Abstract: Traditional weighted fair queue algorithms have the main drawback: the calculation of the weight parameters according to each packet. The paper proposes a new weighted fair queueing algorithm(SPFQ), which uses service probability to schedule packets and a random number to decide which packet to be served next. In addition, a novel adaptive service probability parameter calculation method is used to solve the unfair problem induced by the variable packet length and an adaptive queue management technology to improve the utilization of the server’s queue buffer and reduce the delay burstiness. Simulation results demonstrate the validity and practicability of SPFQ.

Key words: queue scheduling, weighted fair queueing, adaptive queue management, packet switching network

摘要： 传统公平队列调度算法(WFQ、WRR等)普遍存在基于数据包的权重参数计算问题，由此产生的高复杂度使其难以获得广泛应用。该文提出一种新的加权公平队列调度算法，使用服务概率和随机数实现加权公平调度，显著降低了算法的复杂度。同时使用自适应服务概率计算解决了数据包变长度带来的不公平性。通过队列管理技术有效地提高了交换机的缓冲区利用率，并减小了排队延迟抖动。仿真结果证明了算法的有效性和实用性。

关键词: 队列调度, 加权公平排队, 自适应队列管理, 分组交换网络

CLC Number: 

• TP391