Bearing estimation in the presence of near-field sources

Most high-resolution bearing estimation algorithms have been based on the assumption of a far-field source environment. When this assumption is violated, i.e., when some sources are in the near field, these algorithms generally show unsatisfactory performance. An algorithm to tackle this problem using a uniform linear array is proposed. The observed signal covariance matrix of near-field sources is equivalently viewed as a corrupted measurement of the ideal signal covariance matrix of far-field sources located at the same bearings. A far-field approximation (FFA) covariance matrix is reconstructed from the observed covariance matrix to reduce the corruptions. Therefore, existing high-resolution bearing estimation techniques (e.g., MUSIC) can be directly applied to estimate the bearings of the near-field sources. Simulation results demonstrating the effectiveness of this approach are presented.