Thenodesinthewirelesssensornetworksarescatteredinanareaandneedtobeclusteredforsendingandreceivingmessagesamongstthem.Itisevidentthatbyorganizingthesensornodesingroupsi.e.,clustersofnodes,wecanreapsignificantnetworkperformancegains.Clusteringnotonlyallowsaggregation,butlimitsdatatransmissionprimarilywithinthecluster,therebyreducingboththenetworktrafficandthecontentionforthechannel.Clusteringtransmitsaggregateddatatothedatasinkthusreducingnumberofnodestakingpartintransmission.Italsohelpsinenergyefficientroutingofdatapacketsandprovidesscalabilityforalargenumberofnodes.Clusteringalsoreducescommunicationoverheadforsinglehopaswellasmulti-hoproutingalgorithms.Therearemanyclusteringalgorithmsandeachofthemhasowncriteriaforclusteringthespatiallyscatterednodes.SomeofthemareCACC,VAP-E,CFL,K-meansandFCMwhichisused.TheFCMalgorithmisaclusteringalgorithmdevelopedbyDunn,andlateronimprovedbyBezdek.Itisusefulwhentherequirednumberofclustersarepre-determined,thusthealgorithmtriestoputeachofthedatapointstooneoftheclusters[1].WhatmakesFCMdifferentisthatitdoesnotdecidetheabsolutemembershipofadatapointtoagivencluster;instead,itcalculatesthelikelihood(i.e.,thedegreeofmembership)thatadatapointwillbelongtothatcluster.Hence,dependingontheaccuracyoftheclusteringthatisrequiredinpractice,appropriatetolerancemeasurescanbeputinplace.Sincetheabsolutemembershipisnotcalculated,FCMcanbeextremelyfastbecausethenumberofiterationsrequiredtoachieveaspecificclusteringexercisecorrespondstotherequiredaccuracy.