查看文章

The network lifetime is a vital research area of wireless sensor networks (WSNs), K-means are potential solutions that prolongs the network lifetime. However, there are limitations hampering these algorithms, such as the number of clusters is a fixed value and the initial positions of the cluster centroids are predetermined locations. A bad choice of initial centroids leads to a bad clustering and a high energy consumption, thus this paper proposes two phases optimization of initial centroids called X-means; first K-means cluster formed then they are slit into multiple clusters. In general, clusters constructed using tentative CHs selected by K-means algorithm as an initial phase, After that, each cluster gives birth to children, then these children compete to form clusters and the surviving children are the final clusters. The simulation results showed that X-means outperforms the traditional K-means algorithm and it has …