Phase compensation based localization of mixed far-field and near-field sources

For mixed far-field (FF) and near-field (NF) source localization, most existing algorithms are developed for the scenario where the number of available data samples N is much larger than the number of sensors M. Different from these methods, this letter attempts to deal with the scenario where M is large and of the same order of magnitude as N. To obtain a satisfied performance in such a scenario, the phase compensation result of the spike covariance matrix is first utilized to construct a modified spectral function and successively realize the DOA estimation of FF sources. Then, we exploit the oblique projection operation to extract the NF sources. Finally, DOA and range estimation for NF sources is achieved via two one-dimensional (1-D) spectral searches, where the reconstructed manifold vector combining the phase compensation result is used. The effectiveness of the proposed algorithm is demonstrated by computer simulations.