On Exploiting Long Range Dependency of Network Traffic in Measuring Cross Traffic on an End-to-end Basis

In this paper we present three theoretically grounded methods: prediction, reconstruction and interpolation, for measuring cross traffic on the bottleneck link of an end-toend path. The objective is to infer cross traffic as accurately as possible, while not injecting a significant amount of probe packets into the network. In the prediction-based method, we take advantage of the LRD characteristic of the cross traffic to predict the future traffic based on the recent information obtained by probe packets. In the reconstruction method, we rebuild the entire cross traffic process with the information obtained by probe packets. In the interpolation method, we periodically send closely-spaced probe packet pairs to sample cross traffic of the bottleneck link, and infer cross traffic between two sampling points using interpolation. The simulation study indicates that (i) the prediction-based and reconstruction methods can give good mean measurement of cross traffic, while the interpolation method usually captures the instantaneous value of cross traffic better; and (ii) all three methods are adaptive to the dynamic change of cross traffic and are quite robust in the presence of multiple bottleneck links on an end-to-end path.