MACHINE LEARNING TECHNIQUES USED IN WIRELESS SENSOR NETWORKS

Abstract:Wireless sensor network (WSN)refers to a group of spatially dispersedand dedicated sensors for monitoringand recording the physical conditionsof the environment and organizing thecollected data at a central location.Clustering is applied to divide thenetwork into clusters and each clusteris having a cluster head in which weperform data collection. In manycases, while we are collectingdifferent types of data there is chanceof redundancy in the data whileforming a cluster and thus the size ofthe data packets are increased. Thereis a need to aggregate the data at thecluster head such that data can becompressed into less number ofpackets and redundancy is also decreased. The Clustering Techniquesused are “Affinity Propagation” and“Spectral Clustering” and finally acomparision under variousperformance metrics like delay , CPUusage, execution time are donebetween the algorithms under the dataaggregation protocol is done and theefficient algorithm of the two isconcluded. The most efficientalgorithm of the two is considered asthe one used to reduce redundancy inWireless Sensor Networks(WSN)