Research software accompanying Continuous Mobile User Authentication Using Combined Biometric Traits, developed from my master’s thesis at the University of Applied Sciences Upper Austria, Campus Hagenberg.
Instead of only checking identity when a phone is unlocked, this work explores whether touch gestures and accelerometer readings can help verify the user throughout a session.
Three parts of the system
- An Android recorder gathers touch and accelerometer data.
- Python notebooks extract features and train a Hidden Markov Model using
hmmlearn. - An Android verifier imports the trained model and evaluates new input.
The repository includes the publication’s supporting code and data. Read the paper summary and citation.