A new macromodeling approach for nonlinear microwave circuits based on recurrent neural networks

Author

Fang, Yonghua ; Yagoub, Mustapha C E ; Wang, Fang ; Zhang, Qi-Jun

Author_Institution

Dept. of Electron., Carleton Univ., Ottawa, Ont., Canada

Volume

48

Issue

12

fYear

2000

fDate

12/1/2000 12:00:00 AM

Firstpage

2335

Lastpage

2344

Abstract

A new macromodeling approach is developed in which a recurrent neural network (RNN) is trained to learn the dynamic responses of nonlinear microwave circuits. Input and output waveforms of the original circuit are used as training data. A training algorithm based on backpropagation through time is developed. Once trained, the RNN macromodel provides fast prediction of the full analog behavior of the original circuit, which can be useful for high-level simulation and optimization. Three practical examples of macromodeling a power amplifier, mixer, and MOSFET are used to demonstrate the validity of the proposed macromodeling approach

Keywords

Jacobian matrices; MOSFET circuits; backpropagation; circuit CAD; circuit optimisation; circuit simulation; dynamic response; gradient methods; microwave circuits; microwave mixers; microwave power amplifiers; nonlinear network synthesis; recurrent neural nets; CAD; Jacobian matrix; MOSFET; RF mixer; RFIC power amplifier; backpropagation through time; dynamic responses; fast prediction; full analog behavior; gradient-based optimization; high-level simulation; input waveforms; macromodeling approach; nonlinear microwave circuits; output waveforms; recurrent neural networks; time-domain macromodel; training algorithm; Circuit simulation; Computational modeling; Design optimization; Microwave circuits; Microwave devices; Microwave propagation; Neural networks; Nonlinear circuits; Recurrent neural networks; Training data;

fLanguage

English

Journal_Title

Microwave Theory and Techniques, IEEE Transactions on

Publisher

ieee

ISSN

0018-9480

Type

jour

DOI

10.1109/22.898982

Filename

898982