https://github.com/tbabej/uadt

User action detection toolkit. Spy on mobile phone apps using machine learning-based attacks on the encrypted traffic.
https://github.com/tbabej/uadt

Last synced: 12 months ago
JSON representation

User action detection toolkit. Spy on mobile phone apps using machine learning-based attacks on the encrypted traffic.

• Host: GitHub
• URL: https://github.com/tbabej/uadt
• Owner: tbabej
• Created: 2017-05-21T20:29:47.000Z (about 9 years ago)
• Default Branch: master
• Last Pushed: 2019-09-04T15:16:17.000Z (almost 7 years ago)
• Last Synced: 2025-07-21T16:46:02.449Z (about 1 year ago)
• Language: Python
• Homepage:
• Size: 3.7 MB
• Stars: 11
• Watchers: 5
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
Usage

-----

First, make sure you have all the dependencies:

    $ pip install -r requirements.txt

Additionally, you will need to provide training data in the form of pcap files:

    $ ls ~/mypcaps/*.pcap

    one.pcap

    two.pcap

    ...

To generate the dataset of features extracted from the pcap files, use dataset.py:

    $ ./dataset.py ~/mypcaps/

    [1/3] Processing: /home/tbabej/mypcaps/one.pcap

    [2/3] Processing: /home/tbabej/mypcaps/two.pcap

    [3/3] Processing: /home/tbabej/mypcaps/three.pcap

    Writing to mypcaps.csv

To train and evaluate the SVM on this data, use svm.py:

    $ ./svm.py mypcaps.csv --optimize

    Searching for optimal parameters..

    Used parameters: C=2048.0, gamma=0.03125

    Success rate: 0.744769874477

The dataset and svm commands have more options, explore their documentation via:

    $ ./dataset.py -h

    $ ./svm.py -h

Useful

------

Pre-generated .csv training data sets are available in trainingsets/ directory.