Energy Efficient Ensemble K-means and SVM for Wireless Sensor Network

Abstract

A wireless sensor network (WSN) consists of a large number of small sensors with limited energy. For many WSN applications, prolonged network lifetime is important requirements. There are different techniques have already been proposed to improve energy consumption rate such as clustering ,efficient routing , and data aggregation. In this paper, we present a novel technique using clustering .The different clustering algorithms also differ in their objectives. Sometimes Clustering suffers from more overlapping and redundancy data since sensor node's position is in a critical position does not know in which clustering it is belonging. One option is to assign these nodes to both clusters, which is equivalent to overlap of nodes and data redundancy occurs. This paper has proposed a new method to solve this problem and make use of the advantages of Support Vector Machine SVM to strengthen K-MEANS clustering algorithm and give us more accurate dissection boundary for each classes .The new algorithm is called K-SVM.Numerical experiments are carried out using Matlab to simulate sensor fields. Through comparing with classical K-MEANS clustering scheme we confirmed that K-SVM  algorithm has a better improvement in clustering accuracy in these networks.