Research On Self Optimization Of 5G Heterogeneous Network Based On Community Throughput And Load Balancing

The development and popularization of 5G heterogeneous networks have brought faster network speeds and wider service coverage.The number of internet users is continuing to increase,and the distribution of users is becoming more uneven,there will be more hotspot areas where multiple users gather.Traditional user allocation methods cannot meet the communication needs of all users in such scenarios,resulting in a decrease in cell throughput and an extremely imbalanced base station load in the network.Finding a solution that can dynamically optimize network parameters based on real-time network data to improve cell throughput,satisfy the communication requirements of most users,and solve the problem of base station overload in this new environment is the main focus of this study.Self-Organizing Network(SON)technology can automatically configure,manage,and optimize the network to improve its reliability.Through SON technology,the parameters and configurations of different types of base stations in 5G heterogeneous networks can be autonomously adjust based on network operating conditions,to provide optimal wireless coverage and network performance,thereby increasing network capacity and coverage range,improving the communication experience of users and overall network performance.This thesis proposes a dynamic user allocation algorithm that combines machine learning and SON to dynamically allocate base stations to users based on different network environments,improving cell throughput and effectively improving the communication experience of users.Based on this algorithm,a user dynamic allocation algorithm that combines load balancing technology is further proposed in this thesis,which not only improves cell throughput but also solves the problem of base station overload.The specific research content of this thesis includes:(1)In order to improve throughput while taking into account the communication experience of users,this thesis designs a user satisfaction algorithm based on delay,Reference Signal Received Power(RSRP),and throughput.(2)Based on the algorithm in(1),this thesis proposes a dynamic user allocation algorithm that combines SON and reinforcement learning technology to dynamically adjust the connection relationship between users and base stations based on real-time the demand feedback of users,performance indicators of base stations,the location of each user,and network load conditions to maximize the total throughput of the cell.This algorithm not only improves cell throughput but also strives to meet the important indicators required for each user’s communication,thereby improving the communication experience of users.(3)Based on the algorithm in(2),this thesis also proposes a user dynamic allocation algorithm that combines load balancing technology.This algorithm implements base station load balancing based on Qlearning algorithm,which improves the overall communication quality of users in hotspot areas and thus enhances the overall cell throughput and the QoE satisfaction of users.This thesis has conducted simulations for two user distribution scenarios(uniform distribution and Poisson distribution).The simulation results show that the proposed algorithm can effectively improve cell throughput and reduce the number of overloaded base stations,providing reference for the combination of SON and 5G heterogeneous networks.