基于最小风险准则的多传感器检测策略及应用

Strategies analysis of multiple sensors detection based on minimal risk rule

基于带宽、延时、计算等因素,传统多传感器检测以分布式为主,其特点是单传感器根据局部观测做出二元判决并传输给融合中心进行融合。单站条件下,高速通信网络和高性能服务器的应用逐步克服了制约集中式检测的约束,通过融合单传感器的局部观测提升检测性能,避免了大量原始信息的损失。基于最小风险准则推导了分布式检测与集中式检测的算法原理,分析了两种策略在实施过程中的特点与难点,结合单站条件下多雷达联合检测探讨了集中式检测在工程实现时所需考虑的因素。

Distributed detection is studied and implemented mainly for the multiple sensors detectionconsidering such factors as the bandwidth, time-delay, computation capability, etc. It is featured that thebinary hypothesis test with which the fusion center makes a decision is performed based on the localobservation of a single sensor. In a single station, the factors restricting the centralized detection areovercome by applying the high performance communication and the server, where the original observationof a single sensor without any loss can be transmitted to the fusion center to improve the detectionperformance. Based on the minimal risk rule, the theories of distributed detection and centralizeddetection are analyzed. The features and difficulties in implementation of both strategies are discussed.The fundamental factors in engineering of the centralized detection are studied combined with the multipleradars detection in the single station.