Coverage Optimization Strategy of Wireless Sensor Network Based On Swarm Intelligence Algorithm

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

2016 International Conference on Smart City and Systems Engineering

Coverage Optimization Strategy of Wireless Sensor Network Based on Swarm


Intelligence Algorithm
JunBo Xia
(LiaoNing JianZhu Vocational College; Liaoyang 111000,China)
xjb2016ieeecps@163.com

Abstract—When particle swarm optimization is applied in management. The simulation results show that the proposed
coverage optimization of WSN, its coverage rate is fast. algorithm can optimize the coverage working nodes, and can
However, this algorithm easily falls into local optimization and improve the real-time performance of the network node
it will cause premature phenomenon. Using artificial fish- scheduling.
swarm algorithm in coverage optimization of WSN has the
advantage of effective global search performance. However, II. MATHEMATICAL MODEL OF ETWORK COVERAGE
its convergence speed is slow in iteration and it is difficult to
find out the optimal solution. We integrate two algorithms to A. Network Coverage Model
be used in WSN coverage. Global coverage of artificial fish Supposing the monitoring region is a two-dimensional
swarm algorithm is used to search satisfactory solution
surface and N network nodes are randomly dispersed in this
domain. Then, particle swarm algorithm is adopted to perform
region. Node density is large enough in network with
rapid local search, adjust position and direction of WSN node,
and eliminate coverage overlay region and blind area. Thus, redundancy. In order to simplify, we assume that::
the network node distribution is reasonable with maximized Supposing the monitoring region is a two-dimensional
network coverage rate. Finally, artificial fish and improved surface and N network nodes are randomly dispersed in this
particle swarm hybrid coverage algorithm are used to region. Node density is large enough in network with
optimize the network. The simulation results prove that the redundancy. In order to simplify, we assume that:
improved hybrid algorithm can effectively improve the rate of a) Sensor network is a high density static network, that is,
network coverage in WSN. the node will not move after deployment.
b) Various nodes perceive to adopt Boolean perception
Keywords-AFSA; WSN; coverage; particle swarm; deploy model and its physical structure is homogeneous.
c) Various nodes have two status which are work and
I. INTRODUCTION dormancy. Network only contains one center processing
node with strong calculating ability. It can be used to switch
Coverage control of wireless sensor network is a basic wireless sensor network node status.
problem in wireless sensor network and it reflects perception d) All wireless sensor node positions are given.
service quality from sensor network. When wireless sensor The object of node adjustment is to select a group of
network is arranged in field or hostile environment, random optimizing nodes from lots of sensor nodes to keep network
dispersing modes such as air drop are usually adopted. This connection and obtain possibly large coverage rate. Existing
kind of sensing coverage performance exists randomness so studies prove that the network connection can always be
it needs to adopt coverage control strategy to ensure that guaranteed under precondition of full coverage when
wireless sensor nodes can effectively cover the monitoring communication radius between nodes is twice as much as
region. The coverage control of wireless sensor network can the perceiving radius. Thus, the problems in this paper can
be transformed into a restricting optimization. Its optimizing be illustrated as: there is a sensor node set
target is to minimize network energy consumption in unit
time under precondition of guaranteeing sensing coverage S  {si , i  1, 2,..., N } , whose work node number is
quality so as to extend network life. Artificial fish swarm minimized as much as possible when maximizing network
optimization algorithm (AFSA) is a swarm intelligence coverage requirement of one subset and work node set.
thought-based optimizing algorithm. This algorithm is based
on artificial intelligence method to construct artificial fish B. Artificial Fish Swarm Algorithm
and search the optimizing value during optimizing problems. AFSA is an optimizing algorithm to simulate fish swarm
It has insensitive initial value, global coverage, and highly behavior. It can effectively overcome that other intelligence
real time. Randomly deployed network nodes energy cannot algorithm easily falls into local extremum optimization. It
usually regenerate or supply. Thus, it is an urgent and is mainly through finding food, swarm and following
important problem in wireless sensing network on how to behavior for global optimal solution with fast convergence
guarantee enough converge monitoring region and extend speed and strong adaptive ability.
the network life. (1) Food searching behavior. Such step is moving step
In this paper, a new algorithm of particle swarm length value. To hypothesize current status of artificial fish
optimization combined with improved particle swarm as xi , and randomly select one status x j in its visible
optimization algorithm is proposed, which is based on the
algorithm of fish swarm optimization and energy domain, when food concentration of x j is larger than

978-1-5090-5530-2/16 $31.00 © 2016 IEEE 179


DOI 10.1109/ICSCSE.2016.60
current food concentration, it moves forward one step; xij (a, b, c)  xij (t )  av1 j (t  1)  bv2 j (t  1)  cv3 j (t  1)
otherwise, randomly select again. That is:
(1)
 1 The value of a, b, c can be 0 or 1, but not 0 at the same.
 xi (k  1)  xi k  || x  x || ( step)( x j  xi ), FC j  FCi
 j i (3) Compute corresponding fitness of all
 x (k  1)  x k  step, FC FC xij (a, b, c)  xij (t )  av1 j (t  1)  bv2 j (t  1)  cv3 j (t  1) ,

i i j i
and choose the maximum fitness value to assign the position
k is iteration number; xi and x j are status vector;
to xij (t  1) , that is, the new position of particle:
FCi and FC j denote the food concentration of xi and
xij (t  1)  max{ f ( xij a, b, c))} (2)
x j ; xi (k  1) denotes the status of artificial fish swarm The improved particle swarm algorithm process is:
individual i in ( k  1)th iteration z Particle swarm of population size as N is initialized
including randomly obtained initial position and speed of
(2) Swarm Behavior. If it indicates there are more food
particles.
in partner center and it is not crowded, center position of
z To calculate adaptive value of each particle in population,
partners, center position towards partners will move forward
it will be calculated according to fitness function.
one step or food finding behavior will not be performed.
z For each particle, its obtained adaptive value by
 Xc  Xi calculation is compared to historical optimal position. If
 xi|next  X i  Rand () Step , Yc / n f  Yi
 || X c  X i || it is good, this value is taken as currently optimal
position P.
foraging

,otherwise z Towards each particle, its adaptive value by calculation


(3) AF-Follow. X max is the partner which has the is compared to the optimal position Pg of population. If
biggest food concentration in visible region. If it is better, this value is taken as currently global optimal
FCmax  FCi , the food concentration at X max is high position Pg .
and it can be set as center location; otherwise, execute z Adjust the position and speed of current particle.
foraging. z Terminal condition of algorithm. If it reaches the
(4) Bulletin Board. It is mainly used to record the maximum iteration, it stops or return to the second step.
individual status of optimized artificial fish swarm and food B. Realization of Hybrid Algorithm Coverage
concentration of current individual status position in fish Optimization
swarm. Artificial fish will be compared to bulletin board
There are three stages of artificial fish swarm and the
after its one-time movement. If it is superior to bulletin
improved particle swarm hybrid optimization coverage
board, it will replace position in bulletin board.
algorithm. At first, network is initialized. Then, artificial fish
(5) Behavior Selection. For different problems, artificial
swarm algorithm is used to adjust sensor node position and
fish swarm individuals will evaluate environment so as to
direction. Each artificial fish stands for a sensor node
select corresponding behavior.
deployment. Finally, the optimization results of improved
III. FISH SWARM ALGORITHM-BASED OPTIMIZATION artificial fish swarm algorithm reaching the maximum
STRATEGY iteration are taken as input to endow the corresponding
particles in particle swarm to continue optimizing network
A. Improvement of Particle Swarm Algorithm with particle swarm algorithm. Each particle also stands for
a deployment method of sensor node. The processes of
The principle idea is to respectively calculate flying improved artificial fish swarm and particle swarm algorithm
speed of particles according to inertia, historical experience are:
and swarm experience. When position is updated, it will Step 1: Initialize sensor node position and perception
choose multi-position and the best position will be taken as direction in WSN and calculate initial coverage rate Po 0 in
new position of particles. Specific description is: network.
(1) To respectively calculate flying speed of particles
according to inertia, the historical experience and swarm Step 2: Randomly initialize artificial swarm number N
experience, speed updating formula of particles is: in feasible domain. The artificial fish vision is visual, the
maximum step of the largest testing times of each food
v1 j (t  1)  w vij (t ) finding is try number. Congestion factor is 8, and learning
 factor of particle swarm is c and. Initial iteration is set as
v2 j (t  1)  c1 r1 vij ( pij (t )  xij (t )) (1) n=0.
 Step 3: Set bulletin board. Compute each artificial fish

v3 j (t  1)  c2 r2 ( pgj (t )  xij (t )) fitness and compare it with bulletin board. If it is superior to
(2) The particle compute the alternative position of bulletin board, its current value will replace bulletin board
particle according to one or multiple speed, so the position value;
updating equation is

180
Step 4: Each artificial fish updates its position through wireless sensing network node deployment after artificial
food finding behavior, swarm behavior and following fish swarm algorithm optimization. Figure 3 depicts wireless
behavior; sensor network node deployment of improved artificial fish
Step 5: Check whether it reaches the largest iteration. If swarm and particles swarm hybrid algorithm.
step 5 reaches the pre-set iteration, iteration will stop. The
optimal value and optimal position will update in bulletin
board, turn to step 5 or step 2;
Step 6: The optimized result of artificial fish swarm
algorithm that reaches the maximum iteration is taking as
input to endow particles of corresponding particle swarm;
Step 7: The optimal value and optimal position in
bulletin board will be endowed to pbest and g best ;
Step 8: According to the fitness function, the fitness
value of each particle will be calculated, that is, network
coverage rate in monitoring domain;
Step 9: For each particle, its adaptive value by
calculation is compared to historical optimization position
Pt . If it is better, this value is taken as the best position P ;
(a) Deploy result of TRTF algorithm
Step 10: For each particle, its adaptive value after
calculation compares to the optimal position of population
Pg , if it is better, this value is taken as the currently best
global location;
Step 11: The position and speed of current particles are
respectively adjusted;
Step 12: Algorithm termination condition. If it reaches
the maximum iteration, then stop to output the optimal
solution, algorithm stops or return to step 2; Input
monitoring region size, discrete 4x , 4 y values, position of
sensor node in WSN, perception radius, main perception
direction and related parameters value. Output coverage
scheme and coverage rate of final sensor node.
The global optimization result of last-time algorithm is (b) Deploy result of AFSA algorithm
coverage rate after optimization. Corresponding particle
vector is the position and direction of sensor node. Finally,
coverage effect picture after optimization is finally drawn.
IV. SIMULATION EXPERIMENT AND PERFORMANCE
ANALYSIS

A. Deployment Performance Analysis


The coverage area of wireless sensor network is 20×20
square domain. The perceived radius of each sensor node is r
whose size is 2.5×25 and the nodes are randomly dispersed
in coverage area. The maximized iteration of algorithm in
this paper is set as 400 generation. During particles swarm
algorithm iteration, particle number is 30. Flying speed of (c) Deploy result of improved algorithm
particles is [3,3] , parameter 1c =0.9 ˈ2c =0.9. During Figure 1. Deploy result of three algorithms
artificial fish swarm algorithm iteration, artificial fish
number is set as 30, test times of artificial fish is 50 and The coverage rate of wireless sensing network in
vision scope is 1000.The nodes are randomly dispersing in standard PSO algorithm deployment is 83.75. The coverage
test area with 68.5% initial coverage rate. Figure 1, 2 and 3 rate of artificial fish swarm algorithm deployment is 80.25.
are simulating corresponding effect pictures. The coverage rate of artificial fish swarm and particle swarm
In these figures, the center point indicates sensor node hybrid algorithm deployment is 87.5007. The improved
position and circle refers to perceiving area of node. Figure 1 algorithm coverage rate improves 3.75 than standard PSO
depicts to wireless sensing network node deployment after algorithm. On this basis, in terms of coverage rate, artificial
standard PSO algorithm optimization. Figure 2 depicts fish swarm and particle swarm hybrid algorithm have more

181
advantages. From these figures and data, it can be
discovered that hybrid algorithm in this paper can obtain
better coverage rate, deployment result is more even with
few blind points. This is because artificial fish swarm
algorithm has the perfect ability to overcome local
extremum and obtain global extremum.
B. Time Cost Comparison
To compare the algorithm of this paper and other
intelligence algorithm coverage, sensor node quantity and
coverage scope are set as three conditions in table 1. These
three conditions consideration is mainly from setting
relationship between practical coverage scope and sensor (c) Case 3
Figure 2. Time consumption of three cases
nodes to set:
From this figure we can find, when node number
Table 1 Case of the number of different sensor nodes reaches 100 and coverage scope reaches 10%, the difference
Case setting Sensor node number n Coverage range among these four algorithms is not obvious. When node
A n 100 10m×10m
number is smaller than 1000 and coverage scope reaches
B 100  n 1000 50m×50m 25%, algorithm in this paper has advantages. When the node
C n  1000 100m×100m number is more than 1000 and coverage scope reaches 100%,
to compare between algorithm in this paper and four
algorithms, it has the perfect stability and the lest time
consumption.
V. CONCLUSIONS
Since sensor node of WSN has directionality and
perspective, omni-directional sensing model which is
originally adaptive to traditional sensor network cannot be
adapted to WSN. It needs to study new directional sensing
model and coverage optimizing algorithm which is adaptive
to directional perception model. Based on directional
perception model of WSN, this paper studies artificial fish
((a)) Case 1 swarm and particle swarm hybrid-based optimization
algorithm. This algorithm is used to realize the maximized
coverage in monitoring region. By simulating experiment of
analyzing algorithm expression of coverage effect under
different correlated factor restriction, it indicates that our
algorithm has better effect than traditional algorithm in the
improved network coverage rate
REFERENCES
[1] Mohanasundaram R.㧘 Periasamy P.S. Clustering Based Optimal
Data Storage Strategy Using Hybrid Swarm Intelligence in WSN㧘
Wireless Personal Communications, 2015㧘85:1381-1397
(b) Case 2 [2] Bruneo Dario, Scarpa Marco, Bobbio Andrea. Markovian agent
modeling swarm intelligence algorithms in wireless sensor networks,
Performance Evaluation, 2012, 69:135-149
[3] XueHan, Ma, Hong Xu. Swarm intelligence based dynamic obstacle
avoidance for mobile robots under unknown environment using
WSN, Journal of Central South University of Technology (English
Edition), 2008, 15:860-868
[4] Bala Krishna M., Doja M.N. Swarm intelligence-based topology
maintenance protocol for wireless sensor networks, IET Wireless
Sensor Systems, 2011, 1:181-190
[5] Mo Yuanbin, Liu Jizhong, Wang Baolei. A novel swarm intelligence
algorithm and its application in solving wireless sensor networks
coverage problems. Journal of Networks, 2012, 7:2037-2043

182

You might also like