MAVIDS

This project was a PoC and is not actively being worked on. It now requires some TLC to become fully operational.

The Micro Air Vehicle Intrusion Detection System (MAVIDS) is a machine learning-based IDS for UAVs that use the MAVLink communication protocol.

Installation

MAVIDS is only supported on linux using Python 3.7.

Clone the repo:

git clone https://github.com/jasonotu/MAVIDS

Install requirements:

cd MAVIDS
pip install -r requirements.txt

Make migrations and migrate:

python manage.py makemigrations gcsclient
python manage.py migrate

Run the webserver:

python manage.py runserver

Once the webserver is running, you can create an administrative user:

python manage.py createsuperuser

A dashboard user can now be created in the Django backend: http://127.0.0.1:8000/admin/login

Training and tuning

Pre-processing and training can be done using the Jupyter notebooks located in the training folder. For more information on how this is done, see related publications. Training data is also available on IEEE DataPort.

Credits

The PX4 Autopilot was used for testing along with a number of tools from the PX4 project such as Gazebo plugins and ULog scripts.

Publications

Whelan, Jason, et al. "Novelty-based Intrusion Detection of Sensor Attacks on Unmanned Aerial Vehicles." Proceedings of the 16th ACM symposium on QoS and security for wireless and mobile networks. 2020.

Whelan, Jason P. MAVIDS: An Intelligent Intrusion Detection System for Autonomous Unmanned Aerial Vehicles. Diss. 2021.

Jason Whelan, Thanigajan Sangarapillai, Omar Minawi, Abdulaziz Almehmadi, Khalil El-Khatib, February 26, 2020, "UAV Attack Dataset", IEEE Dataport, doi: (https://dx.doi.org/10.21227/00dg-0d12).