一种基于弱实时的加权公平队列调度算法

Weighted fair queuing scheduling algorithm based on weakly hard real-time constraint

针对目前的一些实时应用,例如视频传输应用、网络传输延迟和网络资源利用率相矛盾的问题,结合弱实时理论,提出了基于弱实时约束的广义处理器共享(WGPS)的调度策略以及面向分组模型的基于弱实时约束的加权公平队列(W2FQ)调度算法.该算法以加权公平队列(WFQ)调度算法为基本算法,根据丢失率参数和弱实时约束的要求,适当丢失一些分组,从而可以降低网络传输延迟和减小网络服务带宽.理论分析和仿真结果表明,该算法是一个公平可行的调度算法,而且与WFQ算法相比,实时应用能够得到更及时的传输且保证服务质量不受影响.

In order to weaken the incompatibility between transmission delay and network resource utilization of some real-time applications, such as video transmission application, by integrating weakly hard real-time constraint guarantee, the weak generalized processor sharing （WGPS） scheduling policy based on fluid model and weak weighted fair queuing （W^2FQ） scheduling algorithm based on packet model were proposed. W^2FQ scheduling algorithm improved the weighted fair queuing （WFQ） scheduling algorithm. W^2FQ scheduling algorithm dropped some optional packets according to the miss ratio and weakly hard real-time constraint of real-time applications, therefore, the transmission delay and the network service rate of real-time applications were reduced effectively. Theoretical analysis with network calculus and simulation results show that W^2FQ scheduling algorithm has good performances in fairness and feasibility, and that real-time applications can be transmitted better under W^2FQ than under WFQ.