Research On Key Technologies Of Resource Allocation And Data Access In Edge Computing

In the era of Internet of Everything,massive amounts of data are generated at the edge of the network,causing a sharp increase in the transmission pressure of network data.At the same time,the real-time and privacy protection requirements of new applications are gradually improved.It is difficult to meet the application requirements by using traditional centralized cloud computing to communicate,calculate and process data.In response to this problem,edge computing proposes to use distributed network nodes that are closer to the data generation end to provide distributed computing,storage and control services,thereby reducing the amount of data transmission and responding to user service requests in real time.In recent years,the development of edge computing has raised many research challenges.Therefore,this paper researches on resource allocation and data access in edge computing.The main research work is summarized as follows:First,design PARA which means Performability-Aware Resource Allocation in the edge computing datacenter.Facing the situation of limited resources in the edge data center,aiming at the shortage of coarse-grained allocation of existing resource allocation strategies,this paper puts forward using the characteristics of application service degradability,perceiving the performability of the bidding application in different degradation modes,taking maximizing the performability of the overall application as the optimization goal,constructs a resource allocation model with micros-ervice as the allocation granularity,and then realizes Fine-grained allocation of resources in the edge computing data center.Furthermore,the algorithm for solving the problem is designed.Compared with the scheme with the application as the allocation granularity,PARA can improve the overall performability of the bidding applications by 14.21%,and improve the satisfactory of the applications by 33.36%.Secondly,design LPARA which means Long-term Performability-Aware Resource Allocation in the edge computing data center.For the scenario where the resource-oriented data center is limited,the existing resource allocation strategy is lack of consideration of the dynamic characteristics of resource requirements.This paper proposes to consider the dynamic changes of resource requirements,analyze the impact of resource reallocation on the application performability in the subsequent period,take the optimization of long-term and effective performability as the optimization goal.And on the basis of the PARA,the resource allocation model is extended from a single time period to multiple time periods.And we design the algorithm to solve it.The experimental results show that compared with the static resource allocation strategy,LPARA can improve the long-term performability of the bidding application by 12.11%-19.45% and reduce the resource reallocation overhead by 46.4%-52.3%.Thirdly,design ENSURE which means an Efficie Nt and Sec URE searchable encryption mechanism based on edge computing.For the mobile terminal data access scenario,the traditional searchable encryption mechanism is not suitable for mobile terminals.It proposes to use the trusted edge server at the network edge to optimize the searchable encryption mechanism from the architecture.That is,it offloads the computing intensive tasks performed by the terminal in the process of searchable encryption and the privacy sensitive tasks performed by the cloud to the edge server.Through theoretical analysis and experimental prototype construction,ENSURE can cut off the cloud to obtain keyword information during data access,shorten the access delay by 15%-49% and reduce the energy consumption of mobile terminal by 38%-69%.Finally,design EPFL which means an Edge-based Privacy-preserving Federated Learning mechanism based on edge computing.Data for medical scenes is inaccessible,and a system model for adapting to the needs of medical scenes is proposed.The concept of edge computing is used to reduce the resource occupation of medical terminal equipment participating in training.Based on the concept of federated learning,medical data is invisible but available;differential privacy strengthens the privacy protection in this collaborative learning scenario,effectively reducing the external attacker's reverse access to user sample data.