Novel Frequency Dependent Neural Network Behavioural Model of a Power Amplifier

A novel neural network-based implementation of behavioural models is presented, which aims to decrease simulation time and increase accuracy of existing models. The neural network is used to expand the parameter range of conventional behavioural models of a power amplifier. The neural network maps the transfer characteristics of the various behavioural models (based on the amplitude modulation-amplitude modulation (AM-AM), amplitude modulation-phase modulation (AM-PM) functions) over a specified RF frequency range. The model is used to evaluate the Adjacent Channel Power Ratio (ACPR) for a Wideband CDMA (W-CDMA) signal at different RF carrier frequencies.