Novel neural network-Bessel transform for behavioural modelling of a power amplifier

Abstract

A novel neural network based behavioural model is presented, which aims to decrease simulation time and increase the accuracy of existing models. The neural network model is an alternative to existing behavioural models, using a new neural network-Bessel transform to obtain the behavioural transfer function of the Power Amplifier (PA). The neural network then maps the transfer characteristics of the PA, based on the amplitude modulation-amplitude modulation (AM-AM), amplitude modulation-phase modulation (AM-PM) functions. The model is used to evaluate adjacent channel power for a Wideband CDMA (W-CDMA) signal.