Improved Particle Swarm Optimization Algorithm to Optimize Millimeter Wave Channel Modeling of RBF Neural Network

This paper studies the radial basis function (RBF) neural network optimized by the improved particle swarm optimization (PSO). This neural network is used for channel modeling in complex office environments and compared with the channel modeling results of the basic RBFNN. Predicted and analyzed the path loss, delay spread and power angle distribution of the wireless channel in a 60 GHz indoor office environment. The comparison of multiple sets of parameters shows that the algorithm improves the prediction accuracy of RBFNN and speeds up its convergence better.