Doherty amplifier optimization using robust genetic algorithm and Unscented Transform

Abstract

A robust genetic circuit optimizer using Unscented Transform and Non-dominated Sorting Genetic Algorithm-II is presented. The algorithm provides significant computational cost reduction compared to Monte Carlo method. The Unscented Transform permits circuit performance uncertainties determination from components uncertainties, thus, a search through robustness can be approached. Besides reduced computational effort, a plethora of possibilities offered to circuit designer by the multi-objective search can be observed as result. Doherty power amplifier was used as case study.