Enhancing spectral efficiency in sub-connected hybrid beamforming systems via beam pattern design

Abstract

Fifth-generation systems with a sub-connected (SC) hybrid beamforming architecture are highly beneficial due to their low complexity and high energy efficiency. The fixed number of antenna elements per subarray renders array gain per data stream limited for SC architecture. Enhancing beamforming gain in this architecture is a challenging problem. This work introduces a novel hybrid precoder designed to enhance spectral efficiency in a multi-user SC hybrid beamforming system by leveraging spatial information about users. An initial beam training approach is employed to gather users’ spatial information, which is further used to form user batches. Each batch is assigned a single coarse steering angle and can have multiple refined steering angles. These angles are used to design the analog and digital beamforming vectors for the proposed hybrid precoder. Users within the same batch employ beam pattern multiplication to efficiently utilize a higher number of antenna elements per data stream, leading to enhanced beamforming gains. The analytical beamforming gain in the presence of user batches has been derived and a relationship between spatial location, user batch formation, and achievable beamforming gain is established. It is observed that the average beamforming gain is maximized when the number of antenna elements per subarray approaches the total number of radio frequency chains. Simulation results demonstrate a significant improvement in beamforming gain and validate the superiority of the proposed design over the conventional SC approach in terms of high beamforming gain and improved spectral efficiency.