An efficient optimization method of RF passive components using RBF model

There is a growing interest in the synthesis of radio-frequency (RF) passive elements in electronic design automation communities. In this paper, we propose an efficient optimization method of RF passive components. We first build models for performances extracted from S parameters of passive components by Radial Basis Function (RBF) to ensure the accuracy and the geometric parameters of passive components are incorporated as input variables. Then Pcell is used to generate the layout of inductors or transformers in 65nm process. Finally, the model is optimized by differential evolution algorithm to obtain solutions meeting constraints. Experimental results imply that our proposed method can achieve a target inductance with up to 0.76% relative error of inductance when compared to EM simulations for inductors and up to 0.87% relative error of inductance for transformers.