Toward A Manifold View of the Internet

Date and time: 
Mon, Nov 7 2022 - 10:00am
Location: 
360 Deschutes
Speaker(s):
Loqman Salamatian
Columbia University
Host/Committee: 
  • Reza Rejaie

Measuring the Internet is of primary interest to assess its operation, improve its connectivity, and bring awareness to possible issues. Historically, the principal tool used by researchers to understand Internet infrastructure has been traceroute. Developed in the late 1980s as a troubleshooting tool for network operators, the Internet measurement community has enthusiastically adopted and improved it to study the properties of the routes taken by packets and to infer connectivity in the underlying physical infrastructure. But the past few years have seen a decrease in the utility of traceroutes to measure anything on the Internet, and there is a real possibility that in the (near?) future, the Internet will remain entirely opaque to traceroutes. In this talk, I build from two recent works published at ACM IMC 2022 and ACM Sigmetrics 2022 on a plan to build a manifold vision of the Internet that does not rely on anything that can be tampered with by middleboxes in the network. I describe a new method for assessing the characteristics of network path connectivity that is based on concepts from Riemannian geometry (i.e., Ricci curvature) that also relies on an array of carefully crafted visualizations (e.g., a novel manifold view of a network’s delay space) in private backbones managed by hyperscalers (Google, AWS, Microsoft). I discuss how this manifold representation opens a new way to bridge the gap between the Internet as a physical and logical space. In particular, I illustrate how starting from a geographic model of the Internet, the manifold can result in an entirely new way of thinking about connectivity.

Biography: 

Loqman Salamatian is a third-year Ph.D. candidate at Columbia University supervised by Pr. Ethan Katz-Bassett, Pr. Vishal Misra and Pr. Daniel Rubenstein. His main interests are Internet measurement, complex systems analysis, and Riemannian geometry. He is working on finding new models to understand the interactions between virtual and physical space. He likes to leverage theory to validate qualitative models on top of carefully crafted measurements.

Tags: