在非高斯马尔可夫噪声中随机信号的近似最佳检测

Approximately Optimal Detection of Random Signals in Non-Gaussian Markovian Noise

本文在假设仅已知噪声的边缘分布和相关系数ρ的条件下,探讨了非高斯马尔可夫噪声中随机信号的最佳检测问题。当ρ较小时,以检测器效验的最大化为准则,应用Taylor定理和变分法,在对ρ取三级近似的意义下得到了最佳无记忆检测器的闭式解,并比较了它与白噪声中的最佳检测器在结构上的差异和性能上的改进。

In this paper optimal detection of random signals in non-Gaussian Markovian noise is considered on conditions that only the marginal probability density function f(x) and the dependent coefficient ρ of noise samples are given. The optimum structure of memoryless detectors is obtained by maximizing the third-order approximation of detection efficacy to ρ.