Coherent DOA estimation via Toeplitz-based tensor decomposition

Abstract

Coherent signals encountered in practical electromagnetic environments present significant challenges for high-resolution direction-of-arrival (DOA) estimation. While conventional spatial smoothing-based tensor reconstruction methods neither fully exploit the structural properties of the covariance matrix nor exhibit satisfactory signal decorrelation performance. Therefore, we propose a Toeplitz-tensor-based decorrelation (TTD) algorithm. Specifically, we construct Toeplitz matrices using all the rows of the covariance matrix, which are then stacked as a tensor PARAFAC model. The backward tensor is then built to form an augmented one to improve the performance. Furthermore, we conduct a comprehensive performance analysis for the tensor rank and the computations. Simulation results verify that the TTD not only improves estimation accuracy but also demonstrates stronger resistance to signal correlation effects.