Design And Implementation Of Dependency Task Scheduling System In Edge Computing

With the rapid development of mobile internet technology,a variety of compute-intensive and delay-sensitive applications such as AR/VR,autopilot,industrial control,etc.are gradually emerging.These applications are usually composed of multiple interdependent tasks,which can be expressed as directed acyclic graph(DAG).However,due to the limited computing and communication resources of mobile devices,it is difficult to support such applications well.In addition,the unstable network environment and the high delay caused by long-distance data transmission also make it difficult for cloud computing services to meet the quality of service requirements of these applications.Therefore,edge computing emerges as the times require.It sinks computing,storage,network and other resources to the edge of the network to provide mobile devices with low latency and low-power services.How to efficiently schedule such compute intensive and delay sensitive DAG applications in edge computing systems has become a key problem to be solved.To solve the above problems,this paper proposes a DAG task scheduling algorithm based on container configuration.The specific work is as follows:1)In view of the problem of functional configuration of DAG application services,by introducing a container-based service configuration mechanism,a single edge node provides services for various types of tasks,and based on this,a four-tier network architecture model of application layer,container layer,edge node layer and remote cloud layer is constructed.2)In view of the problem of resource competition incentive of edge network containers,According to the dependency of DAG tasks,the application request model and the request delay model are established.With the request completion rate as the optimization goal,a task scheduling algorithm based on improved ant colony algorithm is designed.The task priority is dynamically adjusted based on the request deadline,and urgent tasks are assigned first to reduce the task delay.Then,the pheromone mechanism of the ant colony algorithm is used to find the scheduling scheme of the global optimal solution of the system.Based on the above work,this paper designs and implements a simulation verification system of edge computing DAG task scheduling algorithm.The scheduling algorithm proposed in this paper is built in the system,providing functions such as task generation,building edge environment,algorithm import,DAG task scheduling simulation,etc.This paper uses the Web to implement the front-end interface of the simulation system,uses Apache Airflow and Mininet to build the network model and algorithm library,and performs the function demonstration and system test.The system simplifies the process of users building edge computing network simulation environment through graphical interface,and provides users with a safe,reliable and scalable simulation verification system of edge computing task scheduling algorithm.