MULTI-OBJECTIVE ENERGY EFFICIENT OPTIMIZATION ALGORITHM FOR COVERAGE CONTROL IN WIRELESS SENSOR NETWORKS

International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol.3,No.4,August 2013
DOI : 10.5121/ijcseit.2013.3403 25
MULTI-OBJECTIVE ENERGY EFFICIENT
OPTIMIZATION ALGORITHM FOR COVERAGE
CONTROL IN WIRELESS SENSOR NETWORKS
Seyed Mahdi Jameii1
and Seyed Mohsen Jameii2
1
Department of Computer Engineering, Shahr-e-Qods Branch, Islamic Azad University,
Tehran, Iran.
Jamei@Qodsiau.ac.ir
2
Department of Computer Engineering, Qazvin Branch, Islamic Azad University,
Qazvin, Iran.
Jamei@Ariapardazesh.ir
ABSTARCT:
Many studies have been done in the area of Wireless Sensor Networks (WSNs) in recent years. In this kind
of networks, some of the key objectives that need to be satisfied are area coverage, number of active
sensors and energy consumed by nodes. In this paper, we propose a NSGA-II based multi-objective
algorithm for optimizing all of these objectives simultaneously. The efficiency of our algorithm is
demonstrated in the simulation results. This efficiency can be shown as finding the optimal balance point
among the maximum coverage rate, the least energy consumption, and the minimum number of active
nodes while maintaining the connectivity of the network.
KEYWORS
Wireless Sensor Networks, Multi-objective Optimization, Coverage, Lifetime
1. INTRODUCTION
Wireless Sensor Networks (WSNs) are very suited for doing the surveillance tasks. The
processing and wireless communication capabilities and battery power of each sensor in this kind
of networks are limited and replacing the battery of nodes is impossible in applications such as
habitat monitoring and monitoring civil structures.[1-2]
Coverage is a key problem in WSNs and it focuses on determining the portion of the field that is
monitored by active nodes [3-7].
For deployment of sensor nodes some of the key objectives that need to be satisfied are

26
The portion of covered area, the number of active nodes, energy consumed by nodes, and network
connectivity are key objectives in the area of WSNs. Selecting the optimal set of active nodes has
been proved as an NP-complete problem in [8].
In the real world, Optimization Problems (OP) is usually with multiple attributes. Commonly,
multiple objectives should be optimized simultaneously; however, there exists conflicts among
the multiple objectives. For example, product quality and cost are two conflicting objectives in
the production activity. In order to achieve the total optimization, some conflicting objectives
should be compromised [9]. Some good algorithms have been put forward such as NSGA-II[10],
PESA [11], PAES [12], SPEA2 [13] ,etc. NSGA has better diversity and faster convergence in
solutions.
In this paper, we propose a NSGA-II based multi-objective algorithm for optimizing all of these
objectives simultaneously. The efficiency of our algorithm is demonstrated in the simulation
results. This efficiency can be shown as finding the optimal balance point among the maximum
coverage rate, the least energy consumption, and the minimum number of active nodes while
maintaining the connectivity of the network. The remaining of this paper is organized as follows:
In Section 2 we present the related work related to coverage in WSNs. In Section 3 we introduce
the NSGA-II algorithm briefly. Section 4 describes the proposed algorithm. Simulation results are
shown in section 5 and the proposed algorithm is evaluated in this section. The paper concludes
with Section presents some 6.
2. RELATED WORKS
Maximizing the coverage and lifetime objectives individually was the main focus of several
studies in the past. Although coverage is the key objective in WSNs but for better efficiency it
should not be optimized separately. The proposed approaches in [14-17] optimize the lifetime and
coverage objectives individually and sequentially, or by constraining one and optimizing the
other. This often results in ignoring and losing “better” solutions since WSN coverage and
lifetime are conflicting objectives [18]. Therefore, there is not a single solution to maximize both
objectives simultaneously and a decision maker [19] needs an optimal trade-off of candidate
solutions.
In a Multi-objective Optimization Problem (MOP), a candidate trade-off solution is often called
non-dominated or Pareto optimal. The set of all Pareto optimal or non-dominated solutions in the
search space, also called Pareto Set (PS), is often mapped to a Pareto Front (PF) in the objective
space [20]. Multi-objective Evolutionary Algorithms (MOEAs) could obtain such an approximate
PF in a single run. This is mainly due to the fact that MOEAs accommodate different forms of
operators to iteratively generate a population of solutions. In the literature, several general
purpose MOEA frame- works are used for dealing with MOPs in WSNs [21–24] such as the Non-
dominated Sorting Genetic Algorithm-II [25] (NSGA-II).
3. NON-DOMINATED SORTING GENETIC ALGORITHM-II
NSGA-II [10] has been demonstrated as one of the most efficient multi-objective optimization
algorithms. Pareto optimality is an integral part of NSGA-II and will be introduced first.

27
3.1. Pareto Concepts
Multi-objective optimization can be expressed as (1):
Min fi (x), i = 1, 2, . . . , m x∈ x
(1)
where fi(x) denotes the ith
objective function, m is the number of objectives and x represents the
feasible search space.
Definition 1: A solution x1 is said to dominate x2 (denoted by x1 ‹ x2) if and only if:
∀i ∈ {1, 2 . . .m} : fi (x1) ≤ fi (x2) ^ ∃j ∈ {1, 2 . . .m} : fj (x1) , fj (x2)
(2)
Definition 2: For S = {xi, I = 1,…., n}, solution x is said to be a non-dominated solution (Pareto
solution) of set S if x∈S and there is no solution x′∈ S for which x′ dominates x.
Definition 3: Assume that set P contains all the non-dominated solutions of S, then
PF= {v | v= [ f1(x), f2(x). . .fm(x)]T
, x ∈ P} is a Pareto front of set S.
3.2. Fitness Assignment Schemes in NSGA-II
In the fitness assignment procedure, NSGA-II allocates a rank value ri to each solution. The non-
dominated solutions are identified and assigned the rank value 1. After removing those solutions
from the population, new non-dominated solutions are assigned rank value 2. This procedure
continues iteratively.
Fig. 1 provides a graphic example. It represents rank values for a population (size 10). First, the
non-dominated solutions 1, 2 and 3 receive rank value 1, then solutions 4, 5 and 6 receive rank
value 2 and the procedure continues.
To promote the solutions in the sparse region, crowding distance Di is assigned to each candidate
solution. Di is the average distance of two points on either side of the solution i along each of the
objectives. For objective dimension two, Di for solution i is determined by a rectangle formed by
two nearest neighbours of i, as shown in Fig. 1. For solution 2, solutions 1 and 3 are the two
neighbours in the same rank, defining the boundary rectangle. D2 would be the average side
length of the rectangle. With assigned ri and Di, any two solutions in the population can be
compared by solution i is superior than solution j ⇔ {ri < rj} or {ri = rj and Di > Dj } (3)

28
Figure 1. Fitness assignment of NSGA-II in a two-objective Space
3.3. Procedure of NSGA-II
In any generation of NSGA-II, there are two steps: evolving and filtering. An evolving process
generates the temporary new population
( + 1)
1
from S(k)
by applying the genetic operators.
1. Coding: A real-coding scheme is adopted because of difficulties of binary representation
when dealing with continuous search space with large dimensions. A decision variable is
represented by a real number within its lower limit and upper limit. Since parts of
decision variables are discrete, a procedure is imposed to round discrete variables of
newly generated solutions to the nearest valid value.
2. Crossover and mutation: A blend of crossover- α operator and normally distributed
mutation operator [26] is employed for the real-coding scheme.
After genetic operators, the filtering procedure combines S(k)
and
( + 1)
1
, and then chooses
solutions by applying (3) to form the new population S(k+1)
.
4. PROPOSED ALGORITHM
In this section, the details of the proposed algorithm are described. At first, we made some
assumptions: the nodes are deployed randomly, each one are static and knows its own location
using some location systems [27]. In the proposed algorithm, such as [28], the transmission radii
of sensors are assumed to be at least twice the sensing radii for assuring the connectivity of the
network.
We introduce a cluster-based optimization scheme which is scheduled into rounds. In each round,
firstly, the target area is divided into several clusters. The LEACH [29] algorithm is used for
clustering and selecting the cluster heads. The cluster-head has full control of its cluster and run
the NSGA-II algorithm for optimizing the following objectives subject to the connectivity
constrain:

29
Objective 1: Maximizing the network coverage:
Max f1(x)= Acoverd / A
(4)
where Acovered is the covered area by the active sensors and A is the whole area of the sensor field.
Objective 2: Number of active sensors that is desirable to be minimize, so can be converted to the
objective for maximization as follow:
Max f2(x)= 1- |K′|/ |K|
(5)
In this equation, |K′| is the number of active nodes and |K| is the number of all nodes.
We have used a bit string with size K for representing the solution. For each sensor node 1 –bit is
assigned in the solution and this bit represents the working state of corresponding node as (6):
x = (x1, x2,…, xi,…, xK)
xi =
1, ℎ
0 ℎ
(6)
In fig.2, the flowchart of the proposed algorithm is shown. The recombination operator used in
this paper is two-point crossover, which is a typical recombination for binary or other string-like
chromosomes, and the crossing points are selected at random. The mutation operator is applied
for each new generated child after crossover. It works by complementing some genes in the
child's chromosome randomly. The mutation operator swaps the bits of each string (0 becomes 1
and vice versa) means that a sleep sensor node becomes active and vice versa.
After a new population has been produced through the genetic operators, selection is done in an
extended space composed of all parent and offspring individuals. This extended sampling space
allows large probability of mutation and crossover while keeping the population relatively stable.
Assign each individual having two fitness functions (coverage rate and number of active sensors),
by introducing the non-dominated sorting, crowded distance operator and elitism. Selecting the
individuals as a parent for producing the next generation is proportional to its fitness value.
Each time there are two solutions of different non-domination ranks, we prefer the higher one. If
there are two solutions with the same non-domination ranks, we prefer the one which has larger
crowded distance. Also the elitism mechanism is used in our algorithm to prevent destroying the
best individual of each generation by the crossover and mutation operators during the evolution
process. This means that the current best individual at each generation of the algorithm can be
easily transferred to the next generation.

30
Report final
population & stop
start
Initialize parent
population of size K
Evaluate objective
functions
Non-dominated sorting
Tournament selection
Crossover & Mutation
Assign high values of
objective functions
Evaluate objective
functions
Combine parent &
child populations, Non-
dominated sorting
Select best N solutions from
the combined population
through Tournament selection
Connectivity
constrain is
satisfied
Stopping
criteria met?
Assign high values of
objective functions
Connectivity
constrain is
satisfied
Yes
Yes
Yes
No
No
No
Figure 2. The flowchart for the proposed algorithm

31
5. SIMULATION RESULT
In this section, our proposed algorithm is simulated using NS-2 simulator [30]. To evaluate the
proposed protocol, the coverage algorithm is implemented using single objective genetic
algorithm (SGA) and the proposed algorithm is compared with it. In the simulations, we assume a
target area with a size of 150×150 m2
. We deployed the sensor nodes randomly in the target area.
The number of nodes, K, in the first experiment are considered as 100, 150, 200, 250, 300, 350,
400, 450, 500 respectively. As shown in Fig. 3 and 4, due to the proposed algorithm accuracy in
selecting the active sensor set, it is able to provide the full coverage in sparse deployment. Also,
with increasing the nodes density, the proposed protocol decreases the energy consumption of the
sensor networks.
In the last experiment, the number of nodes is considered as 100 and coverage rate with different
number of working nodes are depicted in Fig. 5. As can be seen in this figure, with the same
number of working nodes, the proposed algorithm can achieve higher coverage rate compared to
the SGA algorithm. This is because of in the proposed algorithm both of the coverage rate and
number of working sensor are considered simultaneously as objectives.
Figure 3. Coverage Rate of Sensor Set in Different Configuration
Figure 4. The Energy Consumption per Area in Different Configurations
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
100 150 200 250 300 350 400 450 500
TheEnergyConsumptionperArea
Number of Nodes Deployed
SGA
Proposed Algorithm
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
100 150 200 250 300 350 400 450 500
CoverageRate
Number of Nodes Deployed
SGA
Proposed Algorithm

32
Figure 5. Coverage Rate Vs. Number of Working Sensors
6. CONCLUSION:
In this paper we proposed a NSGA-II based multi-objective algorithm for optimizing the area
coverage, number of active sensors and energy consumed by nodes in wireless Sensor networks
while maintaining the connectivity simultaneously. For evaluating the proposed protocol, the
coverage algorithm is implemented using Single objective Genetic Algorithm (SGA) and the
proposed algorithm is compared with it. As shown in the simulation results, many more non-
dominated solutions are found in the proposed algorithm and these solutions are better than the
solutions obtained by the SGA algorithm.
ACKNOWLEDGEMENTS
The authors wish to thank Islamic Azad University, Shahr-e-Qods branch for supporting this
work through grants.
REFERENCES
[1] E. Shih, S. Cho, N. Ickes, R. Min, A. Sinha, A. Wang, A. handrakasan, Physical layer driven protocol
and algorithm design for energy-efficient wireless sensor networks, in: Proc. of the 7th Annual
International Conference on Mobile Computing and Networking, Rome, Italy, pp. 272_287, 2001.
[2] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, A survey on sensor networks, IEEE
Communications Magazine 40, pp. 102_114, 2002.
[3] S. Meguerdichian, F. Koushanfar, M. Potkonjak, and M. Srivastava, “Coverage Problems in Wireless
Ad Hoc Sensor Networks,” Proc. IEEE INFOCOM, 2001.
[4] X. Wang, G. Xing, Y. Zhang, C. Lu, R. Pless, and C. Gill, “Integrated Coverage and Connectivity
Configuration in Wireless Sensor Networks,” Proc. ACM First Int’l Conf. Embedded Networked
Sensor Systems (SenSys), 2003.
[5] X. Bai, S. Kumar, D. Xuan, Z. Yun, and T.-H. Lai, “Deploying Wireless Sensors to Achieve both
Coverage and Connectivity,” Proc. ACM Int’l Symp. Mobile Ad Hoc Networking and Computing
(MobiHoc), 2006.
[6] H. Zhang and J. Hou, “Maintaining Sensing Coverage and Connectivity in Large Sensor Networks,”
J. Ad Hoc and Sensor Wireless Networks, vol. 1, pp. 89-124, 2005.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
20 25 30 35 40 45 50 55 60
CoverageRate
Numberof Working Sensors
Proposed Algorithm
SGA

33
[7] G. Kasbekar, Y. Bejerano, and S. Sarkar, “Lifetime and Coverage Guarantees through Distributed
Coordinate-Free Sensor Activation,” Proc. ACM MobiCom, 2009.
[8] M. Cardei and J. Wu, “Energy-efficient coverage problems in wireless ad-hoc sensor networks,”
Computer Communications, vol. 29, no. 4, pp. 413–420, 2006.
[9] Cui xunxu, Multiobjective Evolutionary Algorithms and their Applications, Beijing:National Defense
Industry Press, pp10-15, 2006.
[10] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multi-objective genetic algorithm:
NSGA-II, IEEE Transactions on Evolutionary Computation, vol. 6, pp. 182_197, 2002.
[11] Come D W. “The parate envelope-based selection algorithm for multiobjective optim ization,”
Lecture Notes in Computer Science, pp839-848, 2000.
[12] Joshua D. Knowles and David W Corne, “The pareto Archived evolution strategy: a new baseline
algorithm for multiobjective optimization,” In 1999 Congress on Evolutionary Computation.
Washington.D.C, pp98-105, 2000.
[13] E.Zitzler, M.Laumanns, L.Thiele. SPEA2, “Improving the strength pareto evolutionary algorithm,”
Technical Report TIK-Report 103, Swiss Federal Institute of Technology Zurich (ETH).May,2001.
[14] S. Meguerdichian, F. Koushanfar, M. Potkonjak, M.B. Srivastava, Coverage problems in wireless
ad-hoc sensor networks, IEEE INFOCOM, vol. 3, pp. 1380–1387, 2003.
[15] P. Cheng, C.N. Chuah, X. Liu, Energy aware node placement in wireless sensor networks, IEEE
Global Telecommunications Conference vol. 5, 3210–3214, 2003.
[16] X. Liu, P. Mohaparta, On the deployment of wireless data back-haul networks, IEEE Transactions on
Wireless Communication vol. 6, No 4, pp. 1426–1435, 2007.
[17] Y. Chen, C.-N. Chuah, Q. Zhao, Network configuration for optimal utilization efficiency of wireless
sensor networks, Elsevier Ad Hoc Networks Vol. 6, pp. 92–107, 2008.
[18] D.B. Jourdan, O.L. de Weck, Layout optimization for a wireless sensor network using amulti-
objective genetic algorithm, IEEE Semiannual Vehicular Technol- ogy 5, pp.2466–2470, 2004.
[19] S. Chaudhuri, K. Deb, An interactive evolutionary multi-objective optimization and decision making
procedure, Applied Soft Computing, Vol.10, No.2, pp.496–511, 2010.
[20] K. Deb, Multi-objective Optimization Using Evolutionary Algorithms, Wiley and Sons, 2002.
[21] R. Rajagopalan, P.K. Varshney, C.K. Mohan, K.G. Mehrotra, Sensor placement for energy efficient
target detection in wireless sensor networks: a multi-objective optimization approach, in: Conference
on Information Sciences and Systems, Baltimore, Maryland, 2005.
[22] S.C. Oh, C.H. Tan, F.W. Kong, Y. S. Tan, K.H. Ng, G.W. Ng, K. Tai, Multiobjec- tive optimization
of sensor network deployment by a genetic algorithm, IEEE Congress on Evolutionary Computation,
pp.3917–3921, 2007.
[23] J. Jia, J. Chen, G. Chang, Z. Tan, Energy efficient coverage control in wireless sensor networks based
on multi-objective genetic algorithm, Computers and Mathematics with Applications 57 (11–12), pp.
1756–1766, 2007.
[24] J. Jia, J. Chen, G. Chang, Y. Wen, J. Song, Multi-objective optimization for coverage control in
wireless sensor network with adjustable sensing radius, Computers and Mathematics with
Applications 57 (11–12), pp.1767–1775, 2009.
[25] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multiobjective genetic algorithm:
NSGA II, IEEE Transactions on Evolutionary Computation Vol.6, No. 2, pp.182–197, 2006.
[26] ABRAHAM A., JAIN L., GOLDBERG R.: ‘Evolutionary multiobjective optimization: theoretical
advances and applications’, 2005.
[27] N. Bulusu, J. Heidemann, D. Estrin, GPS-less low-cost outdoor localization for very small devices,
IEEE Personal Communications 7, pp.28_34, 2000.
[28] H. Zhang, J.C. Hou, Maintaining sensing coverage and connectivity in large sensor networks, Ad-hoc
and Sensor Wireless Networks Vol. 1, pp.89_124, 2005.
[29] W. Heinzelman, A. Chandrakasan and H. Balakrishnan, Energy-Efficient Communication Protocol
for Wireless Microsensor Networks, Proceedings of the 33rd Hawaii International Conference on
System Sciences (HICSS '00), January 2000.
[30] The Network Simulator - NS-2. http://www.isi.edu/nsnam/ns/.

MULTI-OBJECTIVE ENERGY EFFICIENT OPTIMIZATION ALGORITHM FOR COVERAGE CONTROL IN WIRELESS SENSOR NETWORKS

Recommended

More Related Content

What's hot (17)

Viewers also liked (17)

Similar to MULTI-OBJECTIVE ENERGY EFFICIENT OPTIMIZATION ALGORITHM FOR COVERAGE CONTROL IN WIRELESS SENSOR NETWORKS (20)

Recently uploaded (20)

MULTI-OBJECTIVE ENERGY EFFICIENT OPTIMIZATION ALGORITHM FOR COVERAGE CONTROL IN WIRELESS SENSOR NETWORKS