Intellectual Property Navigator

Computer User Authentication using Hidden Markov Model

ID# 2006-3191
Example desktop computer setup

Technology Summary

The disclosed invention is a novel method of authenticating computer access by use of keystroke- based identification. Existing methods which use Keystroke dynamics such as neural networks, statistical classification techniques, and decision trees cannot add or delete users without retraining the entire system. This method uses multiple observation sequences to develop a unique Hidden Markov Model (HMM) for each user. Authentication is made in two stages: (1) Identification and then (2) Verification. User authentication is based on probability score results from each of these two stages and customizable threshold criteria. The invented technique is user authentication system for all applications with keyboard-based access.

Application & Market Utility

HMM keeps track of variability in keystroke patterns and adapts to changes. The level of security can be enhanced or relaxed based on threshold criteria. The system is appropriate for all keyboard-based access applications. It’s non-intrusive and inexpensive to implement. It has the ability to dynamically add or remove users without retraining the entire system.

Generate PDF

Have more questions?

Please contact the Penn State Office of Technology Management.


814-865-6277
otminfo@psu.edu