Study On Millimeter Wave Propagation Characteristics In Tunnel Environment Based On Support Vector Machine

With the application of 5G,millimeter wave technology has been a research hotspot due to its characteristics of wide band,narrow beam,low delay and high rate,which can meet the requirements of 5G.The tunnel environment is an important scene for people to travel and traffic.It is a narrow and limited space.There are multiple reflections,transmission and diffraction,resulting in obvious multipath fading and waveguide effect.Due to the advantages of SVM in solving the problems of small sample,nonlinear and high dimension,it is suitable for predicting the characteristics of multipath fading channel and mine tunnel.In this thesis,the propagation characteristics of millimeter waves are studied in the tunnel environment based on SVM.Meanwhile,genetic algorithm is used to optimize the parameters,which effectively improves the accuracy of the model based on SVM and provides a theoretical basis for wireless signal coverage in tunnel environment.The main research contents and contributions of this thesis are as follows:(1)In this thesis,the research status is summarized,including the measurement and modeling of millimeter wave channel in tunnel environment and channel modeling based on machine learning.The theoretical basis of SBR/ IM method and SVM is introduced.And then the technical route of prediction of propagation characteristics in tunnel environment based on SVM is described.(2)The simulation of long straight tunnel is carried out to verify the correctness of simulation results.Based on the measured results in the known literature,the SVM method is used to train the path loss,and the genetic algorithm is used to optimize the parameters.The results show that the accuracy can be effectively improved after parameter optimization.By comparing the prediction results of the three methods,including SBR/IM method,SVM method and SVM method based on genetic algorithm,it shows that the SVM method based on genetic algorithm has a higher prediction accuracy.(3)Taking Seoul Metro Line 8 tunnel as an example,the channel propagation characteristics of curved tunnel at 28 GHz are studied,including received power,path loss and delay spread.The SVM method based on genetic algorithm optimization is used to predict propagation characteristics,and the result is satisfactory.K-means algorithm is used for clustering,and there is an obvious boundary between clusters.(4)The herring tunnel environment in Xuanwu Lake Tunnel is simulated,and the propagation characteristics of Lo S and NLo S environments are compared to study the characteristics of largescale fading,small-scale fading and multipath clustering.The path losses in LOS and NLOS environments are modeled.It offers a basis for the modeling of millimeter wave channel in tunnel environment.