Performance Evaluation of Clustering Techniques in Wireless Sensor Networks

Abstract

Clustering is one of the essential techniques in wireless sensor network (WSN). Clustering is done to achieve the energy efficiency, improve network lifetime and the scalability of the network. The sensor nodes (SNs) in the network are arranged into various small clusters and each cluster is assigned with a cluster head (CH). Cluster formation is mandatory objective for maximizing the network lifetime to conserve energy. In this work, the problem of clustering is formulated in accordance with dissimilarity factor. The network nodes are deployed and clusters are formed randomly for a large area network. The selection of CHs done dynamically on the basis of residual maximum energy and performance is optimized on the basis of energy consumption. In this paper clustering techniques such as Mean-shift, Fuzzy C Mean (FCM), K-mean (KMEAN) and Hierarchal clustering (HC) are simulated and the results are compared on the basis of dissimilarity factor. HC is showing better results in comparison to the other clustering algorithms. The performance comparison of various clustering techniques is used to find a better formation algorithm for WSN. Better clustering with the proposed HC algorithm will provide better communication in a cost effective manner.