Piecewise Small Signal Behavioral Model for GaN HEMTs based on Support Vector Regression

Abstract

In this paper, a machine learning (ML) technique based, piecewise small-signal behavioral modeling methodology for Gallium nitride (GaN) high electron mobility transistors (HEMTs) is presented. Support vector regression (SVR), as one of the core algorithm of ML technique is employed. Compared with equivalent small-signal circuit model, which extracting the precise parameters of the circuit, the proposed piecewise SVR technique analyze and modeling the measured data by means of mathematical method. The verification is carried out on an 8 × 125 μm GaN HEMT with a 0.25 μm gate feature size. The SVR model shows a big improvement in accuracy of S-parameter behavior prediction when compared with conventional piecewise equivalent circuit model. The relative error of S-parameter prediction is greatly reduced.