A MUSIC-Based Algorithm For Localization of Hybrid Near-Field and Far-Field Sources

In this paper, a new multiple signal classification (MUSIC)-based source localization algorithm is proposed to localize a hybrid of near-field (NF) and far-field (FF) sources. The MUSIC estimator for localizing the FF sources is referred to as the FF-MUSIC estimator. On the other hand, the NF-MUSIC estimator is designed for localizing the NF sources as well. In between the processing of these two estimators, the Capon spatial spectrum is exploited to reconstruct a pure NF covariance matrix. The proposed algorithm requires only several one-dimensional (1-D) spectral searches and does not involve any computations of high-dimensional iterative optimization, higher-order cumulants, and parameter pairing. Unlike the existing algorithms, which are derived based on the second-order Taylor series approximation, i.e., Fresnel approximation, of the spatial phase factor, the proposed algorithm is applicable to any -order expansion of the spatial phase factor. Moreover, it works well for non-uniform arrays, whereas the existing algorithms can only be applied to uniform arrays.