基于Rao检验的信号统计分辨界分析

Analysis of statistical resolution limit based on Rao test

针对基于广义似然比检验(GLRT)的信号统计分辨界(SRL)分析存在计算较复杂、难以得到检测统计量解析表达式等问题,提出了一种基于Rao检测的SRL分析方法。分析了噪声方差已知和未知两种情形下的信号一维方位角SRL以及噪声方差已知时的信号二维方位角及俯仰角SRL,推导得到了Rao检测统计量以及对应SRL所需的最小信噪比(SNR),分析了检测性能、计算复杂度,并与GLRT方法进行了对比。仿真实验验证了Rao检测方法的有效性。

Since the approach of analyzing statistical resolution limit （SRL） based on generalized likelihood ratio test （GLRT） has the problems of high computational complexity and isn＇t easy to get the analytical formula of the detec- tion statistics, a method for the analysis of the SRL based on Rao test was proposed. The unidimensional SRL of azi- muth under the cases of known and unknown noise variance, two-dimensional SRL of azimuth and elevation under the case of known noise variance were analyzed. The Rao test detection statistic and the minimum SNR required for the corresponding SRL were derived, the detection performance and the computational complexity of the Rao test were analyzed and compared with the GLRT method. Simulation results verify the validity of the Rao test.