Data-supported optimization for maximum likelihood DOA estimation

We introduce a new conceptually simple and computationally effective solution to the maximum likelihood (ML) direction of arrival (DOA) estimation problem that consists of maximizing the likelihood function (LF) over a set of points derived from the data. Two different data-supported optimization (DSO) based techniques are formulated which use the data-supported grid points obtained by means of ESPRIT-like and root-MUSIC-like methods, respectively. The first technique is shown to be the method of choice in the short sample size case, whereas the second technique is applicable to situations where the number of snapshots is large. We show that the data-supported grid search of the LF provides the performance similar to that achieved by the genetic algorithm (GA), but at a significantly lower computational cost