Hybrid Node Deployment Algorithm to Mitigate the Coverage Whole Problem in Wireless Sensor Network

Transcription

1 Middle-East Journal of Scientific Research 23 (10): , 2015 ISSN IDOSI Publications, 2015 DOI: /idosi.mejsr Hybrid Node Deployment Algorim to Mitigate e Coverage Whole Problem in Wireless Sensor Network L. Malai and R.K. Gnanamury 1 Asst.Professor/CSE, Vivekanandha College of Engineering for Women, Namakkal , Tamilnadu, India 2 Professor/ ECE, S.K.P Engineering College, Thiruvannamalai , Tamilnadu, India Abstract: Due to random deployment of static sensor nodes and environmental factors existence of holes in wireless sensor networks is inevitable. This holes results in reducing e data transmission performance and in additional energy consumption. Consequently, e hole problem is an important factor for prolonging e network lifetime. The network lifetime can be improved by deploying additional mobile sensors wi e initial static sensor deployment. Finding e number of additional sensors used is a optimization problem. To solve is we are proposing a Hybrid Electromagnetism like algorim (EM) wi genetic operators to obtain e best/optimal coverage of network area by sensor nodes. A new approach calculates e coverage rate of e sensors by using binary detection model and applies e EM wi genetic operators in order to obtain e optimal solution wiin e minimum iterations. A simulation result shows at e Hybrid EM algorim wi genetic operator performs better an basic GA based node redeployment. Key words: Coverage hole Node deployment Coverage hole healing Mobile sensors Hybrid sensor network Lifetime INTRODUCTION Intersection of ese points is called coverage point. Therefore, e main objective of e point coverage In recent years, e rapid developments of wireless protocols is to maximize e number of grid points at communication technology and microelectronics enables could be covered in e field. The coverage issue in wide application of e low-cost, low-power, multi- WSNs depends on many factors, such as e network function and tiny wireless sensor nodes. Hundreds and topology, sensor sensing model and e most important ousands of wireless sensor nodes distributed one is e deployment strategy at is used to distribute roughout a particular area constitute a wireless sensor or row e sensor nodes in e field [4]. The sensor network (WSN).[1] WSNs have widespread application nodes can be deployed eier manually based on a value and prospect in bo military and civil fields, such predefined design of e sensor locations, or randomly by as environmental monitoring, inventory management, dropping em from an aircraft. Random deployment is disaster recovery, object tracking and intrusion detection usually preferred in large scale WSNs not only because it and so on [2]. The Deployment of WSN may vary from is easy and less expensive but also because it might be application to application. It may contain static sensor or e only choice in remote and hostile environments. mobile sensor or hybrid of bo e sensors. However, random deployment of e sensor nodes can One of e key points in e design stage of a WSN cause holes formulation; erefore, in most cases, random at is related to e sensing attribute is e coverage of deployment is not guaranteed to be efficient for achieving e sensing field. In e literature, e coverage problem e required objective in terms of e coverage [3]. in WSNs has been addressed eier as point coverage or In order to overcome e problem of holes area coverage [3]. While area coverage protocols are formulation after initial deployment of e sensor nodes in designed to maximize e area of e sensing field at e sensing field, an efficient algorim at would could be covered, target coverage, on e oer hand, maximize e covered area or coverage point should be assumes at e sensing field is divided into grids. employed. In WSNs where all nodes are stationary, e Corresponding Auor: L. Malai, Asst.Professor/CSE, Vivekanandha College of Engineering for Women, Namakkal , Tamilnadu, India. 2500

2 area of e sensing field and e number of sensor nodes In [7, 8], auors have used Voronoi diagrams to are small, coverage can be maximized by manually detect coverage holes in mobile sensor networks. In [9], deploying additional nodes to e initially deployed ones. auors proposed a coverage hole detection meod However, in large scale WSNs where human intervention (CHDM) for mobile sensors by maematical analysis. It is not possible or when e sensing filed is hostile, is assumed at network consists of mobile nodes each random deployment is e only choice. wi sensing radius r and communication radius 2r. A In random deployment, e holes formulation node p is defined as neighbor of node q if it lies in its problem might be reduced or eliminated after initial communication range. On e basis of central angle deployment using one of two approaches. In e first between neighbor sensors, e auors presented approach, if all sensor nodes are mobile, en an efficient different cases to find coverage holes in communication algorim should be designed such at e coverage is circle around a redundant movable node. To patch hole, maximized while at e same time e moving cost of e a redundant node is moved to an appropriate position mobile nodes is minimized. In is case, e mobility inside e hole. In [10], e auors used e basic EM feature of e nodes can be utilized in order to maximize algorim for dynamic deployment of mobile sensor e coverage. After e initial configuration of e mobile network and avoid e coverage holes. nodes in e sensing field, an efficient algorim such as Researchers also used e hybrid of static and potential field algorim or virtual force algorim can be dynamic sensor nodes in order to improve e coverage. employed for e purpose of relocating e sensor nodes In [11], a bidding protocol was proposed, in which e [5]. static nodes are utilized as bidders and a number of mobile In e second approach, if e sensor nodes are nodes move accordingly to satisfy e coverage hybrid in which some of e nodes are stationary and e requirements. In [12], a distributed protocol was proposed oer are mobile, an efficient algorim should be at considers e different sensing capabilities of e employed in order to find e number and locations of e nodes using realistic sensing coverage model. In is mobile nodes at should be added after e initial protocol, e static nodes determine e uncovered areas deployment of e stationary nodes. One of e using a probabilistic coverage algorim and e mobile algorims at can be employed is an optimization nodes move accordingly using virtual force algorim. In algorim which is used to find an optimal or near optimal [13], several approaches were proposed based on virtual solution. This paper proposed an optimization algorim force algorim and paticle swarm optimization. The which is e hybrid of electromagnetism and GA obtained solutions were analyzed for better deployment operations to find number of sensor nodes at can be in e region of interest. Recently, a biogeography-based added after e initial node deployment in order to optimization algorim was proposed in [14] to maximize maximize e coverage. e coverage area of e network. Genetic algorims have This paper is organized as follows. Section 2 also been used to solve e problem of optimal node discusses e operation of e GA. Section 3 presents e deployment. In random deployment, genetic algorims related work. Section 4 discusses e assumptions and e are applied to determine near optimal positions for components of e proposed approach. Section 5 additional mobile nodes in order to maximize e coverage. presents simulation experiments and discusses e results In [15], a force-based genetic algorim was and Section 6 concludes e paper proposed, in which e mobile nodes utilize e sum of e forces used by e neighbors to choose eir direction. In Related Works: The WSN can have static node or mobile [16], a multi-objective genetic algorim running on a base node or hybrid of static and mobile sensor networks. This station was used. The base station determines where e section presents e existing algorim for each of e mobile nodes can move to maximize e coverage and category. In [6], e auors proposed 3MeSH (triangular minimize e travelled distance. The auors of [17] mesh self-healing hole detection algorim), which is a presented a triangular oriented diagram to detect a hole. distributed coordinate-free hole detection algorim for The auors connected e center of ree adjacent static sensors. If node detects presence of 3MeSH ring sensors to produce triangle and furer presented various defined by all its neighbors, en it is a boundary node. possibilities of occurrence of holes and en calculated For detecting large holes, nodes are allowed to collect e area which is not covered by any of em. This connectivity information from nodes furer away but at uncovered area is coverage hole. The approach is simple e cost of increased complexity. as compared to Voronoi diagram construction. 2501

3 Moreover, it can determine exact area of hole. In [18], e auors proposed a genetic algorim based node (2) deployment in order to optimize e network coverage and connectivity. They model e location of potential mobile sensor as chromosome in e solution apace of GA. The fundamental procedures of EM include initialize, EM based eory efficiently finds which mobile node local search, calculating total force and moving particles. should be moved in order get e maximum coverage. But The generic pseudo-code for e EM is as follows: it requires more number of iteration to find e optimal solution. GA based approaches find e optimum position Algorim 1. EM () of mobile sensor wi e minimum iterations. So in our proposed approach we combine e advantages of bo EM and GA to find e best optimal solutions. The purpose of is hybrid framework is to take e advantage of EM, which yields a high diversity population and GA operator let e algorim converge faster. Electromagnetism-Like Algorim: EM simulates e attraction-repulsion mechanism of electromagnetism eory which is based on Coulomb s law [19]. Each particle represents a solution and e charge of each particle relates to its solution quality. The better solution quality of e particle, e higher charge e particle has. Moreover, e electrostatic force between two point charges is directly proportional to e magnitudes of each charge and inversely proportional to e square of e distance between e charges. The fixed charge of particle i is shown as follows: i i best k where q is e charge of particle i, f(x )), f(x ) and f(x ) denote e objective value of particle i, e best solution and particle k. Finally, m is e population size. The solution quality or charge of each particle determines e magnitude of an attraction and repulsion effect in e population. A better solution encourages oer particles to converge to attractive valleys while a bad solution discourages particles to move toward is region. These particles move along wi e total force and so diversified solutions are generated. The following formulation is e force of particle i. (1) initialize() while (hasn t met stop criterion) do localsearch() calculate total force F() move particle by F() evaluate particles() End While Proposed Approach: In is section, we present our proposed approach The Fig.1 represents e overview of proposed system.. We first present e network assumptions and coverage model and en we discuss e hybrid approach. Network Assumptions: It was assumed at e sensor nodes are randomly deployed and equipped wi GPS and e base station node position is stationary. Furermore, e number of sensor nodes at are initially deployed equals e number of nodes at are required to achieve full coverage as if ese nodes were deterministically deployed. It was also assumed at few mobile nodes are available and can be used to repair e coverage holes after initial deployment of e stationary nodes. Coverage Model and Hole Detection: We assumed at each sensor node wi a sensing radius r can cover an area of circular shape. We also assumed at a Point P j can be detected by sensor S i if P j is wiin e sensing range of S. This can be represented using e binary i model of sensor detection which is given by Fig. 1: Overall view of Proposed System 2502

4 (3) Algorim 3. A Hybrid Framework where D is e distance between e point being sensed P j and e sensor node Si. The coverage function Coverage(S) equals 1 when e target object can be covered or sensed, oerwise it equals 0. If all e points wiin e deployment are covered by at least one node en e network is fully connected network. Oerwise we need to deploy additional mobile sensors to increase e coverage. Hybrid Approach for Mobile Sensor Deployment: The hybrid framework includes modified EM procedures and genetic operators, which adopts selection and mating. The selection operator is binary tournament and uniform crossover operator is applied in e framework. Generic EM provides an excellent diversity while GA is able to converge to a better solution quickly. Thus e hybrid meod takes e advantage of bo sides. The hybrid system starts wi determining which particle is moved by EM or mated by GA crossover operator. The new solution can be obtained from crossing between a better solutions selected by a binary tournament meod. And EM is used to move e inferior solution to a new position. This hybrid approach may encourage solutions converging toward better region quickly and to prevent from trapping into local optima by maintaining e population diversity. Algorim 2 is e pseudo code of e main procedures of e hybrid framework. The Algorim.3 presents our proposed approach. Algorim 2. A Hybrid Framework Initialize () while (hasn t met stop criterion) do local_search() avg calc_avgobjectivevalues() for i = 1 to m do if ((i best) and( f(xi ) < avg) en j a selected particle to mate particle i by binary tournament() uniform_crossover(xi, x j ) else if (f(xi ) > avg) en CalcForce() and Move(xi) end if end for evaluate particles() end while Initialize e Required Parameter: sensor detection radius r, communication range of e sensors comr, e size of e coverage area A, e number of static sensors Ns, Number of mobile sensors Nm, dimension of e population n parameter of GA crossover rate crrate, mutation rate murate, e maximum number of maxiteration iteration iteration are initialized Deploy e Static Sensors Randomly: Calculate e coverage ratio of static deployed nodes using (3) and find e optimum number of mobile sensors required to get e full coverage.initialize random locations for e deployment of mobile sensors. Deploy e Mobile Sensors Randomly Using: for t1=1 to m do for t2=1 to n do <Uniform(0,1) l k + ( u k- lk) end for end for (4) while iter maxiteration Calculate e Objective Function (fx) Values of e Sensors: In order to calculate e charges of e sensors which are deployed in e solution space randomly according to e particle model in e EM algorim, e objective function value of each sensor is calculated based on eir existing positions as follows: where f(xi) is e objective function value of sensor i, is e coordinate value in e k dimension of e sensor i. Calculate Charge of e Sensors: Sensors used in e network are assumed as a charged particle in e solution of e dynamic deployment problem. In order to calculate attraction-repulsion forces between e sensors based on e particle model of e EM algorim which is expressed by Algorim 1, e charge values of each sensor qi are calculated using (1). 2503

5 Detect e Optimal Distributed Mobile Sensors: Sensors which have been placed in optimal locations are detected wi e help (3) Middle-East J. Sci. Res., 23 (10): , 2015 Calculate e Resultant Force of Non-Optimal Sensors: A resultant force, fi, which is calculated according to (2) by adding up e forces applied by e oer sensors in e area to each sensor, is not optimal as a result of not being able to be placed in an optimal location, is applied to e non-optimal sensors. There is no need to calculate e resultant forces of optimal sensors. Because as it is stated on e next step, e 9 step, once and e sensors are placed on optimal locations, eir placements will not be changed so on. Fig. 2: Coverage ratio of different sensing ranges Update e Locations of Non-Optimal Sensors: Each nonoptimal sensor in e area updates its current location according to Algorim 2 by moving in e direction of e resultant force applied on it. iter = iter +1 end while Performance Evaluation: In is section, e performance of e proposed hybrid algorim is evaluated in terms of e amount of coverage (coverage ratio), degree of coverage (k-coverage) and number of additional mobile nodes. Moreover, e effect of e number of randomly deployed static nodes and e sensing ranges on coverage and number of additional mobile nodes were investigated. In e simulation environment, it was assumed at e sensor nodes were randomly deployed in 200m 200m sensor field. The number of static nodes deployed is set to 100. For e experiments e number of deployed mobile nodes is fixed to 30, while e sensing ranges vary from 10m to 20m. In e second experimente number of mobile node varies from 30 to 50. In each experiment, e coverage ratio, k- coverage was measured before and after applying e proposed algorim. Also e results were compared wi pure genetic algorim based node deployment, results shows e hybrid algorim performs better an e existing one. Effect of Sensing Range: Figure 2 shows e coverage ratio when e static nodes are randomly deployed and after adding 30 mobile nodes as a function of e sensing ranges. It is shown at e coverage ratio increases as e sensing radius of e deployed nodes increases. Fig. 3: Coverage ratio of different sensing ranges Fig. 4: Coverage ratio of different sensing ranges Fig. 5: Performance of Hybrid scheme 2504

6 Figure 3 shows e k-coverage when e static nodes 3. Wang, B., Coverage Problems in Sensor are randomly deployed and after adding e 30 mobile nodes as a function of e sensing ranges. As shown, e k- coverage increases as e sensing radii of e deployed nodes increase. This is because e coverage among sensor nodes wi large sensing range is very likely to overlap and hence more targets would be covered by multiple nodes. Figure 4 shows e number of additional mobile Networks -Survey, ACM Computing Surveys, 43(4): Younis, M. and K. Akkaya, Strategies and Tech for Node Placement in Wireless Sensor Networks: A Sur vey, Ad Hoc Networks, 6(4): Howard, A., M.J. Mataric and G.S. Sukhatm, Sensor Network Deployment using Potential Fields: nodes as a function of e sensing range. It is shown at A distributed, Scalable Solution to e Area e number of mobile nodes decreases as e sensing radii Coverage Problem, Proceedings of 6 International of e nodes increase. This is because more targets would Symposium on Distributed Autonomous Robotics be covered as e sensing range of e static nodes increases and hence less mobile nodes would be added to Systems, Fukuoka, June 2002, pp: increase e coverage ratio. 6. Li, X., D.K. Hunter and K. Yang, Distributed Figure 5 shows e performance of EM algorim, coordinate-free hole detection and recovery, in Genetic algorim and hybrid approach in terms of Proceedings of e 2006 IEEE GLOBECOM, San coverage ratio. The figure shows clearly at e hybrid Francisco, Calif, USA, November-December approach outperforms e case of random deployment of 7. Guiling, W., G. Cao and T. La Porta, e static nodes as e additional mobile nodes are Movement-assisted sensor deployment, in located into regions where targets are not covered by e rd Proceedings of e 23 Annual Joint Conference of static nodes. e IEEE Computer and Communications Societies, CONCLUSION IEEE, March Kanno, J., J.G. Buchart, R.R. Selmic and V. Phoha, Detecting coverage holes in wireless sensor This paper presents a hybrid framework to find an optimal solution to e coverage holes problem caused by random deployment of stationary sensor nodes in wireless sensor network. The coverage holes are detected by binary detection and ey are healed by adding mobile sensors wi e existing stationary sensors. The performance of e algorim was evaluated in terms of e coverage ratio, k-coverage and e number of additional mobile nodes requires for various sensing ranges. The results are also compared wi e existing algorims like EM and GA. The simulation results showed at e hybrid algorim can maximize e coverage of e sensing field by finding e minimum number of additional mobile nodes and eir best positions in e field. REFERENCES 1. Malai, L. and R.K. Gnanamury, Energy Efficient Data Collection Framework for WSN wi Layers and Uneven Clusters, International Review on Computers and Software (IRECOS), 9(4): Ahmed, N., S.S. Kanhere and S. Jha, The Holes Problem in Wireless Sensor Networks: A Survey, ACM SIGMOBILE Mobile Computing and Communications Review, 9(2): 4-8. networks, in Proceedings of e 17 Mediterranean Conference on Control and Automation (MED 09), pp: , Thessaloniki, Greece, June Zhao, E., J. Yao, H. Wang and Y. Lv, A coverage hole detection meod and improvement scheme in WSNs, in Proceedings of e International Conference on Electric Information and Control Engineering (ICEICE 11), pp: , April Recep ozdag and Ali Karci, Sensor Node Deployment Based on Electromagnetism-Like Algorim in Mobile Wireless Sensor Networks, International Journal of Distributed Sensor Networks Volume 2015, Article ID , pp: Wang, G., G. Cao and T. La Porta, A bidding protocol for deploying mobile sensors, in Proceedings of e 11 IEEE International Conference on Network Protocol (ICNP 03), pp: , November Ahmed, N., S. Kanhere and S. Jha, A Pragmatic Approach to Area Coverage in Hybrid Wireless Sensor Networks, Wireless Communications and Mobile Computing, 11(1):

7 13. Wang, X. and S. Wang, Hierarchical 17. Babaie, S. and S.S. Pirahesh, Hole detection Deployment Optimization for Wireless Sensor for increasing coverage in wireless sensor network Networks, IEEE Transactions on Mobile using triangular structure, International Journal of Computing, 10(7): Computer Science Issues, 9(1): TMC Omar Banimelhem1, Moad Mowafi1 and Walid 14. Wang, G., L. Guo, H. Duan, L. Liu and H. Wang, Aljoby, Genetic Algorim Based Node Dynamic Deployment of Wireless Sensor Networks Deployment in Hybrid Wireless Sensor Networks by Biogeography Based Optimization Algorim, Communications and Network, 5: , November Journal of Sensor and Actuator Networks, Vol. 1, No : Birbil, S.I. and S.C. Fang, An 15. Sahin, C. et al., Design of Genetic Algorims electromagnetism-like mechanism for global for To- pology Control of Unmanned Vehicles, optimization, Journal of Global Optimization, International Journal of Applied Decision Sciences, 25: (3): Qu, Y. and S. Georgakopoulos, Relocation of Wireless Sensor Network Nodes using a Genetic Algorim, Pro-ceedings of 12 Annual IEEE Wireless and Microwave Technology Conference (WAMICON), Clearwater Beach, April 2011, pp: