OpenVPN fingerprinting

Group 4 - Leonard Eyer, Cristina Stoleriu, Sofia Lohr, Tobias Loch, Matthijs Reyers

This repository contains code for our report on the paper: "OpenVPN is Open to VPN Fingerprinting" by Xue et al. (2022) [1].

Requirements

Install:
- python 3.9
- scapy 2.5.0
- docker 24.0
Download the following datasets:
- VNAT
- ISCXVPN2016

Project structure

The project is structured as follows:

docker: Docker environments to generate datasets with OpenVPN configurations (udp, tcp, udp-tls, tcp-tls)
docker-pcaps: Captured pcap files from the various Docker environments
src: source code for the passive fingerprinting methods

Running Experiments

Extract the contents of the datasets into some folder
Adjust the paths in the config.
Run the experiments by:
```
python experiments.py
```

The config contains the information on how to execute the experiments. Its structure is as follows:

The datasets dictionary contains the path to all pcap files. The actual used datasets are specified in used_datasets
The experiments array contains the algorithms that are run and with what parameters
The output_folder field contains then the folder where for each dataset its results are output

Docker environments

To generate some datasets with known OpenVPN configurations, we created several docker compose environments contained in the docker folder. The captured Pcap files are stored in the docker-pcaps folder.

Every OpenVPN client container runs the docker/client.sh script to generate some network traffic.

References