基于循环排队转发的时间触发流量路由与调度优化方法

Time-triggered traffic routing and scheduling optimization method based on cyclic queuing and forwarding

基于传统以太网的时间敏感网络(Time-SensitiveNetwork,TSN)为时间敏感应用提供了确定性、可靠性的通信保障。为了实现流量调度,TSN工作组提出了各种整形机制,其中基于乒乓队列的循环排队转发机制简化了门控配置,实现流量的有界延迟传输。现有基于CQF机制的流量调度忽略了路由策略对调度优化的影响,针对最短路径优先造成的队列拥塞情况,本文提出了一种基于混合粒子群算法的路径选择算法,算法考虑最小化路径长度和链路负载均衡来确定最优路径,同时提出了一种基于快速排序算法的排序方法提高网络的调度能力。通过实验仿真,算法可以在大规模网络中求解上千条时间敏感流量的调度问题,在可行的执行时间范围内,流量调度成功率可调高约20%。

The Time-Sensitive Network(TSN)based on traditional Ethernet provides deterministic and reliable communication guarantees for time-sensitive applications.The TSN working group has proposed various shaping mechanisms to achieve traffic scheduling,among which the cyclic queuing forwarding mechanism based on a ping-pong queue simplifies gate configuration and enables bounded delay transmission of traffic.Existing traffic scheduling based on CQF mechanisms ignore the impact of routing policies on scheduling optimization.To address the queue congestion caused by the Shortest Path First(SPF)algorithm,this paper proposes a path selection algorithm based on a hybrid particle swarm algorithm.The algorithm considers minimizing path length and link load balancing to determine the optimal path while proposing a sorting method based on the QuickSort algorithm to improve the network's scheduling capability.Experimental simulations demonstrate that the algorithm can solve the scheduling problem of thousands of time-sensitive traffic flows in large-scale networks,and the successful rate of traffic scheduling can be increased by about 20%within the feasible execution time range.