训练样本不足时天线分置MIMO雷达贝叶斯自适应检测

Bayesian adaptive detection for widely distributed MIMO radar with limited training data

摘要:

在未知杂波环境下,为实现良好的目标检测性能,通常需要大量独立同分布样本,以此精准估计未知杂波的协方差矩阵。但在实际工作场景中,对于配备多通道的雷达系统而言,获取足够数量的独立同分布训练样本颇具挑战。为了解决天线分置多输入多输出(multi ple-input multiple-output,MIMO)雷达在训练数据不足时的目标检测难题,采用贝叶斯理论, 将杂波协方差矩阵建模为逆威沙特分布,并采用广义似然比检测(generalized likelihood ratio test,GLRT)准则、Rao准则和Wald准则设计得到了3种贝叶斯检测器。结果表明,所提出的检测器均能实现在训练样本不足时的目标检测,在3种贝叶斯检测器中,基于GLRT准则得到的检测器的检测性能最优。此外,还得出了影响检测性能的关键物理量。

Abstract:

In an unknown clutter environment, in order to achieve satisfactory target detec tion performance, a large number of independent identically distributed (IID) samples are usually required to accurately estimate the covariance matrix of the unknown clutter. However, in actual working scenarios, it is quite challenging for radar systems equipped with multiple channels to obtain a sufficient number of IID training samples. In order to solve the problem of target detection in widely distribubed multiple-input multiple-output (MIMO) radar with insufficient training data, Bayesian theory was adopted and clutter covariance matrix was modeled as an inverse Wishart distribution. Three Bayesian detectors were designed using the generalized likelihood ratio test (GLRT) criterion, Rao criterion, and Wald criterion. The results show that the proposed detectors can all achieve effective target detection with limited training samples, among which, the detector based on the GLRT criterion performs the best. In addition, the key physical quantities that affect the detection performance have also been given.