Towards an Autonomous System Monitor for Mitigating Correlation Attacks in the Tor Network

Abstract

After carefully considering the scalability problem in Tor and exhaustively evaluating related works on AS-level adversaries, the author proposes ASmoniTor, which is an autonomous system monitor for mitigating correlation attacks in the Tor network. In contrast to prior works, which often released offline packets, including the source code of a modified Tor client and a snapshot of the Internet topology, ASmoniTor is an online system that assists end users with mitigating the threat of AS-level adversaries in a near real-time fashion. For Tor clients proposed in previous works, users need to compile the source code on their machine and continually update the snapshot of the Internet topology in order to obtain accurate AS-path inferences. On the contrary, ASmoniTor is an online platform that can be utilized easily by not only technical users, but also by users without a technical background, because they only need to access it via Tor and input two parameters to execute an AS-aware path selection algorithm. With ASmoniTor, the author makes three key technical contributions to the research against AS-level adversaries in the Tor network. First, ASmoniTor does not require the users to initiate complicated source code compilations. Second, it helps to reduce errors in AS-path inferences by letting users input a set of suspected ASes obtained directly from their own traceroute measurements. Third, the Internet topology database at the back-end of ASmoniTor is periodically updated to assure near real-time AS-path inferences between Tor exit nodes and the most likely visited websites. Finally, in addition to its convenience, ASmoniTor gives users full control over the information they want to input, thus preserving their privacy.