传感器观测相关条件下的最优分布式检测融合算法

Optimum Distributed Detection Fusion Algorithm for Correlated Sensor Observations

针对分布式串行检测融合系统的检测性能优化问题,研究了一种各部传感器判决规则联合优化的系统性能优化方法.融合系统由N部传感器构成,系统性能优化准则采用Bayes准则.在各部传感器观测相关的条件下,为了使系统检测性能达到最优,推导出了传感器判决规则联合最优化的必要条件,得出了最优传感器判决规则满足的一组系统方程,并根据系统方程组设计了求解各部传感器最优判决规则的数值迭代算法.仿真结果表明,采用该数值迭代算法联合优化各部传感器的判决规则,可使融合系统的Bayes风险显著低于单部传感器.

The system performance optimization for distributed serial detection system is considered. The fusion system consists of N sensors, and the optimality criterion is Bayesian criterion. In order to optimize the system performance, a joint optimization method of the sensor decision rules is proposed. The optimality condition for joint-optimizing sensor decision rules is derived for correlated sensor observations, and a set of system equations that are satisfied by the optimal sensor decision rules is obtained. A numerical iterative algorithm for the solution of the system equations is proposed to generate the optimal sensor decision rules. Simulation results show that, by optimizing the sensor decision rules jointly using the proposed iterative algorithm, the Bayes risk of the fusion system can be made much lower than that of single sensor.