摘要:
本文讨论了相关高斯噪声自适应非参量检测器。当输入噪声的采样满足独立、同分布(IID)条件时,不管分布是什么形式,非参量检测器均能保持虚警概率恒定(CFAR)。但是,许多类噪声并不能保持IID条件,检测器也就无法保持CFAR。本文提出一种简单、可行的自适应非参量检测器,能自动调节门限,使虚警概率(Pfa)趋于恒定。这种方法的关键在于用递归滤波器的输出值来测量噪声的相关系数(Pd),并由此改变秩值检测器的检测单元的权,使其Pfa趋于恒定。从而使在一定信噪比条件下发现概率(Pd)也趋于恒定。本文给出检测单元加权的非参量检测器的检测性能和渐近性能,然后还给出自适应非参量检测器的近似计算方法和计算机模拟结果。
Abstract:
The problems of adaptive nonparametric detectors are considered. The major reason behind the increasing engineering interest in nonparametric signal detectors is their ability to maintain CFAR (Constant-False-Alarm-Rate) at the detector output for a broad class of noise distributions. The false alarm probability (Pfa) is not constant, when the input noise is not the sequence of IID (Identical Independent Distribution) variance. In order to maintain CFAR the weight of the detection range cell of the detector can be changed continuously by different output value of IIR filter for measuring the correlation coefficient of the input noise. The closed form expressions for detection probability (Pd) and Pfa of the weighted nonparametric detectors are derived. The ARE (Asymptotic Relative Efficiency) of the detectors is investigated. finally, the detection performance of the adaptive nonparametric detector is determined for a finite number of observations.
下载:
