Nonlinear Precoding for Quantized Outphasing RF Chains in Massive MIMO Systems

Implementing large antenna arrays for massive multiple-input multiple-output (MIMO) base stations requires transmit radio frequency (RF) chains with low complexity and high power amplifier (PA) efficiency. One approach to achieve this goal is to relax PA linearity and to digitally compensate for some of the resulting non-linear effects such as an increased error vector magnitude (EVM) and adjacent channel leakage ratio (ACLR). The required compensation method can be implemented in a non-linear precoding module that exploits the large spatial degrees of freedom in a massive MIMO system. In this paper, we propose using outphasing RF chains with low-resolution phase shifters to generate constrained amplitude and phase-modulated signals with very low hardware complexity. To improve transmit power efficiency and to control the ACLR and EVM, we develop a novel non-linear precoding algorithm that incorporates the quantized phase-shift constraints in the outphasing RF chains. The achieved flexibility to trade-off EVM and ACLR in favor of a higher PA efficiency (and vice versa) is demonstrated by means of simulations.