An Embedded-Structured Convolutional Neural Network for Efficient RF Device Behavior Model Extraction

In this paper, a novel embedded-structured convolutional neural network based approach to extract nonlinear behavior RF device model is proposed. The model features a unique embedded neural network architecture designed to extract and store information related to polynomial coefficients, enabling continuous extraction of the nonlinear behavior coefficients. The effectiveness of the proposed method was experimentally validated on a real GaN device. The results demonstrate that the proposed method makes more efficient use of the measurement data, and achieves more efficient nonlinear transistor modeling with reduced data requirements.