Modeling queuing dynamics of TCP: a simple model and its empirical validation

TL;DR

This paper introduces a simple, empirically validated model for TCP queuing dynamics, addressing the lack of accurate models for buffer content behavior under TCP traffic.

Contribution

It proposes a new, straightforward TCP queuing model grounded in experimental data and validated through extensive simulations, improving prediction accuracy.

Findings

Model accurately predicts TCP buffer dynamics

Validated by extensive packet-level simulations

Addresses gap in existing simplistic models

Abstract

Understanding queuing dynamics of TCP is important for correct router buffer sizing as well as for optimizing the performance of the TCP protocol itself. However, modeling of buffer content dynamics under TCP has received relatively little attention given its importance. Commonly used queuing models are based on overly simplistic assumptions about the packet arrival process. As a consequence, there are no quantitatively accurate closed loop TCP models capable of predicting performance even for a single link shared by multiple flows. Our present paper aims to close this gap by proposing a simple TCP queuing model, which is based on experimental observations and validated by extensive packet level simulations.