A Simplified Digital Predistortion Model for Fully-Connected Hybrid Beamforming Massive MIMO Transmitters Using Discrete Particle Swarm Optimization Technique

In this paper, a pruning method using discrete particle swarm optimization (DPSO) is proposed to reduce the complexity of digital predistortion (DPD) for fully-connected (FC) hybrid beamforming (HBF) transmitters. In the DPSO algorithm, the position of each particle corresponds to a subset of model terms, and the subset that maximizes the model accuracy will be found through the cooperation between different particles. A simulation on a 4-stream 64-element FC HBF array is carried out, where the coefficients in each DPD model are reduced by approximately 50% while adjacent channel power ratio (ACPR) and normalized mean square error (NMSE) remain almost unchanged.