A low-complexity robust bearing estimator using reparameterization and covariance match for the distributed source

Many methods were proposed to estimate the bearing of a spatially distributed source. They usually suffer from heavy computational load and are limited to small angle spread due to model approximated error. In this paper, the two problems are considered and the main contribution is two-fold. First, the unimodal symmetric space frequency distribution is introduced to describe the source model. The exact expression of the covariance matrix can be calculated without model approximation even in the case of a large angular spread. Second, using reparameterization and covariance match criteria, a novel estimator is proposed by solving a nonlinear least-square problem in the central space frequency. It can be implemented only by a FFT or a one-dimension search. Numerical examples show its robustness for the large angular spread. A statistical analysis is also included to show the asymptotic efficiency of the bearing estimate.