Hybrid block diagonalization precoding for multi-user weighted sum-rate maximization

This paper studies large-scale multi-input multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) communications in a broadband frequency-selective channel, where a massive MIMO base station (BS) communicates with multiple users equipped with multi-antenna. We develop a hybrid precoding design to maximize the weighted sum-rate (WSR) of the users by optimizing the digital and the analog precoders alternately. For the digital part, we employ block-diagonalization to eliminate inter-user interference and apply water-filling power allocation to maximize the WSR. For the analog part, the optimization of the PSN is formulated as an unconstrained problem, which can be efficiently solved by a gradient descent method. Numerical results show that the proposed block-diagonal hybrid precoding algorithm can outperform the existing works.