Beyond connectivity, modern users require that the networks and protocols they use ensure privacy. However, as network traffic is increasingly encrypted, service providers require new techniques to gain insight into the traffic they serve. In this talk I discuss privacy-enhancing networked systems, as well as systems and techniques for privacy-preserving network traffic analysis. First, I focus on measurements of modern name resolution protocols that have emerged that encrypt DNS traffic. Second, I discuss Oblivious DNS, a protocol that offers privacy protection while maintaining compatibility with existing DNS infrastructure. Third, I describe methods for network traffic analysis that can provide ISPs insight into network performance without breaking encryption. In this work we designed machine learning models that enable operators to infer application performance (streaming video quality of experience) without breaking encryption.
Paul Schmitt is an associate research scholar at the Center for Information Technology Policy (CITP) at Princeton University. His research focuses on networked systems, protocol design, privacy, network traffic inference and performance analysis, and scalable Internet measurement. His work takes a dirty-slate approach to networked systems research, allowing for compatibility and deployability in current environments. He previously received his PhD from UC Santa Barbara in 2017 with Elizabeth Belding and completed a postdoc at Princeton University with Nick Feamster.
ISI Talk Host: Terry Benzel, Director of the Networking and Cybersecurity Division