Research On Offloading Optimization Of Edge Computing For Mobile Network

Mobile edge computing has become a key technology to realize the vision of the 5th generation mobile communication system.By deploying computing and storage nodes on the wireless access network side,it effectively solves the conflict problem of high application service quality requirements and insufficient equipment computing resources and energy resources.This thesis focuses on the optimization of computing offloading in mobile edge computing systems.From the perspectives of edge service providers and users,it aims to meet user task performance requirements while increasing the potential benefits of service providers and reducing user-side total Energy consumption.First,this thesis studies the problem of computing offloading in scenarios where a single mobile edge server has limited resources and is densely deployed.Unlike most of the existing literature that optimizes indicators from the perspective of users,this thesis considers the potential benefits of service providers for scenarios where multiple servers are deployed and the user association is unknown.Taking into account the difference of tasks and the heterogeneity of resources,this thesis proposes a task value evaluation quantitative plan for the income part,so that service providers can make reasonable prices for different task requirements.Taking into account the heterogeneity of resources,this thesis considers resource consumption in multiple dimensions for the cost part,including communication resources,computing resources and energy resources.Then,this thesis uses joint optimization of user associations,offloading decision and resource allocation decision variables to improve the potential benefits of service providers as much as possible on the premise of meeting user performance requirements.In order to solve this problem,this thesis proposes a joint optimization iterative framework,which decomposes the original problem into user association problem and joint offloading strategies and resources allocation problem,and solves them based on matching theory and block coordinate descent algorithm.The simulation results indicate that the proposed algorithm has a faster convergence speed and has a good performance compared to the benchmark algorithm in improving the potential revenue of the service provider.Secondly,this thesis studies the edge computing offloading scenario where server resources are relatively saturated and transmission links become the main performance bottleneck.In the current research,the application of intelligent reflecting surface in edge computing is still lacking,and they focus on optimizing delays and the sum computational bits,and there are deficiencies in the research on user-side energy consumption.This thesis considers the edge computing network scenario assisted by intelligent reflecting surface and non-orthogonal multiple access.This thesis starts from the perspective of reducing user-side energy consumption while combining non-orthogonal multiple access to further improve spectrum efficiency.By jointly optimizing the amount of unloaded data,the user's CPU frequency,upload power,upload delay,and the reflection coefficient matrix of the intelligent reflecting surface,the user-side energy consumption is minimized while meeting the tasks'tolerable delay.This thesis uses block coordinate descent and continuous convex approximation method to divide and transform the original problem to solve it.The simulation results show that the proposed algorithm has good convergence and effectiveness,and the intelligent reflecting surface performs well in edge computing scenarios,which plays a positive role in improving the transmission link state and reducing the total energy consumption on the user side.