Behavioral Modeling of GaN Power Amplifier Based on CG-BPNN

In order to quantify and predict the nonlinear impact of power amplifier (PA) on signal transmission efficiently, the conjugate gradient backpropagation neural network (CG-BPNN) is introduced for behavior modeling in this paper. With the gradient and history information updating from the search the directions, it can reach the global optimal solution with the relatively less iterations. As revealed in the modeling for the GaN HEMT PA and GaN Doherty P A, it can characterize the behavioral properties accurately. Meanwhile, its unstable generalization ability is also significantly improved.