基于遗传算法的分布式传感器网络节点故障检测与恢复(IJCNIS-V6-N12-5)

I.J.ComputerNetworkandInformationSecurity,2014,12,37-46PublishedOnlineNovember2014inMECS()DOI:10.5815/ijcnis.2014.12.05Copyright©2014MECSI.J.ComputerNetworkandInformationSecurity,2014,12,37-46GeneticAlgorithmBasedNodeFaultDetectionandRecoveryinDistributedSensorNetworksLokeshB.BhajantriDepartmentofInformationScienceandEngineering,BasaveshwarEngineeringCollege,Bagalkot,Karnataka,India–587102.E-mail:[e-mail]enceandEngineering,NitteMeenakshiInstituteofTechnology,Bangalore,Karnataka,India-560064.E-mail:[e-mail]—SensornodesarepronetofailureduetoenergydepletionandsomeotherreasonsinDistributedSensorNetworks(DSNs).Inthisregardfaulttoleranceofnetworkisessentialindistributedsensorenvironment.Energyefficiency,networkortopologycontrolandfault-tolerancearethemostimportantissuesinthedevelopmentofnext-generationDSNs.ThispaperproposesanodefaultdetectionandrecoverybyusingGeneticAlgorithm(GA),whensomeofthesensornodesfaultyinDSN.Themainobjectiveofthisworkistoprovidefaulttolerancemechanism,whichisenergyefficientandresponsivetonetworkbyusingGAwhichisusedtodetectthefaultyofnodesinthenetworkbasedontheenergydepletionofnodeandlinkfailurebetweennodes.Theproposedfaultdetectionmodelisusedtodetectfaultsatnodelevelandnetworklevelfaults(linkfailureandpacketerror).Wehaveevaluatedtheperformanceparametersfortheproposedscheme.IndexTerms—DistributedSensorNetworks(DSNs),GeneticAlgorithm(GA),FaultDetectionandRecovery.I.INTRODUCTIONADSNisahighleveldistributedsetofsensorsthatareinterconnectedbyacommunicationnetworkintheenvironment.Thesensorsaredeeplyembeddeddevicesthatareintegratedwithaphysicalenvironmentandcapableofacquiringsignals,processingthesignals,communicating,contextawarecomputingandperformingsimplecomputationtasks.Whilethisnewclassofnetworkshasthepotentialtoenablewiderangeofapplications,italsoposeseriouschallengeslikerouting,datagatheringanddissemination,frequentnetworktopologychange,andfaulttolerance[1].Withalltheseconstraintsanefficientandeffectivemethodtoextractfaultsfromthenetworkischallengingtaskandalsoposemanynewchallengescomparedwithtraditionalnetworksasfollows:powerawareandenergyefficient,exceptionfreeandunattendedoperationorselfconfiguring,andmustrespondtodynamicenvironment,data-centricanddataconcentratedandapplicationspecific,andfaulttolerance[2].DSNsareoftendeployedinhostileandunattendedenvironments.Sensorsmayfailfromimpactofdeployment,fireorextremeheat,animalorvehicularaccidents,maliciousactivity,orbyextendeduse.Thesefailuresmayoccurupondeploymentorovertimeafterdeployment,extensiveoperationmaydrainsomeofthenodespowerorexternalfactorsmayphysicallydamagepartofthenodes.Additionally,hazardsmaychangedevicespositionsovertime,possiblydisconnectingthenetwork.Anyoftheseinitialdeploymenterrors,sensorfailures,orchangeinsensorpositionscausethenetworktobedisconnectedorlackotherdesiredpropertiesandneedtodeployadditionalsensorstofixthenetwork.Itmayexistathardwarelayer,softwarelayer,networkcommunicationlayer,nodelevelandapplicationlayer[3].TherearevariouslevelsoffaultsinDSNenvironmentsuchas:atnodelevelorsinknodelevelandnetworklevelfaults(linkfailureandpacketerror).Inthiswork,ithasbeenconsideredthatanodelevelfaultandreplacethosenodesinthenetwork.II.RELATEDWORKSSomeoftherelatedworksaregivenbelow:Asurveyonfaultmanagementinwirelesssensornetworksispresentedin[4].Thispaperaddressesthechallengebysurveyingexistingfaultmanagementprocessandalsosummarizestheexistingmanagementarchitectures.Theworkpresentedin[5]depictsadistributedfaulttoleranttopologycontrolinstaticandmobilewirelesssensornetwork.Inthisworkadistributedalgorithmforassigningminimumpossiblepowertoallthenodesinthewirelesssensornetworkwasdiscussed.Theworkgivenin[6]presentsadistributedtopologycontrolinwirelesssensornetworkswithasymmetriclinks.Itconsiderstheproblemoftopologycontrolinanetworkofheterogeneouswirelessdeviceswithdifferentmaximumtransmissionranges.Theworkgivenin[7]presentsaBayesiandecisionmodelforintelligentroutinginsensornetworks.Inthispaper,newefficientenergy-awareroutingalgorithmbasedonlearningpatternsispresented.Theprobabilisticdecisionmodelbothconsideredthe38GeneticAlgorithmBasedNodeFaultDetectionandRecoveryinDistributedSensorNetworksCopyright©2014MECSI.J.ComputerNetworkandInformationSecurity,2014,12,37-46estimationoftheavailableenergyattheneighboringnodesandtheimportanceofthemessagestomakeintelligentdecisions.Theworkgivenin[8]describestheselectionofclusterheadsinwirelesssensornetworksusingBayesiannetwork.InthispaperaBayesiannetworkbasedapproachisusedtoselectclusterheads.Theapproachincorporatesenergylevelandsignalstrengthofeachsensornode.ExperimentshavebeenconductedtocomparetheperformanceoftheproposedapproachandLEACHdeterministicclusterheadselectionandtheresultsshowsthattheBayesiannetworkbasedapproachperformsbetterthanLEACHdeterministicclusterheadselectionandchain-basedLEACH.Theworkgivenin[9]addressesfault-toleranttopologycontrolinaheterogeneouswirelesssensornetwork.Itintroducesthek-deg