User authentication for keypad-based devices using keystroke analysis
The use of a Personal Identification Number (PIN) is a common means of ensuring user authentication on numeric keypad devices. However, like passwords and other forms of authentication based upon secret knowledge, PINs have the potential weakness that they may become known to other people.Ord T, Furnell SM
This paper describes a potential approach for strengthening PIN-based authentication, by incorporating a biometric measurement of the user?s typing style when keying in their number. Such keystroke analysis techniques have previously been used in a full keyboard context, but the keypad scenario is considered to represent a more complex problem.
An experimental study is described in which a neural network approach was used to classify and discriminate between 14 test subjects. The main results, using a 6-digit PIN, yielded a False Acceptance Rate (FAR) of 9.9%, with an accompanying False Rejection Rate (FRR) of 30%. Further experiments were able to significantly reduce the error, but at the expense of a longer PIN. The paper also considers potential application areas, in view of the results observed.