Distributed Sequential Event-Region Detection in Sensor Networks

Abstract

In this paper, the problem of event-region detection in wireless sensor networks (WSNs) are explored. Sensor fusion techniques are developed to tackle this problem. However, in contrast to the existing sensor fusion rules in the literature, sensor nodes considered here could observe different phenomena or events, and each node makes its decision with respect to its own hypothesis. In particular, a space-memory fusion rule that considers both space-memory information and local detection performance is derived for event-region detection. In addition, the proposed fusion rule performs in a sequential manner. The need of the fault-tolerance capability for possible sensor faults is also integrated with the design of the space-memory fusion rule. Specifically, the a priori sensor-fault model is used in this paper to address the issue of fault-tolerance capability. Simulation results demonstrate the superiority of the proposed fusion rule in the event-region detection.