Distributed Byzantine fault detection technique in wireless sensor networks based on hypothesis testing

Abstract

Distributed fault detection in wireless sensor network is an important problem where every sensor node identifies its own fault status based on the information from its neighboring sensor nodes. This paper presents a novel distributed fault detection algorithm to detect the soft faulty sensor nodes in sparse wireless sensor networks. In the proposed scheme, every sensor node gathers the information only from their neighboring nodes in order to reduce the communication overhead. The Neyman–Pearson testing method is used to predict the fault status of each sensor node and the neighboring sensor nodes. A voting scheme is applied on the fault status information to obtain the final fault status of each sensor node. The generic parameters such as detection accuracy, false alarm rate, time complexity, message complexity, detection latency, network life time and energy consumption are considered to evaluate the performance of proposed scheme analytically as well as through simulation. The result shows that the proposed scheme significantly improves the performance over the existing algorithms.