Abstract: We used a distributed fuzzy controller method to carry out the fault tolerance and nonuniform clustering for wireless sensor networks. First, the residual energy of nodes, centrality of nodes and distance from nodes to base station were input to the fuzzy controller. The IFTHEN rule was used to reason and output “cluster head opportunity” and “cluster size” value to make the best performance node become cluster head and build a suitable cluster size. And then data transmission by time division multiplexing mode could tolerate the temporary and permanent faults of the cluster heads and member nodes. The simulation results show that the algorithm can effectively balance the load among clusters, reduce the energy consumption of the network, and thus improve life cycle of the network.

Key words: clustering, fault tolerance, wireless sensor network, load balance, fuzzy control