Subspace-Based Super-Resolution Sparse Channel Estimation in Millimeter-Wave Massive MIMO Systems

This chapter introduces super-resolution sparse channel estimation (CE) schemes for both narrowband and wideband mmWave massive MIMO systems with hybrid precoding. Specifically, for narrowband case, a two-dimensional (2D) unitary estimating signal parameters via rotational invariance techniques (ESPRIT) algorithm is adopted to accurately estimate the angle of arrivals/departures (AoAs/AoDs) exploiting the inherent sparsity of angle domain channels. Then, the discussion is extended to the wideband mmWave hybrid full-dimensional MIMO-OFDM systems, in which the introduced closed-loop sparse CE scheme leverages the channel sparsity in both angle and delay domains to enhance performance. This scheme includes the downlink and uplink CE stage, where the multi-dimensional unitary ESPRIT (MDU-ESPRIT) algorithm is used to estimate the AoAs at user devices (UD) in downlink and estimate the AoDs and UDs’ delays at the base station in uplink. Furthermore, the channel parameters acquired at the two stages are paired by a maximum likelihood method and the path gains are then estimated using the least-square approach. The spectrum estimation techniques in hybrid MIMO ensure the super-resolution estimations of the AoAs/AoDs and delays with low training overhead. Finally, the superiority of the considered schemes over state-of-the-art approaches is verified by numerical experiments.