Study On Edge Computing Scheme For Industrial Internet

Edge computing is a comprehensive technology that integrates advanced network technology,data technology and so on across communication,computing,automatic control and other fields.In the Industrial Internet scenario,edge computing can provide assistance in the reception,storage,processing and calculation analysis of real-time data.In addition,it can also make corresponding decision analysis and issue decision results to ensure the real-time industrial application.Similar to traditional IoT,Industrial Internet requires edge computing in data computing and security.The difference is that the real-time and reliability requirements of the Industrial Internet are higher.Real-time,reliability and security are the basic requirements of Industrial Internet.For real-time and reliability of data transmission,IEEE 802.1 Task Group has proposed the TSN standard.In the Industrial Internet scenario,TSN has a higher scalability than the traditional industrial bus.TSN does not need to consider the docking and conversion of different bus protocols.It is modified on the basis of the standard Ethernet.Under the premise of ensuring compatibility,it provides deterministic message transmission to ensure the deterministic delay and reliability of the network.Using the computing and network capabilities of edge computing and the real-time characteristics of TSN is a new idea in the research of edge computing for Industrial Internet.Therefore,this paper mainly studies the edge computing scheme combined with TSN and key algorithm technology,and the specific contents are as follows:(1)This paper analyzes the actual demand of the edge computing scheme for Industrial Internet and considers the indispensability of TSN network for the future Industrial Internet.Combined with Standard 802.1Qcc centralized configuration model,an edge computing scheme combined with TSN is proposed.The scheme mainly considers the configuration and management functions of TSN,involving time synchronization,traffic scheduling,and the corresponding network configuration and management.We also give the module design of CUC node and CNC node in the scheme,explain the protocol and data interaction involved,and then give the scheme working flow.In addition,we also design a validation experiment of our edge computing scheme combined with TSN,verify the effectiveness of the scheme's time synchronization and the performance of traffic scheduling.The latter includes the verification of RTT delay in the horizontal direction and the verification of calculation and configuration performance in the vertical direction.Different scheduling algorithms have different effects on the computing time of CNC nodes.Under the same conditions,Tabu Search Algorithm based on flow Tabu Table is better than ILP Scheduling Algorithm in runtime performance.The test results of configuration time show that the main factor affecting the configuration time is the configuration modification and effective operation of TSN switch itself,and the impact of network hops is very small.Considering that the off-line configuration and dynamic configuration of TSN network are inevitably involved in the practical application,the edge nodes have greater advantages.(2)ILP-based Scheduling and Tabu Search Algorithm,effects of different routing algorithms on TSN scheduling are analyzed.On the basis of flowspan as the scheduling performance target,we propose the maximum network load,that is,MNL as the second routing performance target,and quantitatively study the relationship between MNL and flowspan.Under the given simulation conditions,it is verified that different routing algorithms have different effects on TSN scheduling,and the corresponding MNL values are also different.Tabu search algorithm based on tabu list has similar MNL valuse comparing to ILP algorithm,and the former is slightly higher than the latter in flowspan.Tabu search algorithm is also suitable for dynamic routing scheduling.We verify the proposed dynamic tabu search algorithm and find that it can greatly reduce the running time of the algorithm under the proper K value and MNL required by users.