分布式检测系统的反馈融合算法

Single Step Feedback Fusion Algorithm in Distributed Signal Detection System

通过分析分布式检测系统的工作原理,在原有并行反馈检测系统的基础上提出了一种新的局部检测器反馈融合算法.该算法利用检测器局部判决结果与数据融合中心融合结果的误差对检测器下一步判决门限进行修正,使各局部检测器可以根据所处的环境,采用不同的修正因子修正判决结果,避免了原有检测系统对所有局部检测器使用相同的修正因子所带来的系统性能损失,从而达到提高检测器检测概率的目的.仿真结果表明,该算法能有效提高系统的性能,对于慢起伏噪声信号具有良好的抑制作用,同时也指出了该算法的不足以及改进的方法.

By analyzing the principle of distributed detection system, a novel optimal local fusion rule is presented based on the general parallel feedback detection system. This algorithm modifies the local detector's decision threshold with the error between local decision and the fusion center's decision. With this algorithm, the local detectors can modify their decision with different factor, avoiding the loss of system's performance with a common modifying factor. The result shows that the detection performance is obviously improved, and the present algorithm can overcome the influence of the slow swelling noise. In addition, the shortage of the algorithm is discussed and the method of improving the algorithm is presented.