We are the Network Architecture Lab, and we work on fundamental questions regarding network system design:
- Can we infer network behavior from external observations? We proved the feasibility of network neutrality inference [ paper ]. Earlier, we relaxed the assumption of link independence in network performance inference [ paper ].
- Can we improve Internet transparency? We showed how Internet service providers can produce verifiable traffic receipts—report their own performance in a way that prevents them from lying [ paper ].
- Is Internet transparency compatible with user privacy? We started exploring the trade-off between Internet transparency and Tor anonymity [ paper ]. Georgia won an IRTF Applied Networking Research Prize for this work!
- How should we write software network functions (NFs), such that we can reason about their semantics? We presented the first formally verified NAT written in C [ project page, paper ], then the first framework for verifying software NFs with no verification expertise [ project page, paper ].
- How should we write software NFs, such that we can reason about their performance? We proposed the idea of performance interfaces and showed how to extract such interfaces from software NFs [ code, paper ]. Performance interfaces evolved from performance contracts [ project page, paper ].