Research On Mm-wave Wireless Channel Modeling Based On Genetic Algorithm And Grey Method

With the wide application of wireless technology,millimeter wave communication has become a research hotspot in recent years.Channel measuring,modeling and simulation technology play an important role in the design of millimeter wave communication system.At the same time,the combination of genetic algorithm(GA)and support vector machine(SVM)algorithm in machine learning overcomes the difficulties of traditional channel research,such as difficulties in complex high-dimensional data processing and high computational power requirements for redundant data processing.The introduction of grey theory can make the model suits insufficient data,increase the prediction accuracy,reduce the amount of data required,and reduce the cost of millimeter wave channel measurement.The main works of this paper are as follows.Firstly,based on genetic algorithm,the optimized support vector machine(GA+SVM)model and grey genetic optimization model(GGOM)are established for millimeter wave wireless channel parameters prediction.The theoretical derivations of the two models are also carried out.Secondly,based on the 60 GHz indoor channel measured data,genetic algorithm is introduced to optimize the support vector machine(SVM)algorithm.Specificity,GA+SVM model is used to reduce the dimension of high-dimensional data.The feasibility of this method in channel parameter modeling is also verified.Thirdly,based on the 28 GHz indoor channel measured data,the channel with insufficient data is modeled by the grey genetic optimization model(GGOM).The simulation data is generated by Quasi-Deterministic Radio Channel Generator(QuaDRiGa)platform.It used to verify the reliability of the GGOM under different data distribution modes and different data numbers.The prediction output results of the model are analyzed and compared based on the experimental measurement data.The results show that both the prediction performance and accuracy of the optimized GA+SVM and GGOM are better than that of the traditional BP neural network.In this paper,the millimeter wave wireless channel with high-dimensional and insufficient sample data is studied comprehensively.Based on genetic algorithm,the GA+SVM model and grey genetic optimization model are established,and the feasibility of genetic algorithm and grey theory in the field of wireless channel modeling is verified.It has important reference value for 5G millimeter wave channel simulation and system design.