The anonymity network Tor is under attack. It provides access to an uncensored internet for people in oppressive regimes, and enables for example whistleblowers to report wrongdoing with no fear of retribution. If its anonymity were to be reduced, this would hamper its purpose, both in actual terms, and because this would dissuade users, as the number of users increase its anonymity.
Website fingerprinting can allow a local passive adversary, a very limited attacker, to fathom which websites a user is visiting, reducing Tor’s anonymity. This has continued to improve as AI continues to improve, and researchers publish new website fingerprinting attacks against Tor.
This repository presents a new defense against website fingerprinting. The defense creates camouflage traffic tailored to WWW distributions to hide which page the user visits. It is easily configurable for the level of required anonymity, and does not burden the user overly, neither in installation, nor in the time and size delay it imposes.
If you want all the details and history, the complete thesis repo is at https://github.com/kreikenbaum/wfcomplete. The data files are at https://drive.google.com/open?id=13qM-OuX4RZMhsFbjKKW5ZfoXJIYGSkrI (8GB zipped, more than 20GB unzipped). After downloading, do a zpaq extract data.zpaq. Zpaq keeps the file size down so that it fits on the free Google Drive quota (thank you @Google for hosting this :-)).
If you are only interested in capturing website fingerprinting data, have a look at https://github.com/kreikenbaum/capture. The INSTALL.org file has all the details.
Any feedback, questions, corrections, etc, feel welcome! You can use the issue system.