TMCnet News
A Kind of Wireless Sensor Network Coverage Optimization Algorithm Based on Genetic PSO [Sensors & Transducers (Canada)](Sensors & Transducers (Canada) Via Acquire Media NewsEdge) Abstract: Aiming to the problems of slow convergence speed and being easily early-maturing etc. in existing network based on standard particle swarm algorithm, this paper proposes a wireless sensor network coverage optimization algorithm based on the genetic particle swarm optimization (PSO). The wireless sensor maximum coverage is regarded as the objective function, the genetic algorithm with adaptive crossover and mutation factors is used to search the solution space, and the powerful global search ability of PSO is used to increase search scope to make particle covering more efficient, which both strengthen algorithm optimization ability, improve the nodes coverage, and solve early-mature problem. Comparing with the standard traditional genetic algorithm and new quantum genetic algorithm, simulation results present that the rate of coverage in this algorithm increases by 2.28 % and 0.65 % respectively; and convergence speed is also improved, therefore this method can effectively realize the wireless sensor network coverage optimization. Copyright © 2013 IFSA. Keywords: Wireless sensor network, Coverage optimization, POS, Genetic algorithm. (ProQuest: ... denotes formulae omitted.) 1. Introduction Wireless sensor network is composed of lots of sensor nodes placed in the detection areas, which adopts wireless communication mode to do data communication. In the network all kinds of sensor nodes can collaboratively sensing, collect and analyze the information of related objects in the network coverage area. Covering problem is the key problem of wireless sensor network, the distribution area of sensor nodes determines the effectiveness of network detection. High effective network coverage can ensure the space source of wireless sensor network to get the reasonable distribution [1-2]. At present, scholars have developed a variety of algorithms to optimize sensor network coverage, including distributed algorithm, approximate minimum node set algorithm, set covering algorithm, genetic algorithm, and standard particle swarm algorithm etc. [3-4]. These algorithms all have corresponding defects, for example, the distributed algorithm has a longer network cycle, but it can only solve the problem of unit circle, it can not solve the problem of irregular geometric area, there is certain limitation. Approximate minimum node set algorithm fails to fully consider node distribution, which can not ensure to realize the optimal sensor network coverage. Set covering algorithm operates the coverage through the relationship around central node and surrounding node, there is the unbalanced problem of node load distribution. Genetic algorithm finds the optimal solution through the parallel search method, but its convergence speed is slow, which cannot collect the extract dynamic node on time. Standard particle swarm algorithm searches optimal solution in advance in the detection area, which results the particle search area reduced and node network coverage dropped [5-6]. In order to solve the problems existing in the traditional algorithms, this paper presents a wireless sensor network coverage optimization algorithm based on genetic PSO. The wireless sensor maximum coverage is regarded as the objective function, the genetic algorithm with adaptive crossover and mutation factors is used to search the solution space, and the powerful global search ability of PSO is used to increase search scope to make particle coverage more efficient, which both strengthen algorithm optimization ability, improve the nodes coverage rate, and solve early-mature problem. 2. The Principle of Wireless Sensor Network Coverage 2.1. Construction of Wireless Sensor Network To get a higher sensor nodes coverage rate in WSN, to reduce the number of blind spot of the perception, the density of sensor nodes deployment needs to increase. However, if sensor nodes deployed density is too large, which will produce a large number of redundant nodes to lead to data transmission conflict, the energy will be work out too early, so as to the WSN lifecycle becomes too short. For a sensor node, if its sensing area can be completely covered by the other sensor nodes, and then the node is called redundant nodes, under the status of other nodes is working, the sensor node can enter a dormant state. The redundant nodes are judged through the corresponding algorithm to make the work nodes minimum as far as possible, in order to improve the network coverage. Therefore, in wireless sensor network initialization phase, the choice of work nodes number and node energy consumption need to be considered at the same time, it is a contradiction between WSN coverage and the number of working nodes [7] Wireless sensor nodes have their own specific sensing radius, the detection area within the sensing radius is known as coverage area, the node area outside sensor radius is called blind area. The goal of constructing wireless sensor network is: reasonable allocation of network resources, maximizing the network coverage [8-9]. The physical model of wireless sensor network node coverage is described in Fig. 1. As shown in Fig. 1, node coverage area in wireless sensor network is all the information collection within sensor area that takes node Z as the center of circle and Rz as the radius. The distance between any two nodes RE and RF in communication coverage area should be kept in the diffuse area of RH, to ensure the sensing information from two nodes can effectively communicate. Connected coverage refers to the neighbor nodes of node Z arrange according to the special method to ensure being inducted by sensor nodes [10]. 2.2. The Selection of Coverage Optimization Goal The ultimate goal of wireless sensor network (WSN) is to maximize network coverage. In case to ensure the successful completion of network communication, how to place sensor nodes and using the minimum network nodes to realize maximizing network coverage are the main purposes. Two indicators of sensor node coverage and area coverage are used to judge the coverage of wireless sensor network. 2.2.1. Node Coverage The waiting test area is represented in W, after it is processed by digital discretization, it includes p* q pixels. In W there are q sensor nodes with the same parameters, and the coordinates of each sensor node are known as (x¿, yi} , induction radius is r, the communication radius R = 2r. Wi = {xi,yiir} represents the circle with a center of circle of sensor node coordinate {Xp^.} and the radius r. The pixel points are set as (x, y), the distance between target pixels and sensor node is expressed with h(wi, k) = ^(x, - x)2 + (yi - y)2 , K {ri} represents the coverage probability of pixels (x,j) covered by sensor node Wi. Sensor node detection model can present a certain probability distribution under the influence of the environment, which is as follows: ... (1) In the expression, ru (0 < ru < r) represents the effectiveness parameter tested by sensor nodes, a1,a2, ß1, ß2 respectively represent the testing parameters related with sensor node characteristics, and 01, 02 represents the input parameters, and there are: ... (2) ... (3) Multiple sensor nodes synergistically measure target, which can improve the probability of target measurement, the probability of synergistical measurement is as follows: ... (4) 2.2.2. Area Coverage Rate Monitoring area W contains p*q pixels, a node set comprehensive detection probability Kwpg ( Wpg ) can analyze if each pixel is overwritten, and the area coverage of node set W is set as follows: ... (5) 3. Genetic PSO Optimization Nodes Coverage Algorithm 3.1. Improvement of Traditional Genetic Algorithm Genetic operation contains selection, cross-over, and mutation three kinds of operators, there into, crossover and mutation operators play an important role to genetic algorithm. Cross can extract the effective structure of two parents to construct a new individual; mutation can adjust some individual gene value in the set. Integration of the crossover and mutation operators can enhance the search ability of genetic algorithm, the crossover operator is taken as the main operator, and mutation operator is taken as a secondary operator. The traditional genetic algorithm mainly does the iteration calculation according to the fixed crossover and mutation probability, when the genetic operator is small, it will cause algorithm convergence speed too fast or too slow; when genetic operator is too big, it will lead to the algorithm can not converge. In order to solve the disadvantages of traditional algorithm, the genetic algorithm with an adaptive operator makes the individual can adaptively adjust crossover and mutation probability. Decreasing the crossover probability of parent individual with high similarity, and increasing the crossover probability of parent individuals with low similarity, can enhance cross processing accuracy to complete the global search; at the same time, the individuals with high fitness use low crossover probability, the individuals with low fitness use high crossover probability, which can ensure the diversity of set and increase the convergence speed. The operation method of adaptive crossover and mutation operator is: ... (6) ... (7) Among them, the crossover probability of the i - th and the j -th parent in the q -th generation is described as wt j q, the degree of approximation of the i -th and the j -th parent is described with z;. ., the mutation probability of the i -th individual in the q -th generation is described with pi q, the fitness of the i -th individual in the q -th generation is described with gt q . 3.2. PSO Genetic Optimization Algorithm In the process of node coverage, genetic algorithm has certain disadvantages, in order to overcome these disadvantages, PSO genetic algorithm is introduced. This algorithm combines the PSO particle swarm global search capability and the local optimization function of genetic algorithm, has strong nonlinear mapping ability. PSO algorithm sets up a variety of particles in the space dimensions, the best particle coordinates is the optimal solution of objective function. Based on the current coordinate values, particles constantly adjust their speed and coordinate. In this paper, the spatial dimension is composed of node coverage and area coverage, including two groups of the weight coefficient and two dimensional adaptive values, the weight coefficient of each group is integrated into the model of genetic algorithm and does the readjustment and optimization. In this paper, the algorithm sets up 60 search particles, the solution of searching optimal objective function is the spatial dimension coordinates of each particle. If the position of i -th particle is ... (8) The speed of particles searching optimal in space is ... (9) The optimal position of particle swarm itself is Ki = (Kn,Ki2,...,Ki60), the optimal position of whole particles swarm is L, = (Ln,Li2,Lm) . The particle swarm using expression (10) to adjust the position: ... (10) In the expressions, the first half is used to adjust the speed of particles; the second half represents the self learning process of particles. The expression can optimize the cognizing energy of particles to avoid algorithm appearing "early-mature" phenomenon. The particle swarm's self-learning ability can improve particle swarm global optimization ability. Through the above formulas the particles can improve the effectiveness and efficiency of stage coverage optimization. 4. Simulation Experiments and Analysis 4.1. Simulation Environment Settings Simulation experiment set 50 wireless sensor nodes in a circular measurement range with radius 10 M, the sensing area is the circular area with r = 4m radius, and the communication coverage area of sensor node is the circular area with the radius U = 2r = Sm [9]. The parameters of sensor node detection model in the experiment are ux=u2=2, model checking validity parameters are ru = 0.5m = 2m, different nodes feature parameters respectively are ax = 1, a2 = 0, ßx = 1, ß2 = 0.5, number^ = 50. The experiment uses the computer with main frequency 2.6 GHz under the condition of Matlab to simulation test about wireless sensor network coverage problem. 4.2. Contrastive Analysis on the Performance of This Algorithm and Traditional Algorithm In order to verify the advantages of wireless sensor network coverage optimization algorithm based on genetic PSO proposed in this paper, through the simulation results to compare with the traditional standard particle swarm algorithm (POS) and the method in this paper, the covering performance test data of two algorithms can be described in Table 1. According to the experimental test data in Table 1, the coverage rate of this algorithm and traditional algorithm, as well as the rectangular figure of iteration time variation trend along with the change of node sensing radius, are respectively described in Fig. 2 and Fig. 3. From diagram 2, coverage and sensing radius have the positive correlation, the greater the radius is, the greater the coverage is, and the coverage rate in this algorithm is higher than the traditional genetic algorithm. When the sensing radius is lesser, the coverage rectangular figure shows notable rising trend; when the sensing radius increases, the variation trend of coverage rate rectangle diagram is relatively stable, when the sensing radius reaches a certain value to achieve overall coverage, the coverage rate is 100 %. From the analysis of diagram 3, the iteration number and the sensing radius have the negative correlation, when the sensing radius is larger, the iteration time is smaller, and the iteration number in this algorithm is less than the traditional genetic algorithm. When the sensing radius is small, the variation trend of iteration time rectangle diagram is obvious, the convergence speed of two algorithms has great improvement; when sensing radius increases, the iteration time change is relatively smooth, and the algorithm convergence speed gradually decreases. Finally, through Fig. 2 and Fig. 3, the algorithm in this paper has high coverage than traditional standard particle swarm algorithm, the number of iteration is low, which illustrates the optimal performance of this algorithm is stronger, and the coverage performance is better. 4.3. Performance Comparison with Other Wireless Sensors Coverage Algorithm In order to further verify the superiority of this algorithm, this paper respectively adopts this algorithm, traditional genetic algorithm (CGA), and a new quantum genetic algorithm (NQGA) to do coverage test for sensor network, the test results are shown in Table 2. From the analysis of Table 2, the coverage in this algorithm is 92.26 %, which is higher than other two kinds of algorithms 2.28 % and 0.65 %, and the iteration time of this algorithm is less than other two algorithms. Therefore this algorithm can implement the maximum network coverage through less operation time, which has high network coverage optimization performance. 5. Conclusion Wireless sensor network coverage optimization is in favor of improving network performance and network efficient coverage. This paper presents a wireless sensor network coverage optimization algorithm based on genetic PSO. Wireless sensor maximum coverage is regarded as the objective function, genetic algorithm with adaptive crossover mutation factor is used to search the solution space, PSO powerful global search ability of particle swarm is used to increase search scope, all of these make particle covering more efficient, which strengthen the optimization ability of the algorithm, improve the node coverage, and solve the prematurity problem. Comparing with the traditional genetic algorithm and a new quantum genetic algorithm, there are significant improvements on the effective coverage rate and convergence speed, this algorithm has certain superiority. Reference [1]. Yingchi Mao, The research on wireless sensor network coverage control technology, Journal of Computer Science, Vol. 11, Issue 3, 2007, pp. 20-22. [2]. Xueqing Wang, The research of coverage problems based on grid in wireless sensor networks, Journal of Computer Science, Vol. 33, Issue 11, 2011, pp. 38-39. [3]. Bin Qu, Fangyu Hu, Research on energy-efficient routing protocol of wireless sensor network, Computer Simulation, Vol. 25, Issue 5, 2008, pp. 113-116. [4]. Yonghua Zhou, Peng Li, Zongyuan Mao, A new hybrid method and its application in constrained optimization, Computer Engineering and Applications, 2006, pp. 48-50. [5]. J. C. Platt, Using analytic QP and sparseness to speed training of support vector machines, Advances in Neural Information Processing Systems II, Cambridge, MA: MIT Press, 1999, pp. 557-536. [6]. Xue Wang, Sheng Wang, Junjie Ma, The parallel particle swarm optimization strategy of wireless sensor network node location, Journal of Computers, Vol. 30, Issue 4,2007, pp. 563-568. [7]. Xiangguang Yao, Yongquan Zhou, Yongmei Li, Artificial fish mixed with particle swarm optimization algorithm, Computer Application Research, Vol. 27, Issue 6,2010, pp. 2084-2086. [8]. U. H.-G. Kre, et al. Pairwise classification and support vector machines, Advances in Kernel Methods-support Vector Learning, Cambrige, MA: MIT Press, 1999, pp. 255-268. [9]. Hua Fu, Shuang Han, The distribution optimization of new quantum genetic algorithm based on wireless sensor network sensing nodes, Journal of Sensing Technology, Vol. 21, Issue 7,2008, pp. 1259-1263. [10]. Jiantao Xia, Yiming He, Multi-class classification algorithm combining with support vector and error correction code combination, Journal of North Western Polytechnic University, Vol. 21, Issue 4, 2007, pp. 443-446. Yinghui HUANG School of Electronics and Information, Nantong University, Nantong Jiangsu, 226019, China E-mail: [email protected] Received: 30 June 2013 /Accepted: 25 October 2013 /Published: 30 November 2013 (c) 2013 International Frequency Sensor Association |