Tensor Decomposition-Aided Time-Varying Channel Estimation for Millimeter Wave MIMO Systems

Abstract

In this letter, a tensor decomposition-aided channel estimation scheme is proposed for millimeter wave (mmWave) time-varying channels. Due to the existence of time dimension, we model the received signal as a third-order tensor. By utilizing the property that the variations of angles of arrival/departure (AoA/AoDs) are much slower than the path gains, a two-stage tensor decomposition-based method is proposed for mmWave time-varying channel estimation. In the first stage, by employing the special property of the mmWave time-varying channel, a CANDECOMP/PARAFAC (CP) decomposition-based method is proposed to estimate the AoA/AoDs. In the second stage, a path gains and Doppler shifts estimation scheme is proposed based on the estimated angles. Therefore, the final time-varying channel can be recovered. Simulation results and complexity analysis demonstrate that the proposed scheme outperforms the conventional compressed sensing-based algorithms in term of both estimation accuracy and complexity order when estimating time-varying mmWave channels.