Neural network based power amplifier dynamic modeling for wireless communications

In this paper we present the Neural Network (NN) based dynamic power amplifier (PA) modeling with memory effect. We developed this model with system generator for DSP by Xilinx that could be implemented on DSP chip. The advantage of our system generator based model is to develop highly parallel systems with the most advanced FPGAs, providing system modeling and automatic code generation from Simulink and MATLAB. Our real time modeling method can be adapted for any kind of latest signal type such as cdma-2000 and W-CDMA without any modification of the model and can be adapted to any environmental change such as temperature variation in PA without modify the model. That means our real time model is self adaptable. By using this method we can do a modeling dynamically the non linearity of the PA including memory effects for realistic modulation signals inputs. In this paper, by using our modeling architecture we demonstrate to have an almost same dynamic AM-AM and AM-PM curves of PA.