一种基于紧急程度的自时钟开始时间公平排队分组调度算法

为了克服目前GPS (Generalized Processor Sharing)类调度算法中实时应用分组的排队时延较大且不稳定的局限性，该文提出一种新的分组排队调度算法，该调度算法在计算分组服务标签时添加了一个紧急程度函数，调整了到达分组间的竞争关系，从而可以按照实时性应用的要求来调整到达分组的转发先优级，由此显著降低了实时性应用分组的排队时延和抖动幅度。分析和仿真实验表明，与GPS类其它调度算法相比，该调度算法对于实时应用的分组能提供较低的、更稳定的排队时延保证，同时还继承了GPS类算法的公平性和排队时延有界等特性，而且对系统虚拟时间的跟踪计算更为简捷高效。

An Urgency Based Self-clock Start-Time Fair Queuing Packet Scheduling

In order to get over the limitation of the GPS(Generalized Processor Sharing) like queuing algorithms which the bound of the queuing delay is so long and unstable that is undesirable for the packets of real-time applications, a new packet queuing and scheduling algorithm is proposed. In this algorithm the competitions of the arrival packets are fine-tuned by means of adding a function of flow urgency degree on the computation of their service tags, so the transmitting priorities of the arrival packets can be tuned according to their real-time applications’ needs, therefore the queuing delay and the bounds of oscillation are cut down significantly. According as the analysis and the simulation, this packet queuing and scheduling algorithm can resolve the long and unstable packet latencies problem for real-time applications, and shares both the bounded-delay and fairness properties of the GPS like algorithms, and the computation of its system virtual time is simple and efficient.