Optimal stopping strategies in collaborative event detection in wireless sensor networks
Bratislav Milić · 2012
Measurement noise and unreliability of communication in wireless sensor networks (WSNs) make event detection a non-trivial and error-prone task. In order to reduce errors in the event detection, collaborative event detection was introduced. This paper presents a stochastic model of collaborative event detection in presence of uncertainties caused by measurement noise at sensors and unreliable network. The model allows analysis of event detection accuracy in WSNs as well as synthesis of WSNs with set accuracy targets. An optimal stopping retry mechanism for the event detector is developed from the model. Simulation shows that the retry mechanism reduces probability of false alarms by 35 to 40% with a minimal transport overhead.