A Riemannian Geometry-based Hybrid Beamforming for mmWave Systems

Since millimeter wave communications offer orders of magnitude more spectrum than existing frequency bands, it has been regarded as a crucial enabling approach for 5G systems. Beamforming in millimeter wave systems cannot be accomplished using digital techniques, unlike standard multiple input multiple output networks, because only a limited number of analog-to-digital converters and mixers can be handled due to their cost and power consumption. A hybrid beamforming transceiver architecture combining an analog and digital beamformer has drawn much interest lately as a more affordable option. However, the best way to construct these hybrid beamformers has not been completely determined. This study will offer effective alternate minimization methods for a hybrid beamforming structure, namely the manifold-based geometric fully-connected antennas, in order to approach the performance of the completely digital beamformer, which is complicated. The suggested iterative algorithm outperforms the baseline hybrid beamforming method, namely the MMSE-MP algorithm, by a considerable margin (about 6.8 % for a certain circumstance from 27 to 29 Nats/sec/Hz at 10 dB transmitting power), according to simulation data of bandwidth efficiency. Furthermore, comparing the simulation results of the two hybrid beamforming structures will provide crucial design information based on the suggested methods.