Residual Network Based Direct Synthesis of EM Structures: A Study on One-to-One Transformers

Anonymous

Keywords: direct synthesis, neural network, efficient prediction

TL;DR: We showcase a direct synthesis of EM transformers using neural networks: efficiently predicting circuit geometric designs based on required circuit specification.

Abstract: We propose using machine learning models for the direct synthesis of on-chip electromagnetic (EM) passive structures to enable rapid or even automated designs and optimizations of RF/mm-Wave circuits. As a proof of concept, we demonstrate the direct synthesis of a 1:1 transformer on a 45nm silicon on insulator (SOI) process using our proposed neural network model. Using pre-existing transformer s-parameter files and their geometric design training samples, the model efficiently predicts target geometric designs based on desired circuit specification.