
due to which the packet load distribution of the network
is balanced which leads to balanced energy consumption of the network. Our goal is to minimize the packet load distribution and the energy consumption of 1-hop and 2-hop neighbor nodes of the sink to avoid energy sink hole problem. It is noticeable that packet load on corona 1 and corona 2 nodes in Comb:05(fig. 4-8a) is about 5% less than Comb:06(fig. 4-9a).
The energy consumption of corona 1 and 2 nodes in
Comb:05(fig. 4-8b) is also less than Comb:06(fig. 4-9b) because in Comb:05 medium fraction of data is sent directly to the sink with load weight range
ππ‘π₯
π2
using transmission
which minimizes the data load traffic on corona nodes which are near to
sink. Thus, we find an optimal solution to avoid energy sink hole problem in Comb:05.
69
Chapter 4
100 NRF HomoβBR HomoβBLOAD
Packet Load per unit of time(packet/s)
90 80 70 60 50 40 30 20 10 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(a) Load Distribution Per Corona 0.1 NRF HomoβBR HomoβBLOAD
Energy consumption per unit of time(W)
0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(b) Energy Consumption of Different Coronas Figure 4-4:
Comb:01: {π
70
1
π, π 2 2π, π 3 ππ‘π₯}
4.6. Performance Evaluation
Chapter 4
100 NRF HomoβBR HomoβBLOAD
Packet Load per unit of time(packet/s)
90 80 70 60 50 40 30 20 10 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(a) Load Distribution Per Corona 0.1 NRF HomoβBR HomoβBLOAD
Energy consumption per unit of time(W)
0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(b) Energy Consumption of Different Coronas Figure 4-5:
Comb:02: {π
71
1
2π, π 2 π, π 3 ππ‘π₯}
Chapter 4
100 NRF HomoβBR HomoβBLOAD
Packet Load per unit of time(packet/s)
90 80 70 60 50 40 30 20 10 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(a) Load Distribution Per Corona 0.1 NRF HomoβBR HomoβBLOAD
Energy consumption per unit of time(W)
0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(b) Energy Consumption of Different Coronas Figure 4-6:
Comb:03: {π
72
1
ππ‘π₯, π 2 π, π 3 2π}
4.6. Performance Evaluation
Chapter 4
100 NRF HomoβBR HomoβBLOAD
Packet Load per unit of time(packet/s)
90 80 70 60 50 40 30 20 10 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(a) Load Distribution Per Corona 0.1 NRF HomoβBR HomoβBLOAD
Energy consumption per unit of time(W)
0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(b) Energy Consumption of Different Coronas Figure 4-7:
Comb:04: {π
73
1
π, π 2 ππ‘π₯, π 3 2π}
Chapter 4
100 NRF HomoβBR HomoβBLOAD
Packet Load per unit of time(packet/s)
90 80 70 60 50 40 30 20 10 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(a) Load Distribution Per Corona 0.1 NRF HomoβBR HomoβBLOAD
Energy consumption per unit of time(W)
0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(b) Energy Consumption of Different Coronas Figure 4-8:
Comb:05: {π
74
1
2π, π 2 ππ‘π₯, π 3 π}
4.6. Performance Evaluation
Chapter 4
100 NRF HomoβBR HomoβBLOAD
Packet Load per unit of time(packet/s)
90 80 70 60 50 40 30 20 10 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(a) Load Distribution Per Corona 0.1 NRF HomoβBR HomoβBLOAD
Energy consumption per unit of time(W)
0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
1
2
3
4
5 6 7 Corona nodes
8
9
10
(b) Energy Consumption of Different Coronas Figure 4-9:
Comb:06: {π
75
1
ππ‘π₯, π 2 2π, π 3 π}
Chapter 4 Table 4.4: Performance Tradeoffs
Protocols
Techniques
Parameters achieved
Cost to pay
NRF
Nominal Communication Range Forwarding {π}
Minimum Energy Consumption
Homo-BR
Variable Transmission Range Forwarding {π, 2π}
Hetero-BR
Variable Transmission Range Forwarding {π, 2π}
Stability Period
Maximum Energy Consumption
Homo-BLOAD
Mixed Transmissions Range Forwarding {π, 2π, ππ‘π₯}
Balanced Energy Consumption and Balanced Load Distribution
Energy Consumption Instability Period
Hetero-BLOAD
Mixed Transmissions Range Forwarding {π, 2π, ππ‘π₯}
Stability Period
Energy Consumption
Unbalanced Data Load and Unbalanced Energy Consumption
Data Load Balancing and Stability Period and Minimum Energy Consumption Unbalanced Energy Consumption
In fig. 4-10a, it is shown that FNDT of Homo-BLOAD scheme is 20s, which shows stability of the network.
From fig. 4-8a and fig. 4-8b, we can easily observe that
load distribution is in direct relation with energy consumption of sensor nodes. The stability period of Homo-BLOAD scheme is longer than other two schemes because the total data load is evenly distributed among all nodes in all coronas which results in balanced energy consumption. In Homo-BR scheme, data load is minimized at 1-hop neighbor nodes of sink because the 2-hop neighbor nodes of sink send maximum data load directly to sink using transmission range
2π.
We can also say that corona 2 pays the cost of maximum
energy consumption by minimizing the data load at corona 1 because sensor nodes of corona 2 are 1-hop neighbors of the sink using transmission range
2π.
Due to
maximum energy consumption of sensor nodes in corona 2, they die earlier than other sensor nodes. Thus, stability of Homo-BR scheme is 10s which is 10s less than HomoBLOAD scheme. The stability period of NRF scheme is shorter than Homo-BLOAD scheme because in NRF scheme, the energy consumption of sensor nodes in corona 1 is maximum due to non-uniform data load which causes premature death of sensor nodes in corona 1. FNDT of NRF scheme is 5s, which is 15s less than Homo-BLOAD scheme because the sensor nodes in corona 1 in NRF scheme die within no time due to unbalanced data load. FNDT of the Homo-BR scheme starts 10s before than Homo-
76
4.6. Performance Evaluation
Chapter 4
BLOAD scheme and ANDT ends at 90s.
While the ANDT of the Homo-BLOAD
scheme ends at 100s. Thus, ANDT of Homo-BR and Homo-BLOAD scheme is same because both schemes use variable transmission ranges
{π, 2π}.
The ANDT shows the
instability period and NRF scheme has less instability period than other two schemes as shown in fig. 4-10b. Lifetime of Homo-BLOAD scheme is better than the selected schemes because the balanced load distribution leads to balanced energy consumption of the network. We can observe that the network lifetime of Homo-BLOAD scheme is 100s while, the network lifetime of two selected schemes NRF and Homo-BR is almost 80s and 90s respectively. We can observe that 80% sensor nodes of the network die after 40s in NRF scheme. While, in Homo-BR scheme 50% nodes are still alive at 40s. When sensor nodes near the sink die, the previous coronas sensor nodes cannot report their data to the sink because both NRF and Homo-BR schemes uses transmission ranges
{π}
and
{π, 2π}
respectively. Homo-BLOAD scheme tackle this problem using
adjustable transmission power sensor nodes which can report directly to the sink using transmission range
{π, 2π, ππ‘π₯}
as well as hop-by-hop transmission i.e.
{π, 2π}
and all sensor nodes start direct transmission if no sensor nodes are available in the next hop corona. Therefore, network lifetime of the Homo-BLOAD scheme is longer than other two existing schemes.
4.6.0.1 Implementation of BR and BLOAD Schemes in Heterogeneous Environment FNDT and ANDT of Hetero-BR and Hetero-BLOAD schemes which are implemented in heterogeneous environment are shown in fig. 4-11a and fig. 4-11b.
Stability pe-
riod of Hetero-BR scheme is 15% better than Homo-BR scheme and it is 5% better than Homo-BLOAD scheme. Therefore, the network is stable upto 25s as shown in fig. 4-11b. The energy consumption of Hetero-BR scheme shown in fig. 4-12, is more than the Homo-BR and Homo-BLOAD schemes because all nodes are continuously forwarding and receiving data while in Homo-BR and Homo-BLOAD schemes the nodes near the sink die earlier due to which the nodes of the previous coronas do not forward data to the sink. Thus, the energy consumption of the Homo-BR and
77
Chapter 4
50 45 40
Number of Nodes
35 30 25 20 15 10
NRF HomoβBR HomoβBLOAD
5 0
0
20
40
60 Time(s)
80
100
120
(a) Lifetime of Nodes 50 NRF HomoβBR HomoβBLOAD
45 40
Number of Nodes
35 30 25 20 15 10 5 0
0
20
40
60 Time(s)
80
100
120
(b) Network Lifetime Figure 4-10:
FNDT and ANDT of Homo-BLOAD Scheme
78
4.7. Performance Tradeoffs
Chapter 4
Homo-BLOAD schemes is less than Hetero-BR scheme. The problem discussed above is solved by setting heterogeneous energy levels of each corona nodes. Therefore, we can observe a notable difference in stability of the network, which is high as compare to homogeneous schemes. The implementation of BLOAD scheme in heterogeneous and homogeneous environments is performed and residual energy of both schemes is shown in fig. 4-13a and fig. 4-13b.
It is observed that the energy consumption
of Homo-BLOAD scheme is higher than the Hetero-BLOAD scheme. Therefore, the sensor nodes with high energy consumption are assigned maximum energy level in Hetero-BLOAD scheme. The residual energy of Hetero-BLOAD scheme is maximum than Homo-BLOAD because of the different energy levels assigned to high energy consumption corona nodes. Hence, the scheme having dedicated deployment along with the heterogeneous energy is suitable for continuous monitoring applications in UWSNs.
4.7 Performance Tradeoffs The BLOAD scheme shows tradeoff between data load balancing and energy consumption of the network. We achieve load balancing by adjusting the transmission power level of sensor nodes. Nodes forward data to the sink directly as well as hopby-hop in BLOAD scheme. In order to avoid energy holes, we minimize the data load on 1-hop and 2-hop neighbors of the sink.
Load is distributed among all nodes to
minimize the energy consumption at nodes near the sink.
For load balancing, the
sensor nodes at long distance directly forward data to the sink using direct transmission range and deplete relatively high energy. Thus, the Homo-BLOAD routing protocol achieves balanced load distribution at the cost of high energy consumption as shown in table 4.4. Each sensor node in BLOAD scheme transmits data using variable transmission range. Therefore, energy consumption of the network is maximum in BLOAD scheme as compared to the existing schemes. Sensor nodes in Homo-BR scheme are out of energy in high energy consumption corona due to unbalanced load. In, Hetero-BR scheme, the nodes with high energy consumption are assigned high
79
Chapter 4 energy level and the nodes with low energy consumption are assigned low energy levels. Therefore, stability of the network is achieved by Hetero-BR scheme at the cost of high energy consumption.
4.8 Conclusion of the Chapter Energy balancing in UWSNs is one of the key requirements because of limited energy resources. In UWSNs, the sensor nodes consume high energy when there is an unbalanced load. We proposed a BLOAD scheme to balance load and to avoid of the energy holes in the network.
The BLOAD scheme is specifically designed to solve
the energy hole problem when a node does not find a forwarder node in the next corona in the way to sink. Previously, in NRF and BR schemes, nodes near the sink in corona 1 and corona 2 were out of energy because of unbalanced load and energy hole was formed near the sink and the network was totally disabled. At the end, most of the sensor nodes of the network which were far from the sink were alive and had maximum residual energy. In BLOAD scheme the discussed problem is solved and sensor nodes continuously report data to the sink, even nodes are out of energy in the next corona. The stability in the network is achieved by implementing sensor nodes in heterogeneous simulation environment. The results showed that BLOAD outperforms the existing schemes in terms of stability period and network lifetime. In future, we will work on detecting the energy holes in UWSNs using analytical modelling. The energy hole repair technique is also interesting in proactive and reactive modes of a network.
80
4.8. Conclusion of the Chapter
Chapter 4
50 45 40
Number of Nodes
35 30 25 20 15 HomoβBR HeteroβBR HomoβBLOAD HeteroβBLOAD
10 5 0
0
20
40
60 Time(s)
80
100
120
(a) Residual Energy of Homo-BLOAD Scheme 50 HomoβBR HeteroβBR HomoβBLOAD HeteroβBLOAD
45 40
Number of Nodes
35 30 25 20 15 10 5 0
0
20
40
60 Time(s)
80
100
120
(b) Residual Energy of Hetero-BLOAD Scheme Figure 4-11:
Residual Energy of Homo-BLOAD, Hetero-BLOAD and Hetero-BR
Schemes
81
Chapter 4
50 HomoβBR HeteroβBR HomoβBLOAD HeteroβBLOAD
Energy Consumption per unit of time(W)
45 40 35 30 25 20 15 10 5 0
0
10
20
30
40 50 Time(s)
60
70
80
90
Figure 4-12: Energy Consumption of Homo-BLOAD, Hetero-BLOAD and Hetero-BR Schemes
82
4.8. Conclusion of the Chapter
Chapter 4
25 Comb:01 Comb:02 Comb:03 Comb:04 Comb:05 Comb:06
Residual Energy(J)
20
15
10
5
0
0
50
100
150 Time(s)
200
250
300
(a) Residual Energy of Homo-BLOAD Scheme 25 Comb:01 Comb:02 Comb:03 Comb:04 Comb:05 Comb:06
Residual Energy(J)
20
15
10
5
0
0
50
100
150 Time(s)
200
250
300
(b) Residual Energy of Hetero-BLOAD Scheme Figure 4-13:
Residual Energy for All Possible Combinations of Transmission Ranges
and Load Weights of Both Homo-BLOAD and Hetero-BLOAD Schemes
83
References Promoting heterogeneity, mobility, and energy-aware voronoi diagram in wireless sensor networks. IEEE Transactions on Parallel and
[1] Ammari HM, Das SK.
Distributed Systems, 2008. [2] Zidi C, Bouabdallah F, Boutaba R.
Routing design avoiding energy holes in under-
water acoustic sensor networks. Wireless Communications and Mobile Computing,
2016.
Balance transmission mechanism in underwater acoustic sensor networks. International Journal of Distributed Sensor Networks, 2015.
[3] Cao J, Dou J, Dong S.
An Enhanced Energy Balanced Data Transmission Protocol for Underwater Acoustic Sensor Networks. Sensor, 2016.
[4] Javaid N, Shah M, Ahmad A, Imran M, Khan MI, Vasilakos AV.
On energy hole and coverage hole avoidance in underwater wireless sensor networks. IEEE Sensors
[5] Latif K, Javaid N, Ahmad A, Khan ZA, Alrajeh NA, Khan MI. Journal, 2016.
A New Node Deployment and Location Dispatch Algorithm for Underwater Sensor Networks. Sensors. 2016.
[6] Jiang P, Liu J, Ruan B, Jiang L, Wu F.
Energy balanced strategies for maximizing the lifetime of sparsely deployed underwater acoustic sensor networks. Sensors,
[7] Luo H, Guo Z, Wu K, Hong F, Feng Y. 2009.
Impacts of deployment strategies on localization performance in underwater acoustic sensor networks. IEEE Transactions
[8] Han G, Zhang C, Shu L, Rodrigues JJ. on Industrial Electronics, 2015. [9] Li Z, Yao N, Gao Q.
Relative distance based forwarding protocol for underwater
wireless networks. International Journal of Distributed Sensor Networks. 2014 Feb 11;2014.
End-to-end delay and energy efficient routing protocol for underwater wireless sensor networks. Wireless Personal Communications. 2014
[10] Ali T, Jung LT, Faye I. Nov 1;79(1):339-61.
84
REFERENCES
Chapter 4
A Localization Method for Underwater Wireless Sensor Networks Based on Mobility Prediction and Particle Swarm Optimization Algorithms. Sensors. 2016 Feb 6;16(2):212.
[11] Zhang Y, Liang J, Jiang S, Chen W.
A new multi-path routing protocol based on cluster for underwater acoustic sensor networks. Multimedia Technology (ICMT), International
[12] Liu G, Wei C.
Conference on. IEEE, 2011. [13] A. Sanchez, S. Blanc, P. Yuste, and J. Serrano.
A low cost and high efficient
acoustic modem for underwater sensor networks, in OCEANS, 2011 IEEE - Spain, pp. 1Γ’ΔΕ10, June 2011.
[14] Mari Carmen Domingo.
munication Networks
Overview of channel models for underwater wireless com-
Elsevier Journal on Physical Communication, vol. 1, issue
3, pp. 163- 182, 2008.
On the relationship between capacity and distance in an underwater acoustic communication channel. In Proceedings of the 1st ACM international
[15] Stojanovic, M.
workshop on Underwater networks, Los Angeles, CA, USA, 2006. [16] L.Berkhovskikh and Y.Lysanov.
Fundamentals of Ocean Acoustics
Springer, 1982.
85
New York:
Chapter 5 Conclusion
86
5.1. Conclusion
Chapter 5
5.1 Conclusion Energy efficient routing in UWSNs is important because it is difficult to recharge or replace sensor nodes with limited battery power in harsh underwater environment. Underwater wireless sensor nodes change position dynamically with the water current due to which some areas of the network field become sparse and some areas become dense.
Most commonly due to the high manufacturing cost, high design cost and
high deployment cost sensor nodes are sparsely deployed. An efficient protocol needs to be designed to deal with challenging issues of underwater communication such as high attenuation, low bandwidth, long propagation delay, noise etc. In order to achieve energy efficiency, load balancing and to control routing hole problem in sparse regions, we presented SEEC. Results showed that SEEC achieved energy efficiency and better network lifetime through static clustering in dense regions of the network. The division of the network field into subregions also control routing hole problem. The simulation results showed that network stability, lifetime and energy consumption of SEEC was better than other UWSNs routing protocols.
In addition, an energy
minimization routing protocol was designed for continuous monitoring application in UWSNs. In EMT, a forwarder node was selected in each sector on the basis of physical distance. In forwarder selection phase distance between sink and sensor nodes of a sector was calculated then a node with minimum distance from sink was selected as an eligible forwarder. Each node in a sector forwarded some fraction of data to a selected forwarder node and some fraction of data directly to sink to minimize load. Simulation results showed that EMT performs better than existing BR scheme, designed for continuous monitoring applications in UWSNs. Implementation of EMT in homogeneous and heterogeneous environments showed that Hetero-EMT performed better than Homo-EMT in terms of energy consumption and network lifetime. Apart from this, we proposed a BLOAD scheme to balance load and to avoid the energy holes in UWSNs. BLOAD was specifically designed to solve the energy hole problem when a node did not find a forwarder node in the next corona in the way to sink. Previously, in NRF and BR schemes, nodes near the sink in corona 1 and corona
87
Chapter 5 2 were out of energy because of unbalanced load and energy hole was formed near the sink and the network was totally disabled. At the end, most of the sensor nodes of the network which were far from the sink were alive and had maximum residual energy. In BLOAD, the discussed problem was solved and sensor nodes continuously report data to the sink, even nodes were out of energy in their next hop corona. The stability in the network is achieved by implementing sensor nodes in heterogeneous simulation environment. The results showed that BLOAD outperformed the existing schemes in terms of stability period and network lifetime.
88
