Identifying defective nodes in wireless sensor networks

Abstract

Wireless sensor networks (WSNs) have become ubiquitous, e.g., in logistics, smart manufacturing, smart city infrastructures or vehicular ad-hoc networks. WSNs tend to rely on ad-hoc infrastructures that are prone to a wide range of different defects, e.g., communication failures, faulty sensors or nodes that have been tampered with. Additionally, dealing with defects is challenging, as defects might occur only occasionally. In this paper, we introduce SEDEL, our approach for Sensor nEtwork DEfect Localization. SEDEL helps the WSN operator to pinpoint defective nodes in the routing topology of a WSN. In particular, we let the operator store graph representations of the routing topology, together with information if the WSN has produced errors. Based on this information, SEDEL assigns each WSN node a suspiciousness score that is correlated with the defect probability. Thus, our approach can be used with any kind of defect, and the kind does not have to be known, as long as the operator can decide if a certain processing is correct or not. We have evaluated SEDEL with a real sensor-node deployment. Our evaluation shows that the defective node is assigned a high probability in the vast majority of the experiments.