Deep Learning-Driven Inverse Design of Doherty Power Amplifiers Using Pixelated Combiners and Dual-State Impedance Synthesis
Researchers have developed a novel method for designing Doherty power amplifiers using deep learning, specifically deep convolutional neural networks (CNNs) combined with genetic algorithms (GA) and dual-state impedance synthesis. This approach addresses the complex integration of load modulation, impedance matching, and phase compensation within the output combiner. Prototypes fabricated using this methodology achieved a saturated output power exceeding 44.2 dBm and a peak drain efficiency above 71.2% in the 2.6-2.8 GHz range, with a drain efficiency of 64% at a 6-dB back-off level. After digital predistortion, adjacent channel leakage ratios better than -51.3 dBc were recorded. AI
IMPACT This research could lead to more efficient and powerful wireless communication systems by improving amplifier performance.