The purpose of this workshop is to bring together a diverse group of researchers and practitioners to discuss several aspects of the structure of the Internet and of network traffic on the Internet. In particular, one of the main themes will be a discussion of current tools to measure and infer the connectivity structure of the Internet, and to measure and model the statistics of dynamics of the traffic. Another main theme will be the modeling of both the emergence of network structures and traffic patterns, through progressive competitive/cooperative growth, and the a posteriori statistical analysis of measured traffic data, with explicit connections and studies on available data sets. Challenges and opportunities in constructing multiscale models for such complex networks and traffic patterns will be discussed, from both a mathematical perspective and in view of concrete application to the data sets above.
Inferring connectivity structure of the Internet, and analysis of traffic data.
The connectivity structure of the Internet is complex and mostly hidden. It is challenging to design measurements to infer such structure, and tools to carry them out, that will provide unbiased insight on the structure of the Internet. This has lead to a wide variety of models for the connectivity structure, but the accuracy of such models is often hard to validate. Practitioners from the industry (e.g. ISP’s) and academic researchers alike will present their views. Several tools for measuring and analyzing Internet traffic will also be discussed, as well as current models for interpreting the large data sets collected.
Multiscale representations of large graphs and dynamic patterns.
Generating multiscale representations of large graphs is a problem that arises across a wide variety of disciplines, among which homogenization of PDEs, modeling of physical and biological systems, numerical methods for solving large sparse linear systems or computing eigenvectors of large graphs (e.g. PageRank), exploration of complex graphs and high dimensional data sets by using random walks and spectral graph theory. In this workshop we will bring together researchers in these topics together with industry practitioners with the expectation of fruitful interactions, both in the direction of applying existing ideas to the available data sets of Internet measurements, and in the direction of stimulating the development of new ideas for analysis of these measurements.
Paul Barford
(University of Wisconsin-Madison)
Anna Gilbert
(University of Michigan)
Mauro Maggioni, Chair
(Duke University)
Morley Mao
(University of Michigan)
Rob Nowak
(University of Wisconsin-Madison)
Copyright. All Rights Reserved.