This paper proposes a fully connected neural network (FCNN) model to reduce non-linear distortion in power amplifiers using a basis generation function. The model comprises a feedforward neural network (FNN) and a convolutional neural network (CNN), both of which are designed using polynomial expansion. The FNN generates the basis function, while the corresponding weights are generated by the CNN. The FNN takes the basic elements that form the bases as input, based on the requirements prescribed by dynamic deviation reduction (DDR) model. The proposed model updates the coefficients of the hidden layer concurrently during the training of the FNN and CNN models. The complex multiplication of their outputs yields the trained IQ signal. The proposed model is trained and tested using 300MHz and 400MHz broadband data in an OFDM communication system. And it achieves an ACPR of less than -48dB by 100MHz integral bandwidth for both the training and test datasets.
A Basis Function Generation Based Digital Predistortion Fully Connected Neural Network Model of RF Power Amplifier
2023-10-10
1401866 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch