Neural network predistortion technique for digital satellite communications

Digital satellite communication channels are equipped with nonlinear amplifiers (such as travelling wave tube amplifiers (TWT)) which exhibit severe performance degradation when they work near saturation. The paper proposes an adaptive neural network (NN) predistortion technique to process the transmitted data. The technique is based on two steps. In the first step, the TWT amplitude and phase conversions are modeled. In the second step, an approximation of the TWT inverse transfer function is used to implement the predistorter. The approach is applied to 16-QAM transmissions and is shown through computer simulations to out-perform classical Volterra series techniques