Criticisms of modelling packet traffic using long-range dependence

TL;DR

This paper critically examines the role of long-range dependence in Internet traffic modeling, revealing theoretical flaws in common models and demonstrating that long-range correlations are generally not crucial for queuing performance except under very high load.

Contribution

It challenges the significance of long-range dependence in traffic modeling and provides empirical evidence questioning its impact on queuing behavior.

Findings

Common models have theoretical flaws and do not match real traffic traces.

Long-range correlations are generally not important for queuing performance.

High load conditions may be an exception where long-range dependence matters.

Abstract

This paper criticises the notion that long-range dependence is an important contributor to the queuing behaviour of real Internet traffic. The idea is questioned in two different ways. Firstly, a class of models used to simulate Internet traffic is shown to have important theoretical flaws. It is shown that this behaviour is inconsistent with the behaviour of real traffic traces. Secondly, the notion that long-range correlations significantly affects the queuing performance of traffic is investigated by destroying those correlations in real traffic traces (by reordering). It is shown that the longer ranges of correlations are not important except in one case with an extremely high load.