Research On 5G-R Network Slicing Resource Scheduling And Access Control Algorithm Based On Machine Learning

The rapid development of national science and technology makes our country continue to be in the forefront of the global development of 5G,5G mobile communication technology has gradually been applied to large areas,Railway wireless communication system also has entered the era of 5G-R(5G for Railway)from the GSM-R(GSM for Railway)era.As a key enabling technology for 5G,the network slicing technology based on network function virtualization can automatically allocate different virtual computing resources to users according to different user requirements to meet diverse service requirements of 5G-R networks.This paper firstly defines the network architecture and business requirements of 5G-R,and analyzes the architecture of related open source wireless communication system and the principle of general software radio peripherals.Secondly,based on the actual measurement results,this paper quantifies the differences of computing resources consumed under different bandsizes,and obtains the mapping relationship between computing resources and communication resources.After the problem is formalized as a knapsack problem,the paper aims to maximize the utilization of system resources,and takes the requirements of each slice and the total bandwidth of slices under 5G-R service scenarios as constraints.The optimal allocation of network slice communication bandwidth resources is realized based on machine learning algorithm.Meanwhile,based on software radio technology,software defined network and network function virtualization technology,this paper designs and implements 5G-R zero on-site testing platform,and infuses controlled IP traffic into different test environments based on physical layer basic function interoperability test,vehicle-ground wireless communication transmission performance test and system stability test.To obtain throughput,delay,packet loss and other key indicators,effectively verify the feasibility,reliability and stability of the zero on-site testing platform;Further,based on the field test platform of 5G-R zero,the enhanced Mobile Broadband is designed in this paper.Enhanced mobile broadband)slicing and ultraReliable and Low Latency Communication slicing.The test results of network slicing test platform based on 5G-R zero on-site testing platform show that the slicing scheme designed in this paper is feasible.The simulation results of resource scheduling and intervention control algorithm of network slicing show that different network slicing have different demands on communication resources and computing resources.By evaluating the three optimal allocation algorithms and comparing the static average allocation algorithm,the simulation results show that the Lagrange multiplier method is the optimal network slicing resource scheduling and access control algorithm.