Rapidly convergent «superresolving» direction-of-arrival estimation of noise radiation sources in adaptive arrays

Abstract

The paper summarizes and develops the results obtained by the authors in [11, 33, 49, 50]. We compare the efficiencies of some “superresolving” methods for estimation of a spatial spectrum of Gaussian noise in an antenna array for the case of a finite-size sample in maximum likelihood estimates of their correlation matrices. The comparison is based on an analysis of exact and empirical random parameters distribution laws that determine the methods resolution by statistical and non-statistical criteria. Significant differences in these laws are shown owing to which conclusions about comparative merits of different methods, based on their asymptotic properties analysis, can change to opposite ones under conditions of small samples. Causes of the difference and the possibilities following from their analysis to improve the convergence rate of adaptive methods for noise sources direction-of-arrival estimation are studied.