Advanced MOSFET Model Based on Artificial Neural Network

In this work, we develop a novel MOSFET model for circuit simulation purpose. Instead of using traditional physics-driven model, such as BSIM model, our work uses ANN to model the electrical behavior of the transistor. With unique pre and post-processing procedures, the ANN is trained to model the drain current precisely under various applied voltage, device size and temperature. The model is further successfully implemented in SPICE through Verilog-A language. Both n-type and p-type MOSFETs show good fitting between the BSIM and ANN models. Eventually, an inverter and ring oscillator based on ANN model are demonstrated with static and transient simulations, showing good agreement with the results from BSIM model.