An algorithm for 5-D parameters estimation of near-field sources using parallel factor analysis

To effectively avoid the complicated spectral peak searching and multi-dimensional (M-D) parameter pairing, a novel fourth-order cumulant-based algorithm for estimating five-dimensional (5-D) parameters of near-field (NF) source is presented. Firstly, this approach exploits a non-centro symmetric uniform linear array equipped with concentered orthogonal loop and dipole element, and chooses proper sensor outputs to construct four cumulant matrices; secondly, it forms a parallel factor (PARAFAC) model with unique low-rank decomposition in the cumulant domain by utilizing these matrices; thirdly, the proposed algorithm utilizes the low rank three-way array decomposition of the PARAFAC model to jointly estimate the frequency, direction-of arrival, range and polarization parameters of each NF source. The simulation results reveal that the proposed approach works well in low signal-to-noise ratio (SNR) and small number of snapshots conditions as compared to the subspace method.