分布式多传感器多元假设检验的最优决策融合算法

Optimal Decision Fusion Algorithm for M-ary Hypothesis Testing with Multi-sensor Distributed Detection Systems

研究分布式并行检测融合系统的M元假设检验融合算法。融合系统由融合中心及N部传感器构成。由于融合系统的检测性能由融合规则及各部传感器的判决规则共同决定,因此为了优化系统检测性能,需要联合优化融合规则及各部传感器的判决规则。在各部传感器观测相关的条件下,推导了联合最优化融合规则及传感器判决规则满足的必要条件,并给出了求解最优融合规则及传感器判决规则的数值迭代算法。仿真实验结果表明,采用该融合算法对系统性能进行优化,可获得明显优于单部传感器的检测性能。

The decision fusion algorithm for M-ary hypothesis testing with distributed parallel detection fusion system is considered. The fusion system consists of a fusion center and N sensors. Since the performance of the fusion system is determined jointly by the fusion rule and sensor decision rules, in order to optimize the system performance, the fusion rule and sensor decision rules must be optimized simultaneously. Under the condition of correlated sensor observations, the necessary conditions for joint optimization of fusion rule and sensor decision rules are derived,and a numerical iterative algorithm is also proposed to solve for the optimal fusion rule and sensor decision rules. Computer simulation shows that a system performance that is much better than that of single sensor can be obtained by optimizing the system performance with the proposed fusion algorithm.