Threshold Optimized DVR Model for RF Power Amplifier Using Particle Swarm Algorithm for 5G Application

In this work, the thresholds of the decomposed vector rotation (DVR) model for radio frequency (RF) power amplifiers (PAs) which based on canonical piecewise linear (CPWL) function, has been analyzed and optimized. The particle swarm optimization (PSO) algorithm is employed, and the thresholds of the DVR model is optimized to achieve the optimal performance. The basic theory and modeling procedure of the proposed technique are presented. Both Doherty PA (DPA) and sequential load modulated balanced amplifier (SLMBA) are used for experimental validation. Compared to the conventional Volterra-based model, the standard DVR model, the proposed PSO-based DVR (PSO-DVR) model present huge improvement. Compared with existing simultaneous perturbation stochastic approximation (SPSA) algorithm-based threshold optimization method, the optimization iteration number can be greatly reduced, which means the optimization efficiency is improved.