Research On Offloading Mechanism Of Industrial Internet Computing For Edge Computing

With the acceleration of the modern industrial automation process in China,the surge in the number of intelligent manufacturing machines is unprecedented.Gradually,manufacturers transformed the workshop into a network of integrated physical systems,which called "smart factories." In addition,technological breakthroughs in the field of cloud computing,the Internet of Things and distributed computing lead to the generation of massive data,and the demand for computing resources are also increasingly urgent,while traditional industrial networks cannot meet the demand of computing resources and data in various industrial application scenarios.To address these issues,edge computing emerged as a newly emerging computing mode,by placing services such as computing and storage near end of the device,which can improve the response speed to industrial equipment services and reduce the network cost.This paper analyzes an offloading network consisting of multi-edge nodes with tasks to handle and multiple industrial devices,potentially offloading tasks to multiple computational access points,or using remote cloud servers as additional offloadable points.The goal is to make offloading and resource allocation decisions that minimizes system costs.This paper makes the offloading decision through the network bandwidth estimation,and when facing the large number of industrial tasks that need to be offloaded,the edge resource allocation is particularly important in the computational offloading.This paper achieves the purpose of the edge server load balance through a game theory.The main tasks are as follows:This paper studies and analyzes the development and architecture of Industrial Internet(IIoT)and edge computing,combined with the demand for real-time data and computing resources,and adopts SDN-based architecture to conduct edge compute offloading through flexible traffic strategy,which intelligently manages the network by configuring centralized controllers.The task of industrial equipment for compute offloading in the Industrial Internet is divided into three stages,namely,the offloading decision-making stage,the resource allocation stage,and the task execution stage.After deciding how to offload,the server-side resources are assigned directly for task execution,so the resource assignment and task execution are grouped together.When offloading decisions,this paper estimates the available bandwidth resources by Bayes and change point detection,which calculates the cost of local and remote execution,and finally performs the task at a small cost.When calculating the unloaded resource allocation,this paper proposes a combination of integrated container selection,integration,and the migration scheme to improve the quality of service in a distributed edge computing environment.At the same time,the computing resource allocation at the edge end is realized through the Stackelberg game of multi-host and multi-slave,and the problem of resource pricing is adopted to be solved between industrial equipment and edge resource providers.Finally,the load balancing of workload between industrial equipments and edge nodes can not only relieve the pressure of high burst data volume,but also solve the problems of high task execution delay and high energy consumption.Through simulation experiments verified the rationality and correctness of the algorithm and the network architecture,meanwhile proved the stability requirements while minimizing the system cost.