基于双时间尺度QoS分层的拥塞控制算法

该文针对网络流量的自相似特性提出了一种基于双时间尺度QoS分层的拥塞控制算法，该算法以丢包率作为标准对网络服务质量（Qos）进行分层，质量越高的层次对应越低的丢包率，在小时间尺度上，根据网络传输时延及丢包率等信息，由接收端完成速率的计算工作，并反馈给发送端，发送端根据此速率选择一定层次质量的数据发送；在大时间尺度上，通过预测未来流量水平的变化趋势，控制相邻服务质量层次之间变换的难易程度，以减少发送速率在相邻质量层次间来回振荡。实验表明，与TFRC算法相比，该算法能够有效降低业务流的丢失率，为系统提供更好且更稳定的服务质量。

A New Congestion Control Mechanism Based on Two-time-scale Algorithm with QoS Scaling

Since the traffic in modern telecommunication networks always shows self-similarity or long-range dependence,we propose in this paper a new congestion control mechanism based on two-time-scale rate control algorithm.In the smaller time-scale with QoS(Quality of Service) scaling,the receiver calculates the sending rate according to the RTT and packets lossing rate,then feeds it back to the sender which adjusts the rate immediately and ensures the quality.In the larger time-scale,the algorithm adjusts the distance between the adjacent levels based on the long-range dependency of the traffic.Our simulated result shows that in the self-similar network environment the algorithm can enhance the tracking ability of the network and gain improvement in loss rate when compared with TFRC.It also has a better performance in QoS and smoothness.