A Novel Approach for Fault Diagnosis in Wireless Sensor Networks

Abstract

<p class="MsoNormal"> <span lang="EN-US">Wireless sensor networks (WSNs), belonging to an emerging class of distributed systems, have attracted significant attention over the past few years. As these systems grow in size, they suffer from component failure due to increas</span><span lang="EN-US">ing</span><span lang="EN-US"> in system complexity. Moreover, the fault types in WSNs are quite different from each other. WSNs consist of a number of nodes connected by links; each of the nodes generally ha</span><span lang="EN-US">s</span><span lang="EN-US"> both sensing and processing capabilities. Faults in the sensors are generally of a permanent nature, whereas the processing logic might be intermittently faulty. The mechanism proposed in this paper is shown to diagnose both permanently faulty sensors and intermittently faulty processing unit in wireless sensor nodes with a high degree of accuracy. The proposed algorithm has been evaluated both theoreti<span style="letter-spacing:-.05pt;">cally and through simulation using generic diagnostic parameters such as the Faulty Sensor Detection Accuracy (FSDA),</span> Identifiability, and message complexity.</span> </p