Reliable and Fast Detection of Gradual Events in Wireless Sensor Networks

Abstract

Abstract. Event detection is among the most important applications of wireless sensor networks. Due to the fact that sensor readings do not always represent the true attribute values, previous literatures suggested threshold-based voting mechanism which involves collecting votes of all neighbors to disambiguate node failures from events, instead of reporting an event directly based on the judgement of single sensor node. Although such mechanism significantly reduces false positives, it inevitably introduces false negatives which lead to a detection delay under the scenario of gradual events. In this paper, we propose a new detection method -the bit-string match voting (BMV), which provides a response time close to that of the direct reporting method and a false positive rate even lower than that of the threshold-based voting method. Furthermore, BMV is able to avoid repeated and redundant reports of the same event, thus prolongs the life of the network. Extensive simulations are given to demonstrate and verify the advantages of BMV.

Efficient Data Collection with Spatial Clustering in Time Constraint WSN Applications

“…Paper [17] proposed a light-weight and accurate event detection approach for innetwork decentralized event detection. This approach used decision trees for distributed event detection.…”