骨干网络中的被动模式RTT测量方法*

面向网络管理需求，将不同网络终端各自到测量点的半路径往返时延(HPRTT)定义为测量对象，提出一种较为全面的被动模式测量方法，能够在骨干网络中对各个终端的HPRTT指标进行持续测量，并通过一种自适应模型对测量采样的有效性进行判定。与现有的被动模式RTT测量技术相比，该方法的测量对象和测量结果更加具有网络管理意义，能够在不同的TCP行为模式中获得更多的测量采样并显著提高测量成功率。在Gbps级流量环境下，对大量TCP终端的测量实验结果表明，该方法具有较高的测量效率和准确性。

Passive estimation of round-trip time in backbone network

To meet the network management requirements, this paper defined half-path round-trip time(HPRTT) as the RTT metric from a host to passive measurement point, and proposed a general passive method for continually measuring of HPRTT for different hosts in backbone network, also introduced a self-adaptive filter model to judge the validity of measurement samples. Comparing with current passive RTT estimation methods, this method could obtain more measurement samples from different TCP activity models thus observably improve the measurement success rate. Practical experiments are taken in Gbps backbone and HPRTTs are estimated for large numbers of TCP hosts, the result shows this method works properly and effectively.