Fiber-Missing Tensor Completion for DOA Estimation with Sensor Failure

Abstract

In this paper, we propose a fiber-missing tensor completion-based method for direction-of-arrival (DOA) estimation in sensor failure scenario, where the completion of a three-dimensional (3-D) incomplete tensor signal is pursued to enable accurate DOA estimation. However, due to fibers of missing elements caused by sensor failure in the incomplete tensor signal, it cannot be effectively completed by the conventional low-rank tensor completion methods. To address this problem, a spatial-temporal dimensional augmentation technique is designed to construct a 6-D Hankel tensor with distributed missing elements from the 3-D incomplete tensor signal. Then, a Tucker-based tensor completion method is proposed to complete the augmented Hankel tensor. By operating the inverse Hankelization on the completed Hankel tensor, the resulting completed tensor signal can be effectively decomposed for DOA estimation. Simulation results verify the effectiveness of the proposed method.