A pure cumulant-based method with low computational complexity for classification and localization of multiple near and far field sources using a symmetric array

The authors propose a new method based on spatial cumulants for estimating the parameters of multiple near-field and far-field sources. The Toeplitz property used in some studies is not applicable to fourth-order statistics to separate sources components. Therefore, in this paper, a method is proposed to compute output cumulants of specified sensors in special arrangements, by which the components of the near-field and the far-field sources are effectively separated using differencing. The angle and range estimations, as well as the classification of the sources, are obtained based on the data from two spatial cumulant matrices. One of them contains the angle information of all sources, and the other only contains the information of near-field sources. The parameters extraction algorithm is based on the ESPRIT technique; therefore, the proposed method does not require any spectral search. This leads to a significant reduction in computational complexity. Unlike some approaches, the proposed method does not suffer from array aperture loss. Also, the parameters pairing procedure is done automatically. Analysis and simulation results confirm the good performance of the proposed method in terms of computational complexity, estimation accuracy, correct classification of signals, and aperture loss. © 2019, Electromagnetics Academy. All rights reserved.