Target detection is an important problem in wireless sensor networks. Many studies in the literature have addressed the problem of evaluating the performance of a sensor network by detection probability. However, it is difficult to guarantee detection probability in a sensor network since it depends on the topology of the sensor deployment and the location of the target. A sensor network without a careful arrangement of sensor locations may experience very low detection probability. This paper integrates collaborative fusion and sequential detection to guarantee the quality of the decisions made by a sensor network and analytically derives the average detection latency based on value fusion and decision fusion. Specifically, sensors periodically report their local measurements or decisions to a fusion center. The fusion center makes final decisions only when both of the pre-defined false alarm probability and missing probability are satisfied. Otherwise, it will continue to collect data and repeat the decision making operations. Simple and elegant detection rules are provided for the collaborative sequential detection operations. Extensive simulations are conducted to show the performance of a sensor network in terms of detection latency based on the two fusion mechanisms. Further, the correctness of the analytical results for detection latency is also verified by simulations.
Tags: Saved for Later