A conservation-law-based modular fluid-flow model for network congestion modeling

TL;DR

This paper introduces a modular fluid-flow model for network congestion based on an information conservation law, enabling accurate transient analysis and flexible topology description, with validation against packet-level simulations.

Contribution

It presents a novel conservation-law-based modular modeling framework that captures transient network behavior and includes mechanisms previously ignored, improving accuracy and generality.

Findings

Model accurately predicts transient network behavior

Framework is adaptable to any network topology

Validation shows close match with packet-level simulations

Abstract

A modular fluid-flow model for network congestion analysis and control is proposed. The model is derived from an information conservation law stating that the information is either in transit, lost or received. Mathematical models of network elements such as queues, users, and transmission channels, and network description variables, including sending/acknowledgement rates and delays, are inferred from this law and obtained by applying this principle locally. The modularity of the devised model makes it sufficiently generic to describe any network topology, and appealing for building simulators. Previous models in the literature are often not capable of capturing the transient behavior of the network precisely, making the resulting analysis inaccurate in practice. Those models can be recovered from exact reduction or approximation of this new model. An important aspect of this particular modeling approach is the introduction of new tight building blocks that implement mechanisms ignored by the existing ones, notably at the queue and user levels. Comparisons with packet-level simulations corroborate the proposed model.