Research On Resource Scheduling Algorithm Under 5G New Network

Users' demand for communication systems continues to increase,and new industrial application scenarios such as remote surgery,Internet of Vehicles,and ultra-high-definition video transmission are also emerging in an endless stream.5G communication,as a solution to many practical problems in the communication field,has caused revolutionary changes in all walks of life as soon as it is commercialized.With the gradual popularization of the latest generation of cellular communication technology,how to solve a series of load balancing problems in its advanced network architecture and realize the reasonable scheduling of system resources are becoming a current research hotspot.In response to the above problems,this article has done relevant research on the resource allocation problems in 5G communication networks.Based on the analysis of its communication principles,the communication simulation system under the corresponding scenarios has been built and improved.At the same time,combined with mature algorithms in reinforcement learning,A resource scheduling algorithm with relatively better real-time performance and performance is designed.The main research work of this thesis are as follows.(1)The principle of air interface resource slicing of the wireless access network is studied.By analyzing the characteristics and resource requirements of the three basic scenarios in 5G,a network slicing system model based on the performance of service quality is designed.This model focuses on improving the traditional caching scheme and packet loss mechanism,introducing an early random detection strategy based on priority,and designing a targeted user quality evaluation scheme,which lays the foundation for subsequent algorithms research.(2)A slicing resource allocation scheme based on reinforcement learning is designed.By combining the specific communication scenarios of Deep Q-Network and 5G access network,a DQN-based slice allocation algorithm is proposed.At the same time,the algorithm is optimized by using the priority-based experience playback strategy and Double DQN.Experiments show that compared with the traditional slice polling algorithm and the system without slice adaptation,this algorithm has obvious performance advantages.(3)In order to solve the resource allocation problem under the new SDN architecture in the core network,A network load balancing algorithm based on reinforcement learning is designed.The concrete controller design plan is given,and the overall framework and the design ideas of each module are explained in detail.By testing the system functions,the superiority of the algorithm proposed in this paper is verified in related scenarios.