Computationally Efficient Maximum Likelihood Approach to DOA Estimationof a Scattered Source

Abstract

The problem of Direction-Of-Arrival (DOA) estimation in the presence of local scatterers using a uniform linear array (ULA) of sensors is addressed. We consider two models depending on whether the form of the azimuthal power distribution is explicitly known or not. For both models, the block-diagonal structure of the associated Fisher Information Matrix (FIM) is exploited to decouple the estimation of the DOA from that of the other model parameters. An asymptotically efficient Maximum Likelihood (ML) DOA estimator is derived which entails solving a 1-D minimization problem only. Furthermore, the 1-D criterion can be expressed as a simple Fourier Transform. A numerical comparison with the Cramer-Rao Bound (CRB) illustrates the fact that our computationally very simple DOA estimators are statistically efficient for a wide range of scenarios.