Genetic algorithm based extraction method for distributed small-signal model of GaN HEMTs

Abstract

In this paper, an improved small-signal model parameter extraction method, using genetic algorithm (GA), is presented and implemented for GaN HEMT. The GA optimization is used to generate a high quality reliable starting values for the elements of distributed model. This value are then refined using local optimization technique to find optimal value for each model element. The developed extraction method is validated by simulating S-parameter measurements of a 8x125-µm gate width GaN HEMT over a wide bias range.