The Internet is collectively composed of tens of thousands of individual networks, operated independently. Data flows from one network to another over paths that reflect agreements between network providers to exchange traffic. The ability of the Internet to connect computers around the world depends on the establishment and maintenance of these paths using a complex set of protocols. Hence, understanding the set of paths used in the Internet at any given time is important for the Internet's continued smooth operation. Unfortunately there are relatively few good metrics and algorithms for gaining insight into how the Internet's paths are currently configured and how they are changing over time.
This project is developing new methods to address the lack of good path analysis tools. Using these novel metrics and algorithms, this project's goal is to give insight into a wide range of questions, including identifying when data traffic may be being hijacked for malicious purposes, and when network operators shift their business strategies. The project is developing metrics based on Routing State Distance (RSD), and examining both micro-level questions (at the level of individual Autonomous Systems (ASes)) and macro-level questions (at the level of the whole Internet). At the micro level, the project is examining methods for identifying co-managed ASes, unusually-routed ASes, and shifts in the routing towards individual ASes. At the macro level, the project is assessing the "flattening" of the Internet and large-scale trends in the factors that drive unusual routing behavior. A key component of the project is the identification of where and how to do active measurement to most effectively obtain information about the global routing system.
Broader Impact. The methods being develop have potential to guide improvements to Internet operation, which benefits society broadly. Results from this project will be presented to network operators to help guide future understanding of Internet routing. Additional broader impacts of this project are being felt through educational development and continued support for under-represented groups. As part of this project, the PI is developing learning modules and exercises focusing on network analysis for inclusion in his course "Computer Networks" as well as contributing modules for inclusion in the course "Tools and Techniques for Data Mining" which he is helping to develop. The PI is committed to the engagement of under-represented groups and women in his research, and currently supervises two female Ph.D. students.
