Research On Hybrid Traffic Scheduling Mechanism In Time-Sensitive Networks

In recent years,with the arrival of the Industry 4.0 era and the continuous popularization of intelligent terminal device applications,there have been more and more highly intelligent network application scenarios,such as autonomous driving,telemedicine,virtual reality,etc.The types of information transmitted over the networks has increased rapidly,and the frequency of information interaction between devices has also risen sharply,putting forward higher requirements for the certainty and real-time of information transmission.Traditional Ethernet provides best-effort services for data transmission,and lacks unified scheduling and planning for end-to-end transmission behavior of data packets.Although Ethernet can improve the quality of service(QoS)through congestion avoidance,timeout retransmission and other mechanisms.But once a packet is sent,network actually loses control of the packet,and the resulting delay uncertainty is unacceptable for services with high QoS requirements.To tackle these issues,the IEEE Time-Sensitive Network Working Group proposed a Time-Sensitive Network(TSN)with deterministic delay guarantee function based on standard Ethernet.TSN adds clock synchronization,traffic scheduling,frame preemption and other technologies on the basis of standard Ethernet to realize real-time communication with deterministic limits for the network delay and jitter.This thesis mainly focuses on TSN traffic scheduling technology,and the main contributions are as follows:Firstly,in the LAN scenario where the entire network equipments are precisely synchronized in time and the cycles are completely aligned,a hybrid traffic scheduling mechanism combining the Time-Aware Shaper(TAS)algorithm with the Cyclic Queuing and Forwarding(CQF)algorithm is proposed.The mechanism fully considers the impact of flow scheduling sequence,slot allocation scheme and other factors on network load balancing and scheduling success rate,and improves the schedulability of the network by improving resource utilization.The mechanism mainly consists of three parts:determining the minimum scheduling time slot of the network and adjusting the sampling period of the scheduled traffic(ST),a parity mapping scheme for classifying the stream reservation(SR)traffic,and a traffic scheduling algorithm for adjusting sending time slot of flow at source node.The experimental results show that the mechanism significantly improves the system bandwidth utilization rate and the network scheduling success rate,achieves a good compromise between reducing the complexity of the solution and improving the scheduling performance,and realizes the efficient traffic scheduling of TSN in the same time domain.Secondly,for the WAN scenario with intra-domain time synchronization and inter-domain frequency synchronization,a hybrid traffic scheduling mechanism based on Time-Aware Shaper combined with Cycle Specific Queuing and Forwarding(CSQF)is further proposed.The mechanism first follows the scheme of determining the minimum scheduling time slot of the network and adjusting the sampling period of ST traffic to determine the time scheduling unit and reduce the bandwidth occupation of ST traffic,which is the same as that in our first research outcome.Then,based on the minimum scheduling time slot,under the premise of only ensuring the frequency synchronization between domains,the scheme of realizing periodic cyclic mapping alignment between two adjacent nodes located in different time domains is derived,which lays the foundation for designing the traffic scheduling scheme of crossdomain networks.Finally,combined with segment routing technology,a heuristic algorithm for joint routing and scheduling based on CSQF mechanism is proposed to complete the resource allocation of SR traffic.The experimental results show that compared with the existing mechanisms,the scheme proposed in this thesis has greatly improved the system resource utilization and scheduling success rate,and has good performance even in high load scenarios,and achieves efficient TSN traffic scheduling under cross-domain conditions.