Solution · Behavioural data
For teams building continuous authentication · 2025 to 2026
BEACON
Behavioural fingerprints from gameplay
A 442 GB open dataset of keystrokes, mouse movement, network traffic and screen recordings from competitive Valorant play.
Status: Available for custom data collection
Shows the kind of signals recorded, not real samples
The problem
Continuous authentication recognises a person by how they use a computer. Research on it needs large recordings of real behaviour, captured in sync across the keyboard, the mouse and the network.
What was built
BEACON is our capture suite: a Windows logger that records keystrokes, mouse dynamics, network packets and the screen in sync, and an upload pipeline that checks every file. We used it to build a 442 GB dataset from competitive Valorant play, and we run the same collection for customers on their own users and tasks.
What it does
- Four synchronised streams: keystrokes, mouse dynamics, network packets and screen recordings
- Each session also carries a hardware profile, the game settings and a consent record
- BEACON-Logger, the Windows suite that captured the data, is published on Zenodo
- Uploads are checked with SHA-256 hashes and file-type inspection before they are accepted
What we did
The gameplay data collection, the dataset release and the paper.
- open data
- 442 GB
- synchronised streams
- 4
- DOIs: dataset and logger
- 2