本文研究了分布式CFAR检测在Neyman-Pearson(NP)意义上的最优问题,阐述了基于二元局部判决和基于局部检测统计量的分布式CFAR检测在最优化上的差别.提出了R类和S类局部检测统计量,证明了基于这两种局部检测统计量的在NP意义上的最优分布式CFAR检测具有似然比检验形式.研究了它们的次优融合方法(OS-R-GLRT和OS-S-CA).结果表明,次优融合可以达到与相应的最优融合接近的性能.

This paper studies the optimization problem in Neyman-Pearson(NP) sense in distributed CFAR detection.The difference of optimization between the distributed CFAR detection based on binary local decision and local test statistic (LTS) is elaborated.R and S type LTSs are proposed.It is proved that the optimal distributed CFAR detection based on the above two LTSs in NP sense is in form of likelihood ratio test (LRT).Their suboptimal fusion schemes are developed (OS-R-GLRT and OS-S-CA).The results show that suboptimal fusion schemes could attain similar performance to that of optimal fusion.