Energy Efficient Event Detection Using Probabilistic Inference in Wireless Sensor Networks

Abstract

ABSTRACTA dense deployment in a wireless sensor network results in multiple nodes reporting the same event. Although data correlation is a much sought parameter for accurate event detection, such uncontrolled transmissions result in high in-network traffic, network congestion, and packet loss. Existing event detection schemes usually consider an acceptable trade-off between the detection accuracy and energy conservation. Therefore, a novel delay-based event detection scheme is proposed in this work to minimize the in-network traffic at the leaf level and optimize the overall energy consumption of the deployed network. The optimal data transmission path to the sink is chosen via a fitness function calculated for all the potential cliques in a Markov random field-based sensor network. Extensive simulation analysis of the proposed scheme proves its efficacy in terms of detection accuracy, communication cost, delivery delay, network energy, and packet delivery ratio.