DECOUPLED ESTIMATION OF NOMINAL DIRECTION AND ANGULAR SPREAD BASED ON ASYMPTOTIC MAXIMUM LIKELIHOOD APPROACH

The problem of estimating the nominal direction and its un-derlying angular spread is considered herein. As encoun-tered in spatially distributed source localization, the compu-tation of these directional parameters might be regarded as two consecutive tasks. In this paper, we propose an asymptotic maximum l ikelihood (AML) approach to successively esti-mate both of them. The first advantage of estimation in this way is that it requires only two successive 1-dimensional searches rather than joint 2-dimensional optimization as uti-lized in the asymptotic maximum l ikelihood (AML) esti-mator. Therefore, the decoupled estimation in this way pro-vides more numerical flexibility. Since it belongs to a large-sample approximation of the exact ML method, numerical simulation is conducted in order to validate its asymptotic efficiency producible with respect to Cramér-Rao bound. Although its non-asymptotic performance is inferior to that provided by the joint AML approach, it appeared that, in the region of large number of temporal snapshot, the proposed AML estimator for decoupled estimation is the same as the AML criterion which employed the joint 2-dimensions. 1