A Fault Detection Scheme Based on Self-Clustering Nodes Sets for Wireless Sensor Networks

In this paper, we present a fault detection scheme based on self-clustering nodes sets for wireless sensor networks. In WSN, sensors´ measurements are spatially related, and fault nodes can be detected by local comparisons. We firstly cluster nodes into several sets by the similarity of nodes´ measurements using self-clustering algorithm. Nodes in the several top large-scale sets are mainly credible. Then we set credible nodes´ status as healthy, and run the adjacent nodes comparing algorithm. Faulty sensor nodes can be identified by the comparisons between neighbor nodes and dissemination of the decision made at each node. Time redundancy is used to tolerate transient faults in sensing and communication. Simulation results show that the fault detection accuracy is kept high and the false alarm rate is kept low, even when 35% of the sensor nodes are faulty.