Research On DAG Task Scheduling Algorithm In Edge Computing Environment

The widespread use of Internet of Things(Internet of Things)devices and the rapid development of artificial intelligence have brought enormous pressure and challenges to the traditional cloud computing architecture.In order to reduce network pressure,edge computing task scheduling technology has become a very cost-effective solution to reduce network congestion and improve user experience quality.Scheduling strategy should not only satisfy the dependencies between subtasks,but also greatly improve the system performance.Therefore,it is one of the urgent problems to propose a reasonable and effective scheduling strategy.This paper studies the scheduling strategy of complex application tasks in edge computing.Different from general tasks,we model such complex applications as Directed Acyclic Graph(DAG),which can satisfy the topological relationship of vertices in the graph,Guaranteed dependencies between subtasks.Firstly,the DAG task is modeled and analyzed,the optimization problem is defined from a global perspective,a scheduling algorithm based on priority and sub-task dependencies is proposed,and the scheduling priority order and scheduling strategy of sub-tasks are designed to ensure that the dependencies of each sub-task are reduced.The delay for the system to complete all DAG tasks.Then,from the perspective of the delay of a single IoT device completing the DAG task,the game idea is introduced,and a DAG task scheduling strategy based on the game idea is proposed,so that the DAG task of each IoT device can optimize its own resources while competing for resources.Task completion delay.The two algorithms proposed in this paper are based on the premise of DAG task guaranteeing dependencies,one is analyzed from the perspective of system completion delay,and the other is analyzed from the perspective of single device task completion delay.Both algorithms effectively reduce the delay,reduce reduce energy consumption and improve system performance.The research work of this paper is as follows:(1)Study the scheduling model of typical complex application tasks in the edge computing environment,first model such complex tasks as DAG tasks,then study the system composed of multiple IoT devices and multiple multi-core edge servers,and set the Subtasks can be calculated locally or scheduled to be calculated on a certain processing core of other different edge servers.The subtask scheduling priority queue model is established from the global perspective,and a task scheduling algorithm based on priority and subtask dependency is proposed,which can minimize the system delay while maintaining the dependency of subtasks.(2)Based on the game idea,a game-based DAG task scheduling algorithm is proposed from the perspective of the interests of a single IoT device.Each IoT device wants to formulate a good scheduling strategy to minimize the delay in completing the task itself,but due to limited resources,the game theory idea is introduced,and each IoT device formulates a schedule to make the IoT device complete according to the scheduling strategy of the rest of the devices.delay-minimizing decision.As the game progresses for a limited number of iterations,IoT devices select servers for task scheduling according to the game results each time,and constantly adjust the scheduling strategy.Finally,the scheduling decisions of all devices will reach a stable equilibrium state.