The Need for Realism when Simulating Network Congestion

Published

April 2, 2016

Author(s)

Kevin L. Mills, Christopher E. Dabrowski

Abstract

Many researchers use simulation to investigate network congestion, often finding congestion spread can be modeled as percolation, spreading slowly under increasing load until a critical point, then spreading quickly through the network. The researchers identify precursor signals that arise around the critical point. The signals predict onset of widespread congestion. These findings suggest a theoretical basis for monitoring methods to warn of impending congestion collapse. Yet questions surround the research, as the models are abstract, bearing little resemblance to deployed networks. We explore these questions by examining the influence of realism on congestion spread in simulations. We begin with an abstract model, from the literature, and add elements of realism in various combinations, culminating with a high-fidelity simulation, also from the literature. By comparing congestion patterns among combinations, we make three main contributions. First, we illustrate that congestion spread in abstract models differs significantly from spread in realistic models. Second, we identify elements of realism needed when simulating congestion. Finally, we demonstrate a method to compare congestion patterns among simulations covering diverse configurations. We hope our contributions lead to better understanding of the need for realism when simulating network congestion.

Proceedings Title

Proceedings of the 19th Communications & Networking Symposium (CNS 2016)

Conference Dates

April 3-6, 2016

Conference Location

Pasdena, CA, US

Conference Title

Communications and Networking Symposium (CNS)

Pub Type

Conferences

Download Paper

Local Download

Keywords

Congestion, criticality, networks, percolation, simulation

Citation

Mills, K. and Dabrowski, C. (2016), The Need for Realism when Simulating Network Congestion, Proceedings of the 19th Communications & Networking Symposium (CNS 2016) , Pasdena, CA, US, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=919676 (Accessed April 23, 2024)