Experimental Evaluation of PA Digital Predistortion Based on Simple Feedforward Neural Network
The paper presents the results of experimental studies on evaluation of employing digital predistortion based on simple feedforward neural network for linearization of microwave power amplifiers. The influence of the number of neurons in the hidden layer, the number of delayed input samples at the input of neural network, as well as the number of samples taken for learning a neural network were studied and discussed in the paper. The main goal of this work was to establish the minimal configuration of the neural network which can be used for linearization of power amplifiers excited by wideband and high PAPR signals, e.g. LTE. The results obtained for neural networks were compared with the results obtained for the conventional predistortion method based on memory polynomial.
Paper
Full text
Experimental Evaluation of PA Digital Predistortion Based on Simple Feedforward Neural Network
Semantic Scholar · Engineering · 2020
Abstract
The paper presents the results of experimental studies on evaluation of employing digital predistortion based on simple feedforward neural network for linearization of microwave power amplifiers. The influence of the number of neurons in the hidden layer, the number of delayed input samples at the input of neural network, as well as the number of samples taken for learning a neural network were studied and discussed in the paper. The main goal of this work was to establish the minimal configuration of the neural network which can be used for linearization of power amplifiers excited by wideband and high PAPR signals, e.g. LTE. The results obtained for neural networks were compared with the results obtained for the conventional predistortion method based on memory polynomial.