Adaptive Traffic Fingerprinting for Darknet Threat Intelligence
This addresses the challenge of cyber threat intelligence for organizations needing to monitor darknet activities, though it is a proof-of-concept with incremental improvements.
The paper tackles the problem of identifying malicious traffic on darknets like Tor by developing a novel algorithm that reduces anonymity and detects targeted connections, achieving a false positive rate of 0.001 in simulations.
Darknet technology such as Tor has been used by various threat actors for organising illegal activities and data exfiltration. As such, there is a case for organisations to block such traffic, or to try and identify when it is used and for what purposes. However, anonymity in cyberspace has always been a domain of conflicting interests. While it gives enough power to nefarious actors to masquerade their illegal activities, it is also the cornerstone to facilitate freedom of speech and privacy. We present a proof of concept for a novel algorithm that could form the fundamental pillar of a darknet-capable Cyber Threat Intelligence platform. The solution can reduce anonymity of users of Tor, and considers the existing visibility of network traffic before optionally initiating targeted or widespread BGP interception. In combination with server HTTP response manipulation, the algorithm attempts to reduce the candidate data set to eliminate client-side traffic that is most unlikely to be responsible for server-side connections of interest. Our test results show that MITM manipulated server responses lead to expected changes received by the Tor client. Using simulation data generated by shadow, we show that the detection scheme is effective with false positive rate of 0.001, while sensitivity detecting non-targets was 0.016+-0.127. Our algorithm could assist collaborating organisations willing to share their threat intelligence or cooperate during investigations.