Edge Computing Based Resource Scheduling Optimization Method for Intelligent Manufacturing Workshop

Abstract

The intelligent manufacturing workshops have the characteristics of heterogeneous resources, dynamically arriving tasks, strict deadlines, and the frequently changing states of machines and networks. In order to resolve the corresponding issues of delay, congestion, and instability of scheduling, the paper puts forward a dynamic resource allocation and task scheduling approach based on end-edge-cloud cooperation. Terminal, edge, and cloud nodes, virtual machine capacity, task size, transfer delay, deadlines, and energy consumption are all considered in the process of modeling the computing resources and manufacturing tasks. A multi-objective model is established in order to enhance task completion, resource utilization, satisfaction of service level agreements (SLA) and reduce delay and energy consumption. In the proposed approach which consists of three phases: task sorting, resource pre-allocation, and dynamic scheduling, tasks are adaptively reallocated according to changing of load, network, and node states. According to the experiments done using the simulator named EdgeCloudSim, when 1,000 tasks arrive each minute, this approach can maintain the average delay under 2.7 s and a success rate of more than 93%. The optimal edge offloading ratio is around 0.74.