Statistical analysis of 'superresolution' noise direction finding methods in an array antenna at a finite

摘要

The paper presents the comparison of the effectiveness of some "superresolution" methods for spatial spectrum estimation of Gaussian noise radiation in array antenna at finite number of snapshots, given by the maximum likelihood estimations of their covariance matrices. The comparison is based on the analysis of exact or empirical laws of random parameters distribution that determine resolution capabilities of these methods regarding statistical and non-statistical criteria. Essential differences of these laws are shown due to which conclusions on comparative merits of different methods based on the analysis of their asymptotic properties may change to opposite at the actual conditions of small sampling size. The paper examines the reasons of the differences and possibilities to improve the performance of the adaptive noise direction finding methods resulting from the analysis of these differences.