Weighted Sumrate Maximizing Hybrid Precoder for MU-MISO-OFDM

In multi-antenna transceivers with all-digital beamforming, the number of radio frequency (RF) chains necessarily equals the number of antennas, which entails a large hardware complexity and power consumption when deploying large antenna arrays. Hybrid beamforming, which splits the beamforming into an analog part operating at RF and a digital part operating at baseband, allows using as few RF chains as the number of data streams. A further reduction in the RF hardware cost is achieved by imposing additional constraints on the analog beamformer, such as single-phase-shifter implementation and limited connectivity between the RF chains and antennas. In this contribution we consider a downlink multi-user MISO OFDM communication system and present a novel approach for obtaining the linear hybrid precoder that maximizes the weighted sumrate (WSR) under the above constraints on the analog part. Unlike the WSR maximizing hybrid precoders from the literature, our approach exploits the uplink-downlink duality and performs a gradient-based optimization over a suitable manifold which describes the search space as determined by the imposed constraints. The proposed approach gives rise to a WSR maximizing precoder with a better trade-off between performance and computational complexity compared with WSR maximizing precoders from the literature. From this WSR maximizing precoder, two reduced-complexity heuristic precoders are derived, which outperform state-of-the-art heuristic precoders from the literature, in terms of spectral efficiency and/or computational complexity. The performance of the WSR maximizing precoder provides a useful theoretical benchmark for any heuristic precoder with the same constraints on the analog part.