Angle-Domain Beam Grouping-Based Beam Selection for Massive MIMO Analog Beamforming with NOMA in Cellular Downlink

Abstract

This paper considers an appropriate beam selection method in downlink massive multiple-input multiple-output (mMIMO) analog beamforming (BF) with non-orthogonal multiple access (NOMA) based on intra-beam superposition coding. To obtain a sufficient system-level throughput gain by employing NOMA in downlink mMIMO with analog BF, it is essential to select beams where each beam has many appropriate users that are to be non-orthogonally multiplexed while mitigating the interference among simultaneously selected beams. In addition, analog BF with NOMA suffers from limited channel state information since the feedback interval of the signal-to-noise power ratio (SNR) of candidate beams from users is longer than the scheduling interval. We propose a beam selection method based on an angle-domain beam grouping (ADG), which utilizes a beam group that contains multiple neighboring-angle beams. ADG selects beams that are appropriate for applying NOMA by considering the neighboring beams as well as the candidate beam itself during the beam selection process. ADG is also effective in reducing inter-beam interference by setting the neighboring beams to non-selectable beams. Furthermore, we propose a method that reuses the signal-to-interference and noise power ratio (SINR) measured for user-scheduling and transmission-rate control every beam-update interval into the beam selection. This SINR reuse method further improves the system throughput since the beam selection accuracy over time is improved. Computer simulations show that the proposed methods significantly improve user fairness while maintaining the system efficiency compared to the conventional methods, that they are especially effective when the beam update interval is long, and that the system throughput is significantly improved when using NOMA.