Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: the data best?) • Many of these 7 models considered produce similar macroscopic features (degree distribution, clustering, diameter, etc). • Delve into microscopic details and let the data distinguish between the 7 models. • Must start with models that are accurate statistical fits to data! (different type of model validation). (Acompanying commentary, Rice et al PNAS 2005, DMC does not reproduce giant component.) • Vladimir Filkov lecture next time 2/7/11 “Understanding Internet Topology: Principles, Models, and Validation” D. Alderson, L. Li, W. Willinger, and J. C. Doyle, IEEE/ACM Trans. on Networking, 13 (6), 2005. • I chose this paper since they explicitly claim to deal with validation. Overview • “First principles” approach to router-level Internet modeling. • Consider capability of real routers (annotated graph). • Consider core versus edge requirements. • From this design “optimal” networks. • Compare with sampled topologies of actual internet, and show that constraint-capacity performance curve of real routers fit with their hypothesis. Motivation – Need accurate models of the internet • Testing and evaluating protocols • Protecting against and detecting attacks • Improved designs and resource provisioning → need annotated graphs, with bandwidth capacity explicit (also router buffer capacity). • Given topology generated from a model, which statistical properties to test? • Ascribing meaning to model details. (Why would a random construction relate to an engineered network?) Router level connectivity • Layer 2 (data-link layer) connectivity Open Systems Interconnection (OSI) Reference Model 7 Application Layer 6 Presentation Layer 5 Session Layer 4 Transport Layer 3 Network Layer 2 Data Link Layer 1 Physical Layer Past work – structural topology generators • Random connectivity (e.g., Waxman model) • Transit-stub models of Zegura (Georgia Tech Internetwork Topology Models) • But this miss the broad-scale (power-law-like) distributions in connectivity presumed to b...
View Full Document

This document was uploaded on 03/12/2014 for the course CSCI 289 at UC Davis.

Ask a homework question - tutors are online