Tracing Website Attackers by Analyzing Onion Routers' Log Files

Abstract

The Onion Router (Tor) is one of the major network systems that provide anonymous communication and censorship circumvention. Tor enables its users to surf the Internet, chat, and send messages anonymously; however, cyber attackers also exploit the system for circumventing criminal activity detection. Recently, various approaches that prevent or mitigate abuse of Tor have been proposed in the literature. This paper, which presents one of the approaches, addresses an IP traceback problem. In our model, onion routers that voluntarily participate in attacker tracing detect attack packets (packets carrying an attacker's code or data) recorded in the log files by sharing necessary information with an attacked server over an Ethereum blockchain network. The detection algorithm in this paper uses the statistics of packet travel and relay times and outputs attack-packet candidates. The proposed method attaches a reliability degree to each candidate, which is based on the upper bounds of its Type I and II error rates. A smart contract running on the blockchain network ranks the detection results from onion routers according to the reliability degrees.