Multiobjective Bayesian Optimization for Active Load Modulation in a Broadband 20-W GaN Doherty Power Amplifier Design

This paper proposes a multiobjective Bayesian optimization strategy to deal with the complex matching problem of designing a broadband 20-W gallium nitride Doherty power amplifier (DPA). Applying the proposed method, the DPA’s matching networks are optimized to align the desired and the realized impedance trajectories both at saturation and power backoff from 1.5 to 2.4 GHz. By combining a programming environment with the commercial electronic design automation (EDA) software, the optimization process is implemented automatically. Comparison designs reveal that the proposed method outperforms the built-in optimizer of the commercial EDA software for broadband DPA designs. When tested with a single-carrier 20 MHz long-term evolution signal, the measured results show that the DPA achieves an average efficiency of 45.3%–53.6% at 7 dB backoff across the band, with the adjacent channel power ratio levels better than −45.6 dBc after using digital predistortion.