A Residual Selectable Modeling Method Based on Deep Neural Network for Power Amplifiers With Multiple States

Abstract

A traditional power amplifier (PA) behavioral model typically represents one specific operating state of the PA. As the number of states of PA increases, the depth of the behavioral model based on the deep neural network (DNN) deepens. However, the deepening of the DNN may result in decreased model accuracy. To solve this issue, this letter proposes a residual selectable modeling method to obtain the residual DNN (RDNN), which can be used to build the multistate PA behavioral model. Experimental results show that the multistate PA model constructed by the proposed method can improve the accuracy of the DNN-based PA model. Also, the model accuracy does not decrease with the deepening of DNNs.