A 2.86Gb/s Fully-Flexible MU-MIMO Processor for Jointly Optimizing User Selection, Power Allocation, and Precoding in 28nm CMOS Technology

Abstract

In 5G networks, as growing data usage exponentially, mobile operators need to increase network capacity. To increase the spectral efficiency of massive multiple-input multiple-output (MIMO) system, it is essential to enlarge the number of co-scheduled user equipments (UEs). As increasing the number of co-scheduled UEs up to the number of base station (BS) antennas, the conventional linear precoding schemes such as zero-forcing and maximum ratio transmission show poor capacity, as shown in Fig. 1. As a result, joint user selection, power allocation, and beamforming schemes, including the rank-adaptation zero-forcing (RA-ZF) and generalized power iteration precoding (GPIP) algorithms, are proposed for large-scale massive MIMO systems. However, the prior works on massive MIMO baseband architectures [1]-[4] are no longer suitable for these advanced algorithms; because they do not consider user selection or power allocation. Consequently, it is crucial to develop energy-and computationally efficient BS architecture that realizes the advanced algorithms to achieve a high spectral efficiency gain.