DeTor: Provably Avoiding Geographic Regions in Tor
summary
DeTor builds client-side selection algorithms and a clock-measured speed-of-light alibi that prove a Tor circuit did not traverse user-specified countries, without changing Tor or needing Internet topology. Using pairwise Tor latencies measured with the team tool and simulation, DeTor finds at least one never once circuit for most source–destination pairs (many pairs find hundreds) and finds thousands of never twice circuits for over 80% of pairs; selected circuits also show lower measured RTTs. Code and data are released.
AI-written summary · about
chapters (16)
- 0:00DeTor introduction and goals
- 1:57Nation-state adversary threat model
- 3:23Two geographic avoidance types
- 5:10Design for deployability and proof
- 6:33Proving avoidance with round-trip time
- 8:19Never-once avoidance proof
- 10:27Never-twice against correlation
- 14:25Ellipses method for verification
- 16:30Evaluation setup and methodology
- 17:35Never-once success rate
- 19:46Detour circuits per pair
- 21:09Performance impact of detours
- 23:19Summary of effectiveness
- 24:01Q&A: Real-world deployment
- 26:33Q&A: Threat model and relays
- 28:23Q&A: Bias and scalability
related talks
23mTowards Predicting Efficient and Anonymous Tor Circuits
20mDemystifying the Darknet in 30mns or Less
19mValidaTor: Domain Validation over Tor
01h 16mAnonymity loves Diversity: The Case of Tor
54mAnonymity Loves Resilience: The Case of Tor
33mOnion adventures: how to use onion services and the Tor network in your web endeavors
15mDonar: Anonymous VoIP over Tor
20mSocksTrace: A Proxy Leak Detector for Anonymity-Focused Network Applications