Modeling Techniques For RF And Microwave MOSFETs Based On Neural Network

The MOSFET(Metal-Oxide-Semiconductor Field-Effect-Transistors)plays an important role in the circuits for its low power consumption,mature manufacturing process,high integration,and good coupling ability with analog circuits.In the process of designing and developing circuits,in order to cut the cost,replacing the device with accurate model is an effective way.At the same time,the development of device technology puts higher demands on the device model.Artificial neural network can meet the demands both in speed and accuracy of the model.With the development of the computers' speed and the algorithm,the advantages of artificial neural networks will become more apparent.This thesis focuses on the research of 120 nm MOSFETs modeling technology based on artificial neural network.MOSFET models based on neural network are proposed here.The main content of this thesis includes: 1)the parameter extraction process of the MOSFET small signal model is elaborated.At the same time,the calculation formula of the intrinsic component is deduced.2)in the process of modeling and simulating the S-parameters of the device,two neural network models are proposed: a neural network model based on knowledge and a neural network model based on direct method.Two neural network models are quite different from the traditional model.They not only have faster modeling speed,but also can accurately predict S-parameters under different bias voltages.3)the parameter extraction process of the STATZ model is described.4)based on the STATZ model,a method for establishing an integral neural network model and a method for establishing a differential neural network model are proposed.It lays the foundation of establishing a uniform neural network model that can accurately predict the small-signal and large-signal characteristics of the device...