A Wide-band Power Amplifier based on Genetic Algorithm and Direct Layout Optimization

In this paper, a non-dominated sorting genetic algorithm assisted by Gaussian Process (GP-NSGA-II) is proposed. The Gaussian regression prediction is introduced into the conventional NSGA-II to improve optimization speed by reducing the number of operations.This new method is further integrated with a direct layout optimization technique by combining several commercial electronic design automation (EDA) tools. In virtue of these strategies, layout can be designed automatically. To validate the effectiveness, a wideband GaN power amplifier covering 2-3 GHz is implemented based on a 10W HEMT device. Under input power of 29 dBm, the obtained power-added efficiency (PAE) of the designed circuit is higher than 60% throughout the operational frequency band, and the output power is larger than 41.6 dBm.