Network Application Task Decomposition And Scheduling For Cloud Edge Terminal Computing Force Cooperation

With the rapid development of mobile communication network technology and the continuous emergence of intelligent network applications,the number of intelligent user terminals increases rapidly,and network applications produce more and more resource requirements.The explosive growth of users’ demand for computing power has promoted the formation of a cloud-side-end three-level computing power deployment architecture in the network.The cloud computing power is rich but the delay is high.The edge and terminal computing power are close to users,but the resources are limited,and some nodes will be inaccessible due to failure,offline and high load.The effective collaboration of computing forces at all levels can meet the ubiquitous intelligent computing needs of various network applications.However,the current task decomposition technology has a single decomposition method,which can not adapt to the multi-level computing resource nodes in the network environment and can not provide different task decomposition schemes according to different network scenarios.At the same time,there are inevitably unstable edge nodes in the network scene.When the edge node fails,the tasks working in the edge node need to be relocated to other nodes,resulting in additional overhead and delay.The existing task scheduling research has not considered the problem of task migration caused by the failure of unstable edge nodes.Aiming at the problem that the task decomposition method is single and can not provide adaptation to multi-level computing resource nodes according to different network scenarios,this paper proposes a multi granularity task decomposition algorithm to adapt to multi-level heterogeneous computing resources at the cloud edge.By considering the characteristics of task elements,the relationship between task elements,and the network resource environment,the multi granularity task decomposition method based on weighted graph clustering is used to reasonably divide the computing tasks of network applications.The experimental results show that the proposed algorithm has good performance in optimizing resource allocation and adapting to node resources.Aiming at the problem that it is necessary to quickly select the target node for the subtasks of offline nodes in the network scene with unstable edge nodes,this paper proposes a task scheduling algorithm based on multi granularity decomposition to minimize the cost of task migration.This method fully considers the migration cost of subtasks generated by unstable nodes offline and selects the appropriate decomposition scheme through the granularity hierarchical relationship between subtasks,and the divided subtasks are scheduled to make full use of the computing resources distributed in the network.The simulation results show that in the network environment with unstable edge nodes,compared with the comparison scheme,the proposed scheme can achieve lower total task completion time and cost in different scenarios of edge node stability,and can migrate with lower migration time and cost after a node failure.To sum up,with the development of complex network applications and the continuous development of fragmented network resources,the research on the multi granularity task decomposition of multi-level computing power adaptation at the cloud side and the scheduling scheme for minimizing the cost of task migration in this paper is of great significance to make full use of network fragmented resources and reduce the total time and cost of task completion.