| With the popularization of the fifth generation mobile networks(5G)technology,the metro carriage is a typical high-density and high-capacity complex channel propagation environment.In recent years,the multipath propagation characteristics of millimeter wave in the metro environment has become a research hotspot of scholars at home and abroad.Metro carriage is a narrow and closed environment with multiple scatterers,and its channel environment is very complicated.As an efficient machine learning algorithm,neural network algorithm can better reveal the complex propagation characteristics of metro carriage channel.Therefore BP neural network algorithm based on particle swarm optimization is adopted in this thesis in order to study the millimeter wave channel parameters in the metro carriage environment,such as path loss,received power and delay spread.At the same time,k-means algorithm is used to analyze the clustering characteristics of millimeter wave channel in this thesis.It provides theoretical basis for the construction and optimization of 5G wireless communication system in metro carriage.The main contents of this thesis are as follows:(1)In this thesis,the research background of metro carriage communication is introduced at first.And the research status of millimeter wave frequency band,metro carriage channel environment,neural network algorithm,channel clustering characteristics are introduced.Furthermore,millimeter wave multipath propagation characteristics,ray tracing methods and neural network algorithms are also introduced.Then,the computer realization process of BP neural network algorithm based on particle swarm optimization is introduced in detail.(2)Aiming at the problem that the traditional BP neural network algorithm is easy to fall into the local optimal solution,the BP neural network algorithm based on particle swarm optimization is adopted in this thesis,the particle swarm optimization can calculate the initial threshold and weight,and search in a larger space.The optimized algorithm is used to predict the path loss parameters of the SIMO channel,the experimental results show that the prediction accuracy of the optimized algorithm is significantly higher than that of the unoptimized algorithm.Finally,the correctness and effectiveness of the optimized algorithm are verified.(3)Based on the optimized BP neural network algorithm,the modeling and simulation of the millimeter wave SIMO channel of metro carriage is carried out in this thesis.The channel parameters can be accurately predicted,which indicates that this algorithm is suitable for predicting and analyzing the millimeter wave channel parameters of large scale channel parameters and small scale channel parameters.Then PSO-BP neural network algorithm is used to model and analyze the MIMO channel characteristics,such as the received power,root mean square delay spread and angle spread,the results reveal the multipath component characteristics of MIMO channel.(4)Based on the prediction results of PSO-BP neural network algorithm,multipath clustering characteristics of metro carriage millimeter wave channels in line-of-sight and non-line-of-sight conditions are studied in this thesis.Combining with the k-means clustering algorithm,the clustering characteristic of angle-delay domain and angle-delay-power domain are analyzed.Then,the DB index is used as the optimal clustering index to obtain the optimal number of clusters,and the cluster center coordinates and the number of subpaths contained in each cluster are determined. |
