Distributed event detection in wireless sensor networks

Abstract

In this paper, we presents a distributed event detection algorithm for fault-prone wireless sensor networks. Each sensor node makes a local decision based on the comparison of its readings with those of its neighbors. An event is detected by a combination of local decision based on current sensor readings and local decision with m most recent readings. Discrete measurements over a time interval are used to tolerate most transient faults in sensor readings. The resulting reduction in the number of transient faults greatly reduces the false alarm rate. In addition, the event detection accuracy is maintained high since the tradeoff between false alarm rate and event detection accuracy is mitigated by the enhanced robustness to transient faults. Computer simulation shows that the proposed algorithm outperforms some existing algorithms, such as majority voting.