基于队列长度和时延约束的异步OPS权重调度

Weighted Scheduling Based-on Queue Length and Queue Delay Restriction for Asynchronous Optical Packet Switch

分组调度在异步光分组交换(OPS)节点中起着重要作用,一种智能的调度算法能减小分组的丢失率,增加节点的吞吐量,从而提高服务质量。文章提出了一种基于队列长度和时延约束的解决队列中分组"饿死"和时延问题的权重调度算法。算法通过计算队列长度和分组在队列中等待时间确定调度的权重值,以减小分组在核心节点处的资源竞争和等待时延,调度器计算每个输入端口的N×K个虚拟输出队列的权重值并调度最大权重值队列到可用的输出信道上。分析和仿真结果表明所提的权重调度算法具有高吞吐量和低分组丢失率特点。

Packet scheduling plays an important role in the asynchronous optical switched node. An intelligence scheduling algorithm can reduce packet loss probability and increase the throughput to guarantee the quality of service. A weighted scheduling algorithm based on the queue length associated with waiting delay restriction was proposed to solve the problem of packet queue starvation and time delay. The work has contributed with a much powerful scheduling mechanism which reduces resource contention at the asynchronous OPS core nodes and guarantees a reasonable packet delay thanks to its weighted scheduling algorithm based on queue length and waiting delay for variable length OPS network. The scheduler calculates every input port with N×K virtual output queue（VOQ） weight and schedules the maximal weight queue packets to available output channel. The analysis and simulation results show that the proposed scheduling has high throughput and low packet loss ratio.