Internet Traffic Characterization Based on Active Network Measurement

Existing traffic characterization systems today use passive monitoring techniques. Compared with active measurements, passive measurements have their own disadvantages: they are not easy to implement and need the cooperation of intermediate nodes along the path. Hence they require hardware change, which can significantly increase the cost of measurements. This makes active measurements an attractive option for small-scale performance monitoring. In this paper, we present a new technique to analyze traffic characterization with packet-train probing. We first expanded the traditional packet-pair approach and constructed a simple model to understand how competing traffic changes the probing packet gap in a train for a single-hop network. , We, therefore, present a new theoretical model of traffic characterization based on packet-train probing. This new technique is based on the clear knowledge that traffic characteristics on small time-scale, for packet-dispersion techniques are basically a method of analysis on a small timescale. We used extensive Internet measurements to show that, in addition to providing robust estimators of the Hurst parameter, it can also do multi-fractal analysis. We end this paper by showing, both analytically and experimentally, that the packet-train probing based proposals work correctly on the Internet and offer reliable measurement results.