Signal-noise neural network model for active microwave devices

Abstract

A new method for concurrently modelling the small-signal and the noise performance of active microwave devices is proposed. Here, the device is modelled by a black box whose small signal and noise parameters are evaluated through a neural network, based upon the fitting of both of these parameters over the operational bandwidth of the device. On using the concurrent modelling procedure, it has been found that, not only can the small-signal performance be simulated accurately, but also the prediction of noise performance is in much better agreement with measurements than those of recent published models.