High performance doa trackers derived from paralell low resolution detectors

Abstract

Traditionally, high resolution spectral direction of arrival (DOA) estimation has been associated with algorithms rather than with a processing scheme or architecture. Motivated by previous work on feasible implementations of the expectation-maximization algorithm, the authors show that classical bank filter approach can get similar, even better, performance than the most sophisticated algorithms, in terms of performance versus complexity. In fact, the practicality and robustness required for DOA trackers, both in radar and in the mobile communication scenarios to alleviate data fusion and hand-over respectively, makes evident the use of filter-bank or scanning beams for DOA tracking at the expense of resolution. The herein reported tracker enhances complexity and robustness of these schemes, achieving high resolution from the EM architecture. The result is a low complexity tracker with robustness against coherent sources and a resolution close to singular value decomposition (SVD) based methods.