MIMO Radar Aided mmWave Time-Varying Channel Estimation in MU-MIMO V2X Communications
Sai Huang,Meng Zhang,Yicheng Gao,Zhiyong Feng
DOI: https://doi.org/10.1109/twc.2021.3085823
IF: 10.4
2021-11-01
IEEE Transactions on Wireless Communications
Abstract:Robust channel estimation in time-varying channels is used to guarantee the quality of communication services, especially for Vehicle-to-Everything (V2X) scenarios. To improve the channel estimation accuracy and reduce the pilot overhead, multi-input multi-output (MIMO) radar is deployed to assist millimeter wave (mmWave) channel estimation. In this paper, we propose a MIMO radar aided channel estimation scheme using deep learning (DL) for the uplink mmWave multiuser (MU)-MIMO communications. To allocate pilot resources reasonably, we design a transmission frame structure of joint radar module and communication module, which divides the estimation scheme into two stages, i.e., the arrival/departure (AoA/AoDs) estimation stage and the gain estimation stage. In view of the imperfections of array elements in practice, we propose an AoA/AoDs estimation algorithm based on subspace reconstruction in the AoA/AoDs estimation stage named two-step angle estimation (TSAE) algorithm. In the gain estimation stage, a DL based channel gain estimator is designed. An autoencoder combined with residual structure named residual denoising autoencoder (RDAE) is proposed to eliminate the noise on wireless signals, which is passed into the least square (LS) estimation module to obtain gains. Simulation results demonstrate that the MIMO radar aided and DL-based channel estimator provides the efficient estimation performance of the high-mobility mmWave channel with fewer training resources.
telecommunications,engineering, electrical & electronic