Fuzzy Self-Clustering for Wireless Sensor Networks

Abstract

This paper is proposing a new protocol for uniformly clustering distributed wireless sensor networks based on fuzzy inference system. The objective of this protocol is increasing the lifetime of the sensor network by uniformly distributing the clusters on the wireless sensor network by reclustering the whole network repeatedly. The uniform distribution of clusters on the network provides efficient coverage of that network. Efficient coverage means that the clusters are neither overlapped nor leaving nodes in the network unclustered. Each sensor node uses the energy level, local density within its sensing range and time as parameters for clustering, re-clustering, and merging existed clusters by using fuzzy logic technique. The proposed protocol requires neither a prior knowledge of the sensor locations nor a centralized controller for clustering the sensor nodes. A comparison was conducted and found that the new protocol distributes clusters uniformly as good as ACE protocol and prolong the network lifetime in magnitude of 100% more than ACE.