Implementation of the x-means clustering algorithm for wireless sensor networks

Abstract

K-means clustering algorithms in wireless sensor networks (WSNs) are potential solutions that prolong the network lifetime. However, there are limitations hampering these algorithms, such as size of clusters and the number of clusters. This paper proposes implementing X-means algorithm as a new clustering technique that overcomes K-means limitations; clusters constructed using tentative CHs and tentative area of centroids in an initial phase, After that, if a cluster meets splitting criteria, new centroids selected and new clusters constructed.