NED: An Efficient Noise-Tolerant Event and Event Boundary Detection Algorithm in Wireless Sensor Networks

Abstract

Wireless sensor networks provide an advanced platform to observe the physical world. Different users may be interested in different events derived from a same spatial phenomenon. The constrained and noisy environment of sensor networks, however, challenges successful in-network solutions to monitor and detect events and event boundaries. This paper presents an efficient algorithm, named NED, to support event and event boundary detection in wireless sensor networks. NED encodes partial event estimation results into variable length messages exchanged locally among neighboring nodes. Sensor nodes estimate events and event boundaries based on moving averages to eliminate noise effects. Thus, NED is resource-friendly to constrained sensor networks, and scales well to very large networks. Our experiment results illustrate that NED’s communication cost is flexible and moderate to different noise levels, and NED provides high quality estimation results of event and event boundary detection.