Enhanced Keystroke Dynamics Authentication Using Keystroke Vector Dissimilarity

A keystroke dynamics is a simple example of a biometric authentication by using a typing rhythm on a keyboard without any special device. Previous research used Euclidean distance to distinguish between an authentic user and an imposter [1]. However, the Euclidean distance has a flaw in that a distance could be the same even though the keystroke rhythms are different. Therefore, this paper proposed to enhance the keystroke dynamics authentication by using a keystroke vector and calculating a standard deviation for each key to give different weights such that the variations of each keystroke is taking into account. A dataset is similar in [1] but some data is different. The performance was measured using Equal Error Rate (EER) and the result showed that the EER was 3.33%, compared to 4% in [1].