A Fault Detection Method for Wireless Sensor Networks Based on Credible Sensor Nodes Set

Abstract

In this paper, we present a fault detection scheme based on credible nodes set for wireless sensor networks. While an event occurs, the impact on the nodes around the event source can be described as a distance-related model. And the model can be transformed to a group of linear equations. The coefficients of linear equations are determined by energy value and intrinsic properties of the event source, and they can be worked out through the credible nodes' measured values and location parameters. By using improved hierarchical clustering algorithm, we can determine the credible nodes set through the similarity of nodes' measurements. Then the absolute measurement residuals of all nodes can be calculated by the linear regression algorithm and the measurement residuals will be used for concluding the final status of every node. So we can accomplish the detection. Simulation results show that the fault detection accuracy and the false alarm rate are kept high and low, respectively, even in the case where 40% of the sensor nodes are faulty.