Anomaly Detection In Mesh Networks

Inspiration

A router is a single point of failure along with other communication nodes on a network. In a cyber attack, those single points of failure can bring down the entire system in a short amount of time. The lateral movement speed of adversaries upon exploit can also pose a problem as they can very quickly take over the entire network without limit or mitigation.

What it does

Anomaly detection and threat mitigation using unsupervised machine learning in mesh networks

How we built it

We used posed data sets and our own generated data to validate and create machine learning anomalous detection algorithms, and raspberry pis to attempt to create a mesh network based on Bluetooth protocol.

Challenges we ran into

Raspberry pis had so many issues, they wouldn't connect to the internet, ethernet ports around the room didn't contain access to the network, the Bluetooth kept giving us strange errors, and one of our raspberry pis kept overheating

Accomplishments that we're proud of

We were able to develop a complete working set of python programs that can detect anomalies with high accuracy.

What we learned

Raspberry pis can be very hard to work with when there is no easy access to the internet.

What's next for Anomaly Detection In Mesh Networks

Develop a working mesh net without overbearing issues, and deploy the machine learning software to the mesh net.

Built With

• android
• machine-learning
• matplotlib
• networkx
• python
• raspberry-pi
• scikit-learn
• unsupervised-learning