Model-Driven Deep Learning Based Channel Estimation and Feedback for Millimeter-Wave Massive Hybrid MIMO Systems

This paper proposes a model-driven deep learning (MDDL)-based channel estimation and feedback scheme for wideband millimeter-wave (mmWave) massive hybrid multiple-input multiple-output (MIMO) systems, where the angle-delay domain channels' sparsity is exploited reducing overhead. First, we consider uplink time-division duplexing systems. To reduce pilot overhead estimating high-dimensional channels from limited number of radio frequency (RF) chains at base station (BS), propose to jointly train phase shift network estimator as an auto-encoder. Particularly, by exploiting structured priori model integrated trainable parameters data samples, proposed multiple-measurement-vectors learned approximate message passing (MMV-LAMP) with devised redundant dictionary can recover multiple subcarriers' significantly enhanced performance. Moreover, downlink frequency-division Similarly, pilots BS users be trained encoder decoder, respectively. Besides, further overhead, only received on part subcarriers are fed back BS, which exploit MMV-LAMP reconstruct spatial-frequency matrix. Numerical results show that MDDL-based outperforms state-of-the-art approaches.