Signal source modeling and estimation in direction finding

Abstract

Direction of arrival estimation is an interesting and important topic in array signal processing. In this dissertation, new signal source modeling and estimation techniques for direction of arrival are considered. The direction of a signal source has usually been assumed to be a point, which, however, is invalid in some cases. For example, if the direction of a signal source received at an array is angularly spread, or if the location of a point source is perturbed statistically, the direction of arrival estimation techniques based upon the point source assumption would provide with incorrect results. In order to cope with the problem, we propose distributed source models and estimation techniques when a signal source is angularly distributed and when the location of a point source is perturbed statistically. Distributed sources are considered starting from the most general class, and classified into the parametric and nonparametric sources based upon the distributed source shape. The relationships among the parametric, nonparametric, and point source models are explicitly discussed. In the parametric model, the direction of arrival and distribution parameter are obtained with MUSIC-based estimation technique, and the source distribution are estimated with a direct technique in the nonparametric model. Various statistical properties are also obtained. The application of direction of arrival techniques based upon the conditional and unconditional maximum likelihood estimation are also considered under the parametric model. In the perturbed-location source model, an estimation method based upon eigen-decomposition is investigated to estimate the direction of arrival and perturbation parameter. The asymptotic distributions of the estimation errors of the direction of arrival are also obtained to show the statistical properties of the estimation technique.