Gaussian Process surrogate model for variability analysis of RF circuits
Thong Nguyen, José E. Schutt‐Ainé · 2020
Non-intrusive methods for studying processes involving variables changing such as design optimization, manufacture variation etc. require evaluations of the quantity of interests for a numerous times. These methods, hence, rely on an accurate surrogate model of the process under study. Gaussian Process (GP) is a well-known non-parametric modeling technique for surrogate modeling. This paper explores the effectiveness of GP to model RF applications. The analysis of a milimeter-wave bandpass filter is presented to illustrate the method.