Poster: PredicTor: Predicting Fast Circuits For A Faster User Experience in Tor

The Tor anonymity system provides online privacy for millions of users, but it is slower than typical web browsing. To improve performance in Tor, we propose PredicTor, a path selection technique that uses a kNN classifier trained on a set of 125,000 simulated Tor paths in order to predict the performance of a proposed path. If the path is predicted to be fast, then a circuit is built using those relays. We implemented PredicTor in the Tor source code and show through simulations in Shadow that PredicTor improves Tor network performance by 29% compared to Vanilla Tor and by 21% compared to the previous state-ofthe-art scheme.