Semi-Blind Detection in Hybrid Massive MIMO Systems via Low-Rank Matrix Completion

In massive multiple-input multiple-output (MIMO) systems with hybrid analog/digital architectures, large training overhead is required for conventional pilot-only methods to estimate channel accurately before detecting data. To reduce the training overhead, a semi-blind detection method is proposed for data detection without knowing channel in an uplink multi-user system. The main idea is to exploit the received signal corresponding to both the pilot and data payload for channel estimation or data detection via a low-rank matrix completion formulation. The leveraged low-rank property stems from the fact that the number of active users $K$ is typically much smaller than the number of antennas $N_{a}$ at a base station and the number of time slots $T_{c}$ in a coherence interval. Compared with the pilot-only method, the number of pilots required is reduced from an order of $N_{a}$ to $K$ . Two iterative algorithms are introduced to solve the low-rank matrix completion problem: regularized alternating least squares and bilinear generalized approximate message passing. We further extend the semi-blind detection method to systems with low-resolution analog-to-digital converters. Simulation results show that the proposed methods achieve significant performance gain over the pilot-only method with reduced training overhead for hybrid massive MIMO systems in various settings.