Research On Truthful Mechanism Based Resource Allocation And Pricing Algorithms In Edge Computing

The emergence of edge computing solves the problems of traditional cloud computing in real-time,bandwidth limitation,and application service quality.However,with the rapid increase in the number of wireless access users and the continuous evolution of application scenarios,in a shared resource competition environment,how to maximize social welfare to encourage service providers to share resources while ensuring the credibility of resource allocation and pricing mechanisms,is a new challenge facing resource allocation in the edge computing environment.In response to this challenge,the dissertation aims to maximize social welfare and conducts in-depth research on the resource allocation and pricing mechanism of edge servers in the edge computing environment based on credible mechanism design.The main research work completed in the thesis is as follows:(1)For the problem of continuous task allocation on edge servers,this thesis takes the video surveillance scenario of continuous task submission as an example,and designs a resource allocation algorithm based on residual value strategy and a pricing algorithm based on critical value theory,which expresses the problem It is a nonlinear integer programming model with multiple resource constraints to solve the continuous task allocation problem in a competitive environment.The theory proves the credibility of the mechanism and individual rationality,and the experiment also shows that this method has more advantages than existing research in terms of execution time,social welfare and resource utilization.(2)For the problem that the user’s task submission requirements change with time,this thesis takes the time-varying cloud virtual machine resource allocation scenario as an example,and proposes an integer programming model for offline allocation,and designs the optimal and greedy approximation two kinds of credibility Auction mechanism to solve the problem of resource allocation and pricing in a competitive environment.Among them,the optimal auction mechanism is a dynamic programming(DP)resource allocation and pricing algorithm based on the Vickrey-Clarke-Groves(VCG)mechanism;the greedy auction mechanism is a resource allocation and pricing algorithm based on the predominant resource ratio sorting and dichotomy.It is proved by theory that both algorithms meet the requirements of credible mechanism,and the approximate ratio of the greedy allocation algorithm is calculated and proved.Comparative experiments with the latest research show that the proposed algorithm has higher social welfare,higher service user ratio and shorter execution time.(3)For the online resource allocation problem of real-time task submission faced by computing resource providers,this thesis takes the online allocation scenario of time-varying multi-dimensional resources as an example,and proposes an integer programming model for time-varying multi-dimensional resource allocation.Strategies and advantages of resource ratio sorting of resource allocation and pricing algorithms.The credibility of the mechanism and individual rationality are proved by theory.Comparative experiments show that this method has the characteristics of high social welfare,high resource utilization,and short execution time.(4)For the problem of online resource allocation involving shared bandwidth and heterogeneous server clusters in edge computing scenarios,this thesis describes the problem as an integer programming model for online multi-dimensional resource allocation with shared resources,based on dominant resource ratios,waiting period strategies and the server selection strategy designs a resource allocation algorithm,and designs a pricing algorithm based on the critical value theory.The credibility of the mechanism and individual rationality are proved theoretically,and the influence of the proportion of shared resources,the scale of demand and related strategies on resource allocation is analyzed through experiments.