An efficient second-order method of near-field source localization and signal recovery

In this paper, we propose an efficient algorithm to localize and recover multiple sources in near-field by a uniform linear array (ULA). Firstly, we transform 2-D (bearing and ranging) search into 1-D (only bearing) operation via a series of approximations between near-field and far-field with low bias. This equivalence relation has property of rotational invariance between adjacent subarrays in signal subspace, which allows us to employ ESPRIT method for the far field bearing estimation. Then the near field range and bearing can be obtained by constructing another symmetric array. Finally, utilizing the near-field manifold, the signal can be recovered without ambiguities of order and scale. The algorithm does not require high-order statistics computation and avoid the global peak searching. The performance of the algorithm is tested by simulations.