This thesis analyzes the threat of autonomous system (AS) and Internet exchange(IX) level adversaries on Tor, currently the most widely deployed andused anonymity overlay network. Of particular interest is the possibility ofa single AS or IX point observing both the path from the client to the entrynode and the path from the exit node to thenal destination. Experimentalresults indicate that a non-trivial number of circuits are vulnerable tosuch compromise. A novel AS-level path prediction algorithm is developedin order to allow the client to choose paths without vulnerabilities. The pathprediction algorithm sacri ces some accuracy in the top path prediction inorder to decrease the hardware requirements necessary to predict AS-levelpaths and is simple enough to run on standard client hardware. We validatethe accuracy of the path predictorrst compared to classical path predictionalgorithms, then compared to traceroute data taken from Planet Lab. Thesimulator predicts paths with similar sets of ASes and linksnding 90% of theactual ASes seen in the traceroute data. The e ects of choosing paths utilizingthe new path predictions is then investigated tond that load balancingis adversely a ected. The entropy loss due to the new path selection methodis also investigated, speci cally the entropy of the client from an adversaryobserving the exit/destination path. Wend that choosing paths with ournew path selection algorithm results in minimal entropy loss. Overall, theresults demonstrate that the new path simulator is a lightweight solution todefend against AS and IX-level compromise of anonymous communicationpaths on the Internet and should be considered for deployment to maintainthe privacy guarantees of such systems.
【 预 览 】
附件列表
Files
Size
Format
View
Protecting anonymity in the presence of autonomous system and internet exchange level adversaries
PDF
download
878987797
028-85220240
