Constrained Tensor Decomposition-Based Hybrid Beamforming for Mmwave Massive MIMO-OFDM Communication Systems

Hybrid beamforming design for OFDM systems is hugely challenging since their analog precoders and combiners are shared among all subcarriers. This paper proposes a novel two-stage joint hybrid precoder and combiner design for maximizing the average achievable sum-rate of frequency-selective millimeter-wave massive MIMO-OFDM systems. In the proposed approach, the analog precoder and combiner design is formulated as a constrained Tucker2 tensor decomposition problem, which allows maximizing the sum of the effective baseband gains over every subcarrier while suppressing inter-user and intra-user interferences. The solution is obtained by a projected alternate least-square-based algorithm, which is suitable for both single-user (SU) and multiuser (MU) systems. The digital precoder and combiner for SU systems is obtained from the effective baseband channel' SVD. In contrast, for MU systems, it is obtained from the regularized channel diagonalization method, which balances multiuser interference and noise suppression. Numerical simulation results validate the effectiveness of the proposed method.