RBF neural network channel modeling of millimeter wave based on adaptive particle swarm optimization algorithm

Abstract

In this paper, the method of radial basis function (RBF) in machine learning is applied to the modeling of millimeter-wave channel based on millimeter wave indoor wireless channel measurement data. A RBF neural network channel parameter prediction model based on adaptive particle swarm optimization (APSO) is established, and the prediction results of the traditional RBF algorithm are compared. Specifically, the RBF model optimized based on APSO is used to learn and predict the characteristics of large-scale channel parameters (LSCP), such as path loss and delay spread. The results show that the predicted channel parameters of the APSO-RBF model is consistent well with the actual measured value. The learning performance and prediction effect of this algorithm are better than the traditional RBF algorithm, that is, the RBF algorithm has a smaller root-mean-square error (RMSE), and the predicted curve has a larger fitting degree with the original measured curve. In addition, the APSO-RBF model has good adaptability to the change of channel parameters in the case of large data fluctuation, and can achieve good prediction effect for 5G millimeter wave channel parameters.