An Artificial-Intelligence (AI) Assisted Mm-Wave Doherty Power Amplifier with Rapid Mixed-Mode In-Field Performance Optimization

Abstract

This paper presents an artificial-intelligence (AI) assisted “intelligent” millimeter-wave (mm-wave) Doherty power amplifier (PA) architecture that achieves robust adaptive operation over antenna VSWR. A built-in machine-learning core utilizes online function estimation and reinforcement learning algorithms for dynamic Doherty performance optimization. As a proof of concept, a 3-bit mixed-signal Doherty PA (MSDPA) at 28GHz is used as a hardware platform to demonstrate the performance improvement with the AI core. Over 2:1 antenna VSWR variation, simulations using extracted Doherty PA schematic show up-to 4.9 dBm P 1dB improvement and up-to 7% PAE improvement at P 1dB .