Localization of Mixed Far-Field and Near-Field Sources via Cumulant Matrix Reconstruction

In this paper, we propose a new localization algorithm using a uniform linear sensor array for scenarios where both the far-field (FF) and near-field (NF) sources may coexist. First, the direction of arrivals (DOAs) of FF sources are obtained using the FF MUSIC spectrum. Then, a special fourth-order cumulant matrix of the array output which is only characterized by the DOAs of sources is constructed. After the kurtosis of FF signals is estimated, the NF cumulant matrix (NFCM) is reconstructed by removing the related FF components. Next, the DOAs of NF sources are estimated by exploiting the reconstructed NFCM and cumulant domain rotational invariance structure. Finally, with the estimated NF DOAs, the ranges of NF sources are estimated via 1-D spectral search. The proposed algorithm does not require NF DOA search and is computationally more efficient than the previous methods. In addition, it realizes a more reasonable classification of the source types. Numerical results show that the proposed algorithm achieves higher estimation accuracy than the traditional methods.