/obfs-detection

Scripts for CCS 2015 "Seeing through Network Protocol Obfuscation"

Primary LanguagePython

The scripts for CCS2015 paper "Seeing through Network Protocol Obfuscation"

Dependencies

Install Tor Stem and Selenium driver

sudo pip install stem selenium

Install scikit-learn

sudo apt-get install build-essential python-dev python-setuptools \
                     python-numpy python-scipy \
                     libatlas-dev libatlas3gf-base
sudo pip install -U numpy scipy scikit-learn

Use the framework to collect Tor traces

  1. Download the source code of the newest Tor Browser Bundle from https://www.torproject.org/projects/torbrowser.html.en, and unzip it. The resulting directory should be "path/tor-browser_en-US".
  2. Download the Alexa Top 1M domain list from http://s3.amazonaws.com/alexa-static/top-1m.csv.zip, or create your own file that contains target domains. The format of the file must be "unique_ID, domain_name". The unique IDs should be numeric values.
  3. Follow the instructions in https://github.com/Yawning/obfs4 to build obfsproxy4, change the output to obfs4proxy.bin (or obfs4proxy4.exe) and put it in the "path/tor-browser_en-US/Browser/TorBrowser/Tor/PluggableTransports/"
  4. Put tor_trace_collection.py and conf.py in the "path/tor-browser_en-US/Browser/", and configure the conf.py.
  5. Disable the TorLauncher extension in the TBB.
  6. Run "python tor_trace_collection.py -h" to see how to use it.
  7. The pcaps for a given type of PT will be stored at "PCAP_ROOT_DIRECTORY/ROUND_NUMBER/PT_NAME/"