Energy Aware Swarm Optimization for Wireless Sensor Network. S. Thilagavathi 1, Dr. B. G. Geetha 2. India

Similar documents
Using Particle Swarm Optimization for Enhancing the Hierarchical Cell Relay Routing Protocol

Simulation Based Analysis of FAST TCP using OMNET++

RAP. Speed/RAP/CODA. Real-time Systems. Modeling the sensor networks. Real-time Systems. Modeling the sensor networks. Real-time systems:

CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION

Parallelism for Nested Loops with Non-uniform and Flow Dependences

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique

Adaptive Energy and Location Aware Routing in Wireless Sensor Network

Determining the Optimal Bandwidth Based on Multi-criterion Fusion

A Load-balancing and Energy-aware Clustering Algorithm in Wireless Ad-hoc Networks

Layer Based and Energy-Balanced Clustering Protocol for Wireless Sensor Network

Overview. Basic Setup [9] Motivation and Tasks. Modularization 2008/2/20 IMPROVED COVERAGE CONTROL USING ONLY LOCAL INFORMATION

Positive Semi-definite Programming Localization in Wireless Sensor Networks

Efficient Distributed File System (EDFS)

DEAR: A DEVICE AND ENERGY AWARE ROUTING PROTOCOL FOR MOBILE AD HOC NETWORKS

Performance Comparison of a QoS Aware Routing Protocol for Wireless Sensor Networks

IJCTA Nov-Dec 2016 Available

Load Balancing for Hex-Cell Interconnection Network

Long Lifetime Routing in Unreliable Wireless Sensor Networks

3. CR parameters and Multi-Objective Fitness Function

Extending Network Life by Using Mobile Actors in Cluster-based Wireless Sensor and Actor Networks

Performance Improvement of Direct Diffusion Algorithm in Sensor Networks

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration

NETWORK LIFETIME AND ENERGY EFFICIENT MAXIMIZATION FOR HYBRID WIRELESS NETWORK

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

A New Token Allocation Algorithm for TCP Traffic in Diffserv Network

Load-Balanced Anycast Routing

Wireless Sensor Network Localization Research

Constructing Minimum Connected Dominating Set: Algorithmic approach

MobileGrid: Capacity-aware Topology Control in Mobile Ad Hoc Networks

DESIGNING TRANSMISSION SCHEDULES FOR WIRELESS AD HOC NETWORKS TO MAXIMIZE NETWORK THROUGHPUT

THere are increasing interests and use of mobile ad hoc

EVALUATION OF THE PERFORMANCES OF ARTIFICIAL BEE COLONY AND INVASIVE WEED OPTIMIZATION ALGORITHMS ON THE MODIFIED BENCHMARK FUNCTIONS

An Optimal Algorithm for Prufer Codes *

Application of Improved Fish Swarm Algorithm in Cloud Computing Resource Scheduling

Video Proxy System for a Large-scale VOD System (DINA)

ARTICLE IN PRESS. Signal Processing: Image Communication

Natural Computing. Lecture 13: Particle swarm optimisation INFR /11/2010

Routing in Degree-constrained FSO Mesh Networks

Cluster Analysis of Electrical Behavior

Mobile Adaptive Distributed Clustering Algorithm for Wireless Sensor Networks

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS

Network Coding as a Dynamical System

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision

Complexity Analysis of Problem-Dimension Using PSO

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

An Entropy-Based Approach to Integrated Information Needs Assessment

Online Policies for Opportunistic Virtual MISO Routing in Wireless Ad Hoc Networks

Private Information Retrieval (PIR)

Resource and Virtual Function Status Monitoring in Network Function Virtualization Environment

Evaluation of an Enhanced Scheme for High-level Nested Network Mobility

Clustering Algorithm Combining CPSO with K-Means Chunqin Gu 1, a, Qian Tao 2, b

Performance Evaluation of Information Retrieval Systems

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation

Network Intrusion Detection Based on PSO-SVM

A Semi-Distributed Load Balancing Architecture and Algorithm for Heterogeneous Wireless Networks

Vol. 4, No. 4 April 2013 ISSN Journal of Emerging Trends in Computing and Information Sciences CIS Journal. All rights reserved.

QoS Bandwidth Estimation Scheme for Delay Sensitive Applications in MANETs

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;

sensors ISSN

CS 534: Computer Vision Model Fitting

Module Management Tool in Software Development Organizations

Wishing you all a Total Quality New Year!

A Hybrid Genetic Algorithm for Routing Optimization in IP Networks Utilizing Bandwidth and Delay Metrics

A Binarization Algorithm specialized on Document Images and Photos

Cost-Effective Lifetime Prediction Based Routing Protocol for Wireless Network

IMPROVEMENT of MULTIPLE ROUTING BASED on FUZZY CLUSTERING and PSO ALGORITHM IN WSNS TO REDUCE ENERGY CONSUMPTION

Cognitive Radio Resource Management Using Multi-Agent Systems

Backpropagation: In Search of Performance Parameters

A Distributed Three-hop Routing Protocol to Increase the Capacity of Hybrid Wireless Networks

Classifier Swarms for Human Detection in Infrared Imagery

Advanced radio access solutions for the new 5G requirements

Classification Method in Integrated Information Network Using Vector Image Comparison

Concurrent Apriori Data Mining Algorithms

Delay Variation Optimized Traffic Allocation Based on Network Calculus for Multi-path Routing in Wireless Mesh Networks

Efficient Content Distribution in Wireless P2P Networks

OPTIMAL CONFIGURATION FOR NODES IN MIXED CELLULAR AND MOBILE AD HOC NETWORK FOR INET

QoS-aware routing for heterogeneous layered unicast transmissions in wireless mesh networks with cooperative network coding

Fuzzy Filtering Algorithms for Image Processing: Performance Evaluation of Various Approaches

Cost-Effective Lifetime Prediction Based Routing Protocol for Mobile Ad Hoc Network

Outline. Type of Machine Learning. Examples of Application. Unsupervised Learning

Analysis of Particle Swarm Optimization and Genetic Algorithm based on Task Scheduling in Cloud Computing Environment

A Notable Swarm Approach to Evolve Neural Network for Classification in Data Mining

The Codesign Challenge

Optimizing Document Scoring for Query Retrieval

Distributed Grid based Robust Clustering Protocol for Mobile Sensor Networks

Energy Efficient Computation of Data Fusion in Wireless Sensor Networks Using Cuckoo Based Particle Approach (CBPA)

Virtual Machine Migration based on Trust Measurement of Computer Node

Dynamic Bandwidth Provisioning with Fairness and Revenue Considerations for Broadband Wireless Communication

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task

Cost-efficient deployment of distributed software services

Feature Reduction and Selection

Recommended Items Rating Prediction based on RBF Neural Network Optimized by PSO Algorithm

Research of Dynamic Access to Cloud Database Based on Improved Pheromone Algorithm

Efficient Load-Balanced IP Routing Scheme Based on Shortest Paths in Hose Model. Eiji Oki May 28, 2009 The University of Electro-Communications

THE PATH PLANNING ALGORITHM AND SIMULATION FOR MOBILE ROBOT

Distributed Topology Control for Power Efficient Operation in Multihop Wireless Ad Hoc Networks

A Decentralized Lifetime Maximization Algorithm for Distributed Applications in Wireless Sensor Networks

IMPROVING THE SPEED OF DYNAMIC CLUSTER FORMATION IN MANET VIA SIMULATED ANNEALING

A Saturation Binary Neural Network for Crossbar Switching Problem

Transcription:

Energy Aware Swarm Optmzaton for Wreless Sensor Network S. Thlagavath 1, Dr. B. G. Geetha 2 1. Research Scholar, Department of Informaton Technology, Insttute of Road and Transport Technology, Erode, Inda 2. Professor, Department of Computer Scence and Engneerng, K. S. Engneerng College of Technology, Thruchengode, Inda thlagavath12a@gmal.com Abstract: A Wreless Sensor Networks (WSN) s a dstrbuted network of wreless nodes wth bult n sensors for measurement of physcal parameters lke temperature, humdty and so on. WSNs nbult characterstcs of lmted avalable power source and low complexty processors dfferentate such networks from other wreless networks ncludng MANETs. WSN routng s specfcally challengng when the node has moblty. The man objectve of any WSN routng protocol s to provde effectve and effcent communcaton for the network wth mnmal power utlzaton. In ths paper, a hybrd Partcle Swarm Optmzaton (PSO) algorthm s proposed based on resdual energy for fndng optmal number of clusters and cluster head (CH). The suboptmal solutons found durng CH become the key nodes for formaton of multple routes between the CH and snk node usng Iteratve deepenng depth-frst search approach. Results show mproved performance of the network compared to Leach protocol. [S. Thlagavath, B. G. Geetha. Energy Aware Swarm Optmzaton for Wreless Sensor Network. Lfe Sc J 2013; 10(7s): 654-660] (ISSN:1097-8135).. 103 Keywords: Wreless sensor networks (WSN), Leach, Partcle Swarm Optmzaton (PSO), Wreless Routng 1. Introducton Wreless sensor networks (WSN) s a collecton of nodes formed nto a network [1] wth each node havng processng capablty and contan memory, a RF transcever, has a power source, and accommodates dfferent sensors and actuators. The nodes communcate wrelessly and self-organze after ad hoc deployment. WSN are wdely deployed at an ncreasng pace and s expected that n a decade, WSN access va the Internet s the norm. The new technology has unlmted potental for many applcaton areas ncludng mltary, envronmental, crss management, medcal, transportaton, and entertanment. Multhop routng s an mportant servce needed for WSN. Internet and MANET routng technques fal to perform well n WSN. Internet routng assumes relable wred connectons and hence there are rare packet errors, but ths s not the case wth WSN. Though most MANET routng solutons depend on symmetrc lnks between neghbors t s not true for WSN. These dfferences necesstated the nventon and use of new solutons. For WSN, often deployed ad hoc, routng begns wth neghbor dscovery. Nodes send rounds of messages and construct local neghbor tables whch nclude mnmum nformaton about each neghbor s ID and locaton. The nodes should know ther geographc locaton pror to neghbor dscovery. Informaton n tables should nclude nodes remanng energy, delay through that node, and lnk qualty estmate. Once tables exst, most WSN routng algorthms messages are drected from source locaton to an address based on geographc coordnates, not IDs. A typcal routng algorthm whch works smlarly s Geographc Forwardng (GF) [2]. In GF, a node knows ts locaton, and a message about t routng has the destnaton address. Ths node then computes as to whch neghbor node makes most progress to the destnaton by usng geometry based dstance formula and message s forwarded to the next hop. In GF varants, a node also accounts for delays, lnk relablty and remanng energy. Another WSN routng paradgm s drected dffuson [3]. Ths soluton ntegrates routng, queres and data aggregaton. A query s dssemnated ndcatng data nterest from remote nodes. A node wth approprate requested data reples wth an attrbute-value par. Ths attrbute-value par s attracted to the gradents based requestor, set up and updated durng query dssemnaton and response. Along the way from source to destnaton data s aggregated to lower communcaton costs. Data can travel over multple paths ncreasng routng robustness. However geographc forwardng and drected dffuson are not effectve when the wreless sensor node s moble. WSN routng s challengng because of nbult characterstcs whch dfferentate t from other wreless networks ncludng MANETs or cellular networks. Followng are some of the man characterstcs nherent to WSN: Sensor nodes are tghtly constraned as regards energy, processng, and storage capactes. Hence they requre careful resource 654

management. In most applcaton scenaros, WSN nodes are statonary after deployment except for a few moble nodes. In WSNs, sometmes data recept s more mportant than learnng the IDs of nodes whch forwarded the data. Also n contrast to typcal communcaton networks, all sensor network applcatons requre sensed data flow from multple sources to a specfc BS. But ths does not prevent data flow n other forms. WSN are applcaton specfc.e., desgn requrements of sensor networks change wth applcaton. WSN nodes poston awareness s mportant as data collecton s generally locaton based. It s not practcal to use Global Postonng System (GPS) hardware for ths task. Fnally, data collected by many WSN sensors s based on common phenomena and so there s the probablty that the data s redundant. Ths factor needs explotaton by routng protocols to mprove energy and bandwdth utlzaton. The man objectve of any routng protocol s to provde effectve and effcent communcaton for the network wth mnmal energy utlzaton. Clusterng has been wdely pursued by the research communty n WSN to acheve the network scalablty objectve. Each cluster has a leader, referred to as the clusterhead (CH). Load balancng s used to allocate traffc amongst dfferent paths to avod formng congested areas and at the same tme allow the energy consumed to be dstrbuted among the entre network [4, 5]. For energy effcent WSN routng protocol, the man objectve functon would be the energy constrants wthn the network. Beng a NP complete problem, an deal route can be found usng metaheurstc algorthm [6]. Ths paper explores route optmzaton technque usng an mproved Partcle Swarm Optmzaton (PSO), a metaheurstc algorthm known for fast convergence whch s crtcal for a hghly moble WSN. To deal ssues lke clusterng, optmal deployment, data aggregaton and node localzaton, the PSO has been successfully appled [6]. 2. Related Works There are varous energy aware algorthms avalable n lterature. Some of the works are revewed n ths secton. Kwon, et al., [7] studed the problem of the lfetme maxmzaton n WSN. The end-to-end transmsson success probablty constrant wth a cross-layer strategy whch concerns a combnaton of physcal layer, MAC layer and routng protocol s proposed. The problem consdered s dvded nto sub-problems at every layer. The optmal algorthm along wth another substtutve heurstc algorthm nvolvng only less complexty for every sub-problem s also proposed. Usng smulatons, an exchange relaton present between the relablty constrant and the network lfetme maxmzaton s llustrated. The strategy developed by a combnaton of the proposed algorthm at every layer extensvely enhances the lfetme of the network. For varous energy consumpton models, the effect of the retransmsson control on the energy competence was examned. The results obtaned by smulatng the proposed model reveals that lttle gan s acheved wth low power yeld n the multple retransmssons. Along wth the transmsson power, the power converson effcency of the amplfer maxmzes and also when there s short lnk dstance. Therefore, the proposed model s effcent and also relable. The data transformed to a snk by ntermedate sensors are mostly aggregated n order to decrease the cost. A subset of the sensors called aggregators performs ths compresson. The challengng problem faced s to deploy an approprate number of aggregators strategcally as most of the sensors are equpped wth small and unreplenshable energy reserves thereby to reduce the energy consumpton durng transportng and aggregatng the data. Chen et al., [8] ntally studed the sngle-level aggregaton and then proposed an Energy-Effcent Protocol for Aggregator Selecton (EPAS) protocol. Subsequently, to an aggregaton herarchy s t generalzed and further the work was done to extend EPAS to Herarchcal EPAS. For aggregator selecton, a fully dstrbuted algorthm was ntroduced by dervng the optmal number of aggregators. The results obtaned by smulaton of the proposed algorthm revealed that n WSN, the energy consumpton for data collecton can be decreased sgnfcantly. Zhang et al., [9] proposed a new onlne routng scheme named Energy-effcent Beaconless Geographc Routng (EBGR) that facltates fully stateless, loop-free, wthout the help of pror neghborhood knowledge the energy-effcent sensorto-snk routng at only low communcaton overhead. Intally, n EBGR every node calculates ts best nexthop relay poston the bass of the energy-optmal forwardng dstance, and usng the Request-To- Send/Clear-To-Send (RTS/CTS) handshakng mechansm, every forwarder chooses the neghbor closest to ts best next-hop relay poston as the nexthop relay. Under EBGR, the lower and upper lmts on hop count and the upper lmt on energy consumpton for sensor-to-snk routng are establshed. To provde energy-effcent routng n the presence of unrelable communcaton lnks n lossy sensor networks the EBGR can be capably 655

used. In wreless sensor networks wth hghly dynamc network topologes, the results obtaned from smulaton of the proposed scheme reveals the hgh-throughput over the other exstng protocols. In sngle-hop sensor networks, the transmssons are performed between every sensor and a fuson center, drectly whereas n the multhop sensor networks transmssons are performed between neghborng sensors that s much effectve n spectrum and energy. Huang et al., [10] ntroduced a dgtal transmsson energy plannng algorthm along wth an analog transmsson energy plannng algorthm n multhop sensor networks for the purpose of progressve estmaton nvolvng the knowledge of a routng tree from every sensors to a destnaton node. The proposed progressve estmaton algorthms ncludng ther transmsson energy plannng wthn a fnte tme mnmzes the network transmsson energy when ensurng any prespecfed estmaton performance at the destnaton node. When the transmsson tme-bandwdth product avalable for each lnk and each observaton sample s not too restrcted, the dgtal transmsson shows superor performance n transmsson energy compared to the analog transmsson. For wreless networks, many power-aware routng approaches are developed consderng that the nodes are prepared to gve up ther power reserves as a whole for the network. But, n many practcal utlty applcatons n whch the nodes are present n groups, a node s prepared to gve up power reserves wthn the nodes present n ts group and not to other nodes n the network. For reducton n power consumpton of each group, resources wll be shared among other groups also. Consequently, a coalton s created by the groups and through whch they are able to route each other s packets also. There are varous propertes for sharng among groups than sharng among ndvduals and mutually benefcal sharng between groups and nvestgate far. Guha et al., [11] proposed a pareto-effcent condton specfcally for group sharng on the bass of maxmn farness named as far coalton routng. In order to compute the far coalton routng, dstrbuted algorthms are also proposed. Therefore, benefcal sharng of resources mutually among dfferent groups s acheved by far coalton routng whch s also valdated by performng a range of smulatons. The most wdespread and lucratve are fxedpower wreless sensor networks. But the trbulatons experenced by these networks are energy constrants, RF nterference and envronmental nose. In order to obtan energy effcency, relablty and scalablty n message delvery, the routng protocols for these networks should preval over these trbulatons. But, the accomplshment of these necesstes creates conflctng demands. Loh, et al., [12] proposed a new routng protocol EAR whch s able to accomplsh effcent, relable and scalable performance only wth a mnmzed concesson of energy effcency. On the bass of four parameters.e., a weghted combnaton of dstance traversed, expected path length, energy levels and determnaton and mantenance of best routes dynamcally and the lnk transmsson success hstory are employed for desgnng the routng protocol EAR. The proposed method s compared wth four exstng protocols. The smulaton results obtaned show that the proposed EAR llustrates superor performance. Hence, the protocol based on a combnaton of routng parameters s better than the protocols based on only hop-count and energy or those mplementng floodng. The better performance of the proposed EAR s demonstrated n terms of packet latency, packet delvery rato, energy consumpton and scalablty. In several wreless sensor applcatons the Topk montorng s an essental component. Mnj Wu et al., [13] usng the semantcs of top-k query proposed FILA, whch s an energy-effcent montorng scheme. The common dea used to suppress unnecessary sensor updates s acheved by nstallng a flter at each sensor node. For the precson and effcacy of FILA, the basc ssues are flter settng and query re-evaluaton. In order to handle concurrent sensor updates, a query re-evaluaton algorthm s developed. Especally, to decrease the probng cost, optmzaton schemes are ntroduced. For the purpose of balancng energy consumpton and extendng the network s lfetme a skewed flter settng scheme s desgned. Eager and lazy are two flter update strateges proposed for supportng varous applcaton scenaros. Order-nsenstve top-k montorng, approxmate top-k montorng, and top-k value montorng are some of the top-k query varants for whch the algorthm s extended. Usng real data traces, the performance of the proposed FILA s evaluated extensvely. The results show that: 1) In terms of both network lfetme and energy consumpton, FILA outperforms the exstng TAGbased approach and Cache approach constantly. 2) Furthermore to returnng the top-k result set, for each of top-k sensor readngs FILA s able to acheve a tghtly bounded approxmaton. 3) For the traces evaluated, the overall performance of the lazy flter update scheme s superor to the eager scheme. 4) For the HM samplng scenaro, the skewed flter settng s better than the unform flter settng. Qng Cao et al., [14] proposed a new multcast protocol called ucast, n sensor networks for energy effcent content dstrbuton. The ucast s desgned to sustan several multcast sessons partcularly 656

durng whch the number of destnatons s small n a sesson. Any state nformaton relevant to the ongong multcast delveres at ntermedate nodes s not retaned n ucast. But, usng a scoreboard algorthm at ntermedate nodes the multcast nformaton n the packet headers and parse these headers are drectly encoded. The llustratons n ths paper are to handle multple addressng and uncast routng approaches, ucast s potent and suffcent. ucast s well-organzed, robust and scalable n the face of modfcatons n network topology, such as ntated wth the energy conservaton protocols. The performance of ucast s evaluated by smulaton systematcally and compared to other protocols and based on the Berkeley motes platform gathers prelmnary data from a runnng system. Therefore, ucast s effcently employed n varous uncast routng protocols. 3. Methodology 3.1 Partcle Swarm Optmzaton (PSO) The Partcle Swarm Optmzaton (PSO) algorthm s an adaptve algorthm founded on a psycho-socal representaton; a populaton of ndvduals also referred to as partcles, adaptng through returnng stochastcally toward prevously successful regons [15, 16]. There are two prmary operators n Partcle Swarm: Velocty update and Poston update. Durng generaton each partcle s accelerated toward the partcles n the prevous best poston and the global best poston. A new velocty value for each partcle s calculated at every teraton and ths s based on ts current velocty, dstance from ts prevous best poston, and dstance from ts global best poston. The value of velocty s then utlzed to compute the next poston of the partcle n search space. Ths procedure s then terated a specfc number of tmes, or tll a mnmum error s notced [17, 18]. The PSO algorthm can be vsualzed as follows: The PSO smulates the behavour of brd flockng randomly searchng for food n an area and there s only one pece of food avalable n the area under search. All brds are gnorant of the locaton of the food. But the dstance to the food through each teraton s calculated. The effectve procedure s to follow the brd nearest to the food. PSO learned from the above scene and used t to solve optmzaton problems. In PSO, each sngle soluton s a "brd" n search space, called "partcle". All partcles have ftness values evaluated by the ftness functon to be optmzed, and have veloctes whch drect the partcle flght. The partcles fly through the problem space by followng current optmum partcles. PSO ntalzes wth a cluster of random partcles or solutons and searches for optma through updatng of generatons. In each teraton, the two "best" values of the partcle s updated. The frst one s called pbest whch s the best soluton (ftness) acheved tll then. Another "best" value tracked by the partcle swarm optmzer s the best value, obtaned tll then by any partcle n the populaton.e the global best and s called gbest. Another best value used s lbest whch s the best value of the partcle partcpatng wth ts topologcal neghbors. PSO s computatonally economcal as t requres only prmtve mathematcal operators. PSO optmzes clusterng n a network because these knds of networks have lmted resources. Partcle postons and veloctes are generated randomly n the begnnng. The algorthm then proceeds teratvely and updates all veloctes and postons of all the partcles as follows: v wv c r p x c r p x d d d d d d 1 1 2 2 g x x v d d d where d s the number of dmensons, s the sze of the populaton, w s the nerta weght, c 1, c 2 are postve constants called cogntve parameter and socal parameter respectvely, r1 and r2 are random d v values n the range [0, 1]. s the new velocty of the th partcle computed based on the partcle s prevous velocty, dstance between the prevous best poston and current poston and dstance between d x the best partcle of the swarm. calculates the new poston of the partcle. In classcal PSO, f gbest s far away from the global optmum then the partcles tend to get trapped n the local optmum n the gbest regon. To avod ths, the partcles are moved to a larger search space to fly and pbest poston of a parcle s updated based on the pbest poston of all the partcles n the swarm. Ths ncreases the ablty to avod local optmum and mproves dversty of the swarm. The updatng velocty of the partcle s gven by: V w v c rand pbest x d d d d d * * * fd where f f 1, f 2,..., f d refers to the pbest s the pbest that the partcle used and fd dmenson of partcles pbests. Two partcles are selected randomly and the partcle s whose velocty s updated s excluded. The ftness value of the partcles pbests are compared and the dmenson of the better one s selected to update the velocty [19]. 3.2 Proposed Methodology The proposed methodology s mplemented n two steps to modfy the exstng Leach routng. In the frst step, cluster head s selected usng PSO and n 657

the second step, multple routes are selected usng sub optmal nodes and teratve deepenng depth-frst search approach. Fgure 1 shows the flowchart of the flow chart of the proposed optmzng of PSO. No Fgure 1: Flowchart of optmzng usng Partcle Swarm Optmzaton A clusterng algorthm s used for selecton of the CH based on the weght of each node [20]. The weght of the node v s gven by W v as follows: W w D w S w M w P v 1 v 2 v 3 v 4 v where D v =degree of dfference, S v = sum of dstances of the members of the CH M v =avg speed P v = accumulatve tme of a node beng a CH The CH chosen has a sum of weghts equal to 1 and wth mnmum weght W v. Each node has a unque ID whch s used to encode the partcles and partcles have the IDs of all the nodes n the network. The CH algorthm stops when all the nodes are assgned as CHs or members of a CH. Followng are the condtons that the algorthm teratvely uses on each node n the partcle to check the sutablty of a node to be CH or not: Is the node a CH Begn Intalze all partcle wth Max-Mn Evaluate Ftness Value (Wv) Select Partcle for Velocty update Update the pbest and gbest Update velocty and poston of each partcle PSO Fnal Sln? End Yes Is the node member of CH Is the number of neghbors less than the maxmum neghbors allowed. The W v value s used to fnd the ftness of each partcle. Process s contnued to maxmum number of teratons. On convergng of the algorthm, soluton for the global best partcle s reported. The clusterng algorthm observes the followng rules: the Receved Sgnal Strength (RSS) gves the dstance of the node from the event; f the RSS s above the threshold value RSS, then the node s located wthn the event area; resdual energy of the neghborng nodes are known. The energy consumpton s less when the dstance s less, thus CH located nearest to the event requres the least energy for data transmsson. The lfetme of the network can be prolonged by mnmzng the energy consumpton n the cluster. Ths can be gven as follows: max T,mn C, jse where E Se s the energy consumpton of the cluster and T C s the workng tme of the cluster. Resdual energy s taken nto consderaton durng selecton of CH. The chance of a node to become CH s hgh, when t has hgher resdual energy, more neghbors and strong sgnal strength. The objectve functon of the CH s obtaned as follows: E Se k1 k2 k3 q1 E * K * SE where E represents the resdual energy, K set of neghborng nodes, SE sgnal strength detected and k 1, k 2, k 3 are the weghts controllng E, K and SE. The CH dynamcally chooses a route for transmttng data that depends on path metrc, such as energy consumpton. An energy constrant metrc s used whle searchng for the multple routes among the CHs and snk node. The Energy Constrant metrc computes the nter-flow nterference and the varaton n the transmsson rates and loss ratos of wreless lnks. The EC metrc s desgned as follows: * IEC c ETT c N c N c j j j where N (c) s the set of neghbors of node c s the channel c j N c N c s the total number of nodes that may be nterfered wth by the transmsson actvtes between Node and Node j over channel c. ETT j (c), the expected transmsson tme, whch computes dfferences n transmsson rates and loss ratos of lnks. The EC metrc s the combned channel tme of all nodes used by the transmssons of the flow at lnk (, j, c). An mportant mplementaton ssue of EC s 658

the estmaton of N(c). When nodes are wthn each other s nterference range, the transmsson rate s reduced f both the nodes transmt smultaneously [21]. The Iteratve deepenng depth-frst search approach works as follows: Frst, a depth-frst search to depth one node s performed. Then, dscardng the nodes generated n the frst search, start over and do a depth-frst search to level two. Next, start over agan and do a depth-frst search to depth three and so on, contnung ths process untl a goal state s reached. as Iteratve deepenng depth-frst search approach expands all nodes at a gven depth before expandng any nodes at a greater depth. It s guaranteed to fnd a shortest-length soluton. Fgure 3: Route dscovery tme for Leach wth and wthout optmzaton 4. Expermental Setup The proposed method was smulated n OPNET for evaluaton. The evaluaton setup conssts of 18 sensor nodes and one snk spread over an area of 4 sq. Km. The maxmum hop count n the setup s 5 hops. Maxmum avalable bandwdth s 11 Mbps and transmsson power for each node s 0.005w. Fgure 2 shows the smulaton envronment. The smulatons are run for 300 sec. Fgure 4: Number of Clusters formed for Leach wth and wthout optmzaton The number of clusters formed by the proposed Leach s slghtly more than the regular Leach routng. Fgure 2: The expermental setup. Fgures 3 to 5 gve the smulaton results for routng and routng wth proposed optmzaton. Route dscovery tme, Number of clusters formed and packet delvery rato are shown respectvely. From Fgure 3, t s observed that the route dscovery tme for the proposed Leach routng s hgher than the regular Leach. The dscovery tme s more as CH dynamcally chooses a route for transmttng data that depends on path metrc n the proposed Leach and also as multple routes are selected usng sub optmal nodes. Fgure 5: Packet Delvery Rato for Leach wth and wthout optmzaton It s observed from the Fgure 5, that the packet delvery rato mproves drastcally n routng wth proposed optmzaton. Ths leads to mprovement n the performance of the network. Though the route dscovery tme and the number of clusters formed s more n the proposed Leach routng, the packet delvery rato ncreases n tune of 3.5% for network of more than 200 nodes. 659

5. Concluson The routng protocol of any network ams to provde effectve and effcent communcaton. In WSN, Sensor nodes are tghtly constraned as regards energy, processng, and storage capactes, requrng careful resource management. In most applcaton scenaros, WSN nodes are statonary after deployment except for a few moble nodes. In ths paper, a hybrd Partcle Swarm Optmzaton (PSO) algorthm s proposed based on resdual energy and multple routes between the cluster head (CH) and snk node are found based on energy constrant metrc. The suboptmal solutons found durng CH become the key nodes for formaton of multple routes usng Iteratve deepenng depth-frst search approach. The proposed method s evaluated through smulaton. The smulaton results demonstrate the effcency of the proposed optmzaton for fndng routes and mprovng the packet delvery rato n WSN network. References 1. J. Hll, R. Szewczyk, A, Woo, S. Hollar, D. Culler, and K. Pster, System Archtecture Drectons for Networked Sensors, ASPLOS, November 2000. 2. B. Karp, Geographc Routng for Wreless Networks, PhD Dssertaton, Harvard Unversty, October 2000. 3. C. Intanagonwwat, R. Govndan, and D. Estrn, Drected Dffuson: A Scalable Routng and Robust Communcaton Paradgm for Sensor Networks, Mobcom, August 2000. 4. Jan Tang, Guolang Xue, and Wey Zhang, Interference-Aware Topology Control and QoS Routng n Mult-Channel Wreless Mesh Networks, n ACM Mobhoc, 2005. 5. Yalng Yang, Jun Wang, and Robn Kravets, Interference-aware Load Balancng for Multhop Wreless Networks, Tech. Rep. UIUCDCS-R- 2005-2526, Department of Computer Scence, Unversty of Illnos at Urbana-Champagn, 2005. 6. Raghavendra V. Kulkarn, and Ganesh Kumar Venayagamoorthy, Techncal Correspondence Partcle Swarm Optmzaton n Wreless-Sensor Networks: A Bref Survey, IEEE Transactons On Systems, Man, And Cybernetcs Part C: Applcatons And Revews, Vol. 41, No. 2, March 2011. 7. Hojoong Kwon, Tae Hyun Km, Sunghyun Cho, and Byeong G Lee, A Cross Layer Strategy for Energy-Effcent Relable Delvery n Wreless Sensor Networks, IEEE Transactons On Wreless Communcatons, Vol. 5, No. 12, December 2006 8. Yuanzhu Peter Chen, Arthur L. Lestman, and Jangchuan Lu, A Herarchcal Energy-Effcent Framework for Data Aggregaton n Wreless Sensor Networks, IEEE Transactons On Vehcular Technology, Vol. 55, No. 3, May 2006 9. Habo Zhang and Hong Shen, Energy-Effcent Beaconless Geographc Routng n Wreless Sensor Networks, IEEE Transactons On Parallel And Dstrbuted Systems, Vol. 21, No. 6, June 2010 10. Y Huang and Yngbo Hua, Energy Plannng for Progressve Estmaton n Multhop Sensor Networks, IEEE Transactons On Sgnal Processng, Vol. 57, No. 10, October 2009 11. Ratul K. Guha, Carl A. Gunter, and Saswat Sarkar, Far Coaltons for Power-Aware Routng n Wreless Networks, IEEE Transactons On Moble Computng, Vol. 6, No. 2, February 2007 12. Xaoyng Zheng, Chunglae Cho, and Ye Xa, Optmal Swarmng for Massve Content Dstrbuton, IEEE Transactons On Parallel And Dstrbuted Systems, Vol. 21, No. 6, June 2010 13. Peter Kok Keong Loh, Hsu Wen Jng, and Y Pan, Performance Evaluaton of Effcent and Relable Routng Protocols for Fxed-Power Sensor Networks, IEEE Transactons On Wreless Communcatons, Vol. 8, No. 5, May 2009 14. Mnj Wu, Janlang Xu, Xueyan Tang, and Wang- Chen Lee, Top-k Montorng n Wreless Sensor Networks, IEEE Transactons On Knowledge And Data Engneerng, Vol. 19, No. 7, July 2007 15. Qng Cao, Tan He, and Tarek Abdelzaher, ucast: Unfed Connectonless Multcast for Energy Effcent Content Dstrbuton n Sensor Networks, IEEE Transactons On Parallel And Dstrbuted Systems, Vol. 18, No. 2, February 2007 16. Latff, N.M.A.; Tsmends, C.C.; Sharf, B.S.;, "Performance Comparson of Optmzaton Algorthms for Clusterng n Wreless Sensor Networks," Moble Adhoc and Sensor Systems, 2007. MASS 2007. IEEE Internatonal Conference on, vol., no., pp.1-4, 8-11 Oct. 2007 17. Matthew Settles, An Introducton to Partcle Swarm Optmzaton, 2005 18. Eberhart, R. C., Sh, Y.: Partcle swarm optmzaton: Developments, applcatons and resources, In Proceedngs of IEEE Internatonal Conference on Evolutonary Computaton, vol. 1 (2001), 81-86. 19. S. Agarwal, J. Padhye, V. Padmanabhan, L. Qu, A. Rao, and B. Zll, Estmaton of Lnk Interference n Statc Multhop Wreless Networks, n ACM Internet Measurement Conference (IMC), 2005. 1/8/2013 660