Abstract

This study is devoted to the substantiation of practical recommendations on the selection of parameters for the band-diagonal regularization of maximum likelihood estimates of Hermitian correlation matrices of Gaussian interferences. These parameters include the diagonal regularization parameter and band regularization parameter, which represents the compensation multiplicity, i.e., the number of stages of adaptive lattice filters, for ensuring the high speed of adaptation of radar protection systems against the noise jamming and clutter corresponding to the theoretical (ultimate) limit. It has been shown that during the adaptive spatial processing of signals against the background of noise jamming, the selection of diagonal regularization parameter depends on the dimension of problem, while during the time processing of signals against the background of clutter, it also depends on the shape of correlation function of such interferences. The limitation of the number of stages of adaptive lattice filters as compared with the number of time channels makes it possible to simplify the adaptive processing of signals against the background of clutter and at the same time to increase the speed of adaptation.

中文翻译：

---

高斯干涉相关矩阵及其逆矩阵最大似然估计的带对角正则化参数的选择

摘要

本研究致力于证实有关高斯干扰厄米特相关矩阵的最大似然估计的带对角正则化参数选择的实用建议。这些参数包括对角线正则化参数和频带正则化参数，它们代表补偿的多重性，即自适应格子滤波器的级数，以保证雷达保护系统对噪声干扰和杂波对应的理论值的高速自适应。 （最终）限制。已经表明，在噪声干扰背景下对信号进行自适应空间处理时，对角正则化参数的选择取决于问题的维度，而在杂波背景下对信号进行时间处理时，也取决于这种干扰的相关函数的形状。与时间通道数相比，自适应格子滤波器级数的限制使得可以简化信号在杂波背景下的自适应处理，同时提高自适应速度。