Improving performance of node clustering in wireless sensor networks using meta-heuristic algorithms and a novel validity index

Abstract

The use of wireless sensor networks has significantly increased in the last decade. These networks consist of a large number of small sensors, which are efficient tools for data collection from different environments. The data collected by these sensors will be usually transmitted to a base station that will provide the end user’s data. In order to improve scalability of such networks, sensor nodes can be grouped into non-overlapped clusters. These clusters create a hierarchical design in wireless sensor networks that leads to better energy utilization and thus increase the network’s lifetime. Using validity indexes and meta-heuristic algorithms are common ways to improve performance of clustering. In this paper, we provide a new validity index called ASCS by enhancing the Chou and Su (CS) validity index, and improve the performance of the meta-heuristic algorithms for clustering using this new validity index as their objective functions. Differential evolution and harmony search are two algorithms that will be used for this purpose. The experimental results show the better performance of ASCS index comparing to Davies and Bouldin (DB) and CS validity indexes in determining appropriate number of clusters and determining proper clusters’ members.