Efficient PA modeling using neural network and measurement setup for memory effect characterization in the power device

This paper proposes an artificial neural network (ANN) model for power amplifiers (PAs), which exhibits memory effects. This model is based on a complex-valued multi-layered perceptron (MLP) time delay neural network (TDNN). The developed PA models are based on TDNN with unity and nonuniform time delay taps between instantaneous and previous input signal of the PA. Modeling results for both types of PA models are compared to show the advantage of the newly developed model. Furthermore, a measurement setup for analyzing and measuring the nonlinearity and memory effects in GaN active power device is also proposed.