Research On Computing Power Distribution Mechanism Based On Edge-cloud Collaboration

With the rapid development of IoT technology,network traffic has grown exponentially,and terminal devices on the edge side carry emerging services that require high quality and low latency,such as interactive games,autonomous driving,and face recognition.Limited by local computing resources and storage resources,terminal equipment cannot support such services.At the same time,the traditional cloud computing model cannot meet the real-time requirements of the business due to long link transmission.As a new computing mode,edge computing deploys the computing resources and storage resources of the central cloud to the edge of the network,processes and stores data at the edge of the network,and provides near-end services for users,which can solve the problem of realtime business needs in cloud computing.However,when the number of users and the scale of data are further expanded,edge nodes tend to be saturated.It is very important to formulate an efficient computing power allocation strategy to make full use of the limited resources of edge nodes.Therefore,this paper introduces the central cloud into the edge computing architecture as a global scheduler,and studies the computing power allocation in the edge computing and cloud collaboration scenarios.The main work is as follows:(1)A computing power allocation scheme to improve tasks with different priorities is proposed.In this paper,in the edge computing architecture,the central cloud is introduced as a global scheduler for overall management,forming a threetier system architecture of "end-edge-cloud".Modeling,including tasks,task offloading request information,task priority,local terminal,edge server,central cloud and system processing cost,takes the weighted sum of task processing delay and energy consumption as the system processing cost.Aiming at the computing power distribution under the edge-cloud collaborative system,this paper introduces a new fitness value rule to improve the computing power distribution for tasks with different priorities,and combined with artificial bee colony algorithm(Artificial Bee Colony algorithm,ABC)to optimize the system processing cost under the "endedge-cloud" architecture.(2)An improved computing power allocation algorithm for edgecloud collaboration is proposed.For the computing power allocation algorithm under the current edgecloud collaborative system,the artificial bee colony algorithm has higher development ability when facing high-dimensional problems,that is,when the task volume is large.Weakness and poor convergence,this paper proposes an improved artificial bee colony algorithm based on chaos with single and multi-dimensional dynamic population(CIOMABC).Firstly,a new crossover operator is designed to improve the problems of the single multi-dimensional dynamic population strategy Artificial Bee Colony Algorithm(OMABC)in the existence of scout bees,which have poor optimization of inferior solutions and low solution search efficiency.Then,based on the characteristics of randomness,ergodicity and regularity of chaotic sequences,the chaotic local search operator is integrated into the framework of the IOMABC algorithm,that is,the artificial bee colony algorithm based on chaos with improved single and multi-dimensional dynamic population(CIOMABC).Convergence accuracy and search accuracy are further improved.(3)Design and implement a simulation platform based on EdgeCloudSim for algorithm verification.Based on the EdgeCloudSim edge computing simulation tool,this paper builds a simulation platform for edge-cloud collaborative computing power distribution,including task generation modules with different priorities,task scheduling modules,network communication modules and task processing module,etc.,and input the computing power allocation algorithm based on task priority proposed in this paper and the improved algorithm in this paper into the task scheduling module.Finally,according to the analysis and comparison of the experimental results obtained by the simulation,it proves the correctness of the fitness value rule based on task priority and the effectiveness of the algorithm improvement proposed in this paper.