Research On Multi-queue Forwarding In Time-sensitive Network

Time-sensitive network realizes the deterministic transmission technology of industrial Ethernet in a simple and open architecture of standard Internet.The deep integration of the new generation of technological revolution and industry makes the Industrial Internet one of the key technologies supporting the transformation of the industrial system.Its application scenarios are no longer limited to single-point digitalization,but extend to scenarios where multiple elements and agents are connected to achieve interconnection between platforms.The industrial data transmitted by the industrial Internet has changed from a simple data and control traffic model in a single production line to a complex traffic model with multiple-service coexistence,massive data,and differentiated requirements.However,the existing Cyclic Queuing and Forwarding mechanism only provides a single level of service for traffic,which limits the types of traffic that can be scheduled,and cannot take into account both high-deterministic transmission and high schedulability.How to realize the mixed transmission that can provide differentiated services for traffic,taking into account the quality of service and scheduling ability.Based on the shortcomings of existing scheduling mechanisms in above scenario,this paper studies the multi-queue forwarding mechanism in time-sensitive network under the background of industrial Internet.Firstly,this paper transforms the existing mechanism into a multicycle one,forming a Multi-cycle Cyclic Queuing and Forwarding mechanism,enhancing its adaptability and flexibility in dealing with widely different transmission needs.The Multi-cycle Cyclic Queuing and Forwarding mechanism avoids the loss caused by making trade-offs among multiple metrics,and adapts to scenarios with different proportions of cyclic streams.It is an inclusive scheme that can be applied to multiple environments without changing the scheduling mechanism.Secondly,a mathematical model is established for the Multi-cycle Cyclic Queuing and Forwarding scheduling problem,and a heuristic algorithm based on particle swarm optimization is proposed according to the characteristics of the problem.This algorithm combines the idea of divide and conquer,divides the population into multiple groups according to individual similarity,and divides different roles to give consideration to local search and global search using known information.It is suitable for high-dimensional scheduling planning problems with exponential quantity variables such as this problem.Finally,considering the online scheduling requirements in dynamic networks,an incremental scheduling algorithm using the multi-period characteristics of the proposed scheduling mechanism is proposed for periodic flows.The priority goal is to shorten the delay caused by the new configuration distribution and avoid frame loss.Then all the above mechanisms and algorithms are implemented in the built simulation platform and further simulation verification is completed.