Inferring Invisible Internet Traffic
The Internet is at once an immense engineering artifact, a pervasive social force, and a fascinating object of study. Unfortunately many natural questions about the Internet cannot be answered by direct measurement, requiring us to turn instead to the tools of statistical inference. As a detailed example I’ll describe a current project in traffic measurement. We are asking the question: using traffic measurements taken at one location in the Internet, can we estimate how much traffic is flowing in a different part of the Internet? Surprisingly, the answer is yes. I’ll explain why this is possible (with a connection to problems like the Netflix Prize), how it can be done, and how this result could be used to give a network operator an edge over its competitors.
Mark Crovella is Professor and Chair of the Department of Computer Science at Boston University. His research interests center on improving the understanding, design, and performance of parallel and networked computer systems, mainly through the application of data mining, statistics, and performance evaluation.
Professor Crovella is co-author of “Internet Measurement: Infrastructure, Traffic, and Applications” (Wiley Press, 2006) and is the author of over one hundred and fifty papers, with over 19,000 citations (Google
Scholar). Between 2007 and 2009 he was Chair of ACM SIGCOMM. In 2010 his paper “Self-Similarity in World Wide Web Traffic: Evidence and Possible Causes” received the SIGMETRICS Test of Time Award, and in 2013 his paper “Routing State Distance: A Path- Based Metric for Network Analysis” won the IETF Applied Networking Research Prize. Professor Crovella is a Fellow of the ACM and of the IEEE.