A novel cluster-header selection method in wireless sensor networks

Clustering has been proved to be an effective approach to organizing the wireless sensor network into a connected hierarchy for routing of sensing data. For efficient constructing in the actual circumstance, the distributed algorithm is more practical than the centralized algorithm. However, the distributed algorithm can not guarantee even distribution of cluster heads over the network, and it can cause wasteful energy consumption due to a lot of control messages needed to be exchanged among nodes. In this paper, we present a probabilistic scheme for selecting cluster heads based on only local information at each node without exchanging control packets among its neighbors. Computer simulation results show that our scheme offers a longer network life time than other existing schemes.