基于局部多元判决的稳健的分布式检测算法  

Robust distributed detection algorithm based on local m-ary decision making

作　　者：刘向阳[1] 彭应宁[1] 王秀坛[1] 

机构地区：[1]清华大学电子工程系

出　　处：《清华大学学报（自然科学版）》2007年第7期1155-1158,共4页Journal of Tsinghua University(Science and Technology)

摘　　要：为提高基于二元判决的Ⅳ雷达分布式检测算法在各个雷达的信噪比未知且可能时变条件下的检测性能，提出了基于局部Ⅳ门限量化的分布式检测算法和衡量不同检测算法性能的一般性方法。仿真结果表明：该算法性能优于各种基于“N选k”融合的检测算法，在特定情况下比基于OR融合的方案有2～2．5dB的信噪比改善，相对于极大似然准则下的集中式检测器的信噪比损失不到0．8dB。在各个传感器的信噪比差异明显时，与基于OR融合的方法有同样的稳健性，因而是一种稳健且高性能的检测算法。The detection probability of an N-radar distributed detection scheme based on local binary decision making with unknown and perhaps time-variant signal-to-noise ratios （SNR） is improved using an algorithm based on local quantization with N thresholds. A general method was developed to measure the detection performance of different detection algorithms for this case. Results show that the algorithm outperforms all detection schemes based on k out of N fusion and has similar detection performance to the centralized detector with the maximum likelihood criterion. For these conditions, the SNR is improved by 2 ～ 2.5 dB compared with the OR Boolean fusion based scheme. Compared with the centralized detector, the SNR losses are at most 0.8 dB. For large differences in the SNRs at various sensors, the algorithm is as robust as the OR Boolean fusion based scheme.

关 键 词：多雷达 分布式检测 局部多元判决 稳健检测 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]