Research On The Computing Resource Management And Scheduling Method For Industrial Edge Nodes

With the intelligent transformation of traditional industry,the number of industrial field devices joining the network continues to increase,and intelligent applications continue to spread,resulting in an increasing demand for computing resources in the industrial field.As a computing model closer to the edge of industrial networks,edge computing has gradually become a research focus in the field of industrial Internet of Things.Different from the centralized computing model of cloud computing,edge computing can provide close-range,low-latency,and highly reliable localized computing resources for industrial field devices.Due to the uneven distribution of industrial field equipment and the diversity of applications,the amount of tasks offloaded to the edge side varies greatly,which is most likely to cause load imbalance in edge computing.Therefore,in order to solve this problem,this thesis studies the computing resource scheduling strategy for industrial edge nodes,and realizes the flexible scheduling of computing resources among industrial edge nodes.The main research work of this thesis is as follows:1.On the basis of fully investigating the research status at home and abroad,this thesis proposes a computing resource virtualization architecture for industrial edge nodes.The architecture uses container technology to abstract and integrate computing resources to form a virtual resource layer to provide support for resource sharing.At the same time,starting from the requirements of real-time resource monitoring,intelligent management,and flexible scheduling,the core functional modules are researched and designed based on this architecture.2.Aiming at the problem of unbalanced load among industrial edge nodes,a computing resource scheduling strategy based on load prediction is proposed.This strategy mainly includes load prediction model and resource scheduling mechanism.Among them,the load forecasting model combines exponential smoothing and gray system theory,which can have good forecasting effects in periods of steady load and sudden load changes.The resource scheduling mechanism determines the scheduling priority by considering the degree of node overload and the priority of industrial computing,which effectively alleviates the imbalance of load and meets the high real-time demand of industrial computing.3.This thesis builds a test platform for industrial edge node resource scheduling by using Raspberry Pi,CC2530 wireless communication module and PC.In this thesis,the proposed resource scheduling strategy based on load prediction is tested and comparatively analyzed.The test results show that after adopting this scheduling strategy,the resource utilization rate of the cluster is increased by 7.12%,and computing resources are used more effectively;the resource balance is reduced by6.05%,and the use of CPU resources and memory resources within the node is more balanced;the load balance of the cluster is reduced by 15.69%,and the load imbalance has been greatly improved.