Algebraic Algorithm for Mixed Near-Field and Far-Field Sources Classification and Localization

Abstract

Using uniform linear array (ULA), a passive localization algorithm is presented for mixed far-field (FF) and near-field (NF) signals scenarios. Based on the high-order cumulant (HOC) technique, a special Hermite matrix is constructed by three fourth-order cumulant matrices, which are calculated by dividing the ULA into two sub-arrays. Then, the special matrix of signals is decomposed to obtain the source subspace. According to ESPRIT algorithm, two transformation matrices of all sub-arrays can be obtained. Meanwhile, the two transformation matrixes could be used to calculate the range and angles of arrival (AOA) of NF sources, as well as AOAs of FF sources. Moreover, compared with two-stage MUSIC (TSMUSIC) and four-order cumulant MUSIC method, the proposed algorithm has higher accuracy for localisation of both FF and NF sources without any spectral search.