Research On Resource Allocation Mechanism For QoE-Oriented Edge Computing

In recent years,edge computing technology has emerged as the times required to solve the drawbacks of traditional centralized data processing,such as high latency,high energy consumption,and network congestion.Edge computing technology uses the processing power and storage resources on edge devices to process and store data,reducing the dependence on cloud computing centers and achieving faster response and lower energy consumption.Therefore,how to improve resource utilization and optimize performance indicators through edge computing resource allocation and task scheduling has become a hot and challenging research topic.However,most of the existing research results focus on resource allocation methods for Quality of Service(QoS),which are usually based on network performance indicators and cannot fully reflect the specific needs of users.They often lose effectiveness when processing application requests with different user needs.In contrast,resource allocation methods for Quality of Experience(QoE)focus more on user experience.They can optimize resource allocation based on users’ differentiated needs,providing a better user experience and improving user satisfaction.Based on this,the main work of this thesis is as follows:(1)Firstly,a QoE-oriented resource allocation method for edge computing is proposed for the multiple-user single server edge computing scenario.By modeling edge computing scenarios,analyzing task models,local computing models,edge computing models,and general QoE evaluation models,this thesis proposes a resource allocation optimization problem to maximize the overall QoE.A two-stage resource allocation method for QoE is proposed to address this problem.A low-complexity adaptive resource scheduling method is proposed in the local computing phase,which can quickly generate optimal computing resource scheduling and offloading strategies based on QoE requirements.In the computing phase of an edge computing system,a deep reinforcement learning method based on the the improved Deep Deterministic Policy Gradient(DDPG)is proposed.Compared with the traditional DDPG method,this method has a faster convergence speed and higher reward value.The analysis of simulation results shows that the proposed two-stage resource allocation method can better meet the QoE needs of users and achieve overall QoE maximization compared to other algorithms.(2)Secondly,the above scheme is applied to the smart city scenario,and a QoE-oriented edge computing resource intelligent allocation system is designed and implemented.This system can provide optimized resource allocation schemes tailored to different task needs and the current resource status of users to improve resource utilization and user satisfaction.In module design mainly includes a registration and login module,smart city map display module,data statistics module,edge computing server management module,user management module,task management module,and resource allocation module.Among them,the smart city map display module intuitively displays the geographic location and status information of smart city edge computing servers and users through maps.The data statistics module visually displays the data generated by the system through charts.The task management module implements the personalized creation of differentiated QoE requirement tasks.The resource allocation module can provide an optimized resource allocation plan based on users’ differentiated QoE needs,providing a better user experience and improving user satisfaction.