WRR算法在多类别实时数据流调度中的优化

摘要

With the development of converged networks, the quality of service (including the available bandwidth, end to end delay, jitter and packet loss rate) for some real-time data streaming applications (voice flow, video flow, etc. ) become more and more important. The traditional WRR algorithm can only meet the fairness of the real-time queue, but it can not assure that the multi-class real-time data is low latency and low jitter. The BSTLRR algorithm is based on the WRR scheduling algorithm. The BSTLRR scheduling algorithm not only in the low delay and low jitter multi-class real-time data stream frame is superior to the WRR scheduling algorithm, and to some extent, the priority queue ensures the fairness of scheduling.%随着融合型网络的发展,服务质量(包括可用带宽、端到端的时延、抖动和丢包率)对一些实时数据流应用(语音流、视频流等)越来越重要.由于传统的WRR算法只能满足各个应用队列的公平性要求,而不能保证多类别实时数据的低时延和低抖动性要求,所以本文在WRR算法的基础上提出了BSTLRR调度算法,此算法运用了二层循环和借用时隙的两个思想.BSTLRR调度算法不仅在调度低时延和低抖动的多类别实时数据流帧方面要优于WRR调度算法,而且在一定程度上也保证了各优先级队列调度上的公平性.