Analysis of free-text keystroke dynamics for Arabic language using Euclidean distance

Abstract

Online services play an increasingly important role in executing tasks in many aspects of our lives. These services depend greatly on authentication mechanisms in order to give sufficient protection to the provided data. Since the traditional username and password scheme suffers from various security and usability drawbacks, this paper investigates a free-text keystroke dynamic authentication approach which provides security with high level of usability. In particular, this study extends previous studies in which the Arabic language is incorporated by investigating the utilization of different timing features, as well as applying Euclidean distance classification method. Based on a controlled experiment, the False Accept Rates (FAR) was 0.2, while the False Reject Rates (FRR) was 0.0. The results showed that the system performance is enhanced compared to other studies.