296 / 2020-01-05 20:17:00

Efficient Channel Estimation via Redundancy-Reduction for Hybrid MmWave Massive MIMO Systems

sparse channel estimation; redundancy reduction; hybrid structure; mmWave massive MIMO

IEEE SAM 2020: Special Sessions > Sparse and Low-Rank Methods for Array Processing and Wireless Communications

In this paper, a novel channel estimation solution is developed for hybrid mmWave massive MIMO systems. Unlike the conventional compressive sensing (CS) based channel estimation which essentially solves an underdetermined problem through high-computation optimization solutions, we transform such an originally underdetermined problem to a determined counterpart, which is then solved by low-complexity deterministic formulations. It is the inherent redundancy of the channel covariance that enables the problem reformulation. By utilizing the redundancy-reduction (RR) transformation, we formulate the channel covariance and the observed covariance in the linear mapping of a transformed covariance in the RR domain. Further, we design an efficient precoding principle to guarantee such a linear mapping matrix with full column rank. As a result, the estimate of the transformed covariance can be directly obtained from the observed covariance through the least squares (LS). Next, angle estimation methods can be employed for the path angle pairs retrieval, and then the path gain can be obtained sequentially through LS. Compared with conventional CS based approaches, the proposed RR-LS method becomes much more efficient as the number of antennas increases. It is not only thanks to the deterministic solution of the transformed problem, but also due to the reduced dimension from the RR transformation.