Energy efficient grid based k‐means clustering algorithm for large scale wireless sensor networks

Abstract

SummaryHierarchical based routing is a kind of group‐based routing that consists in creating of a virtual hierarchy among network sensor nodes. This class of routing technique is generally designed for large‐scale networks. It aims to efficiently increase network lifetime by cutting the whole network into clusters. However, traditional clustering techniques show some limitations and do not take into consideration the self‐organized with dynamic topology inherent in wireless sensor networks (WSNs). These limitations can lead to an unbalanced cluster head distribution that affects the whole network energy consumption. In this paper, we propose a grid‐based k‐means clustering protocol (named GBK), which combines grid‐based routing with k‐means algorithm in order to overcome the above mentioned weaknesses. From the supervised zone area size parameter, the base station determines the optimal grid size based on our optimization study. Afterwards, the k‐means algorithm is executed in every grid cell generating a cluster head per cell grid where the nearest node to the grid cell centroid is elected. An enhancement of this proposed GBK algorithm named GBK‐R is also proposed to extend network stability of the GBK algorithm by node scoring calculation that take as parameters the node remaining energy in addition to the distance to centroid. Our proposed GBK and GBK‐R algorithms allow for an enhanced network stability and increase the network lifetime as demonstrated by our performance evaluation study. In addition, this GBK clustering algorithm provides a better network topology control and a better control of the random nature of node distribution by generating cluster heads with bounded localization.