Radial basis function neural network models for power-amplifier design

A method for learning the dynamic responses of a nonlinear power-amplifier with a radial basis function (RBF) neural network is presented. The training data of the RBF neural networks are samples of the input and output waveforms of the power-amplifier. A new training scheme employing efficient gradient-based optimization methods is developed to train the RBF neural networks model. Simulation results are presented to demonstrate the superiority of the RBF neural networks compared with recurrent neural networks and other networks.