A new multimodal behavioural biometrie authentication framework for system security

Abstract

Unauthorized user access to a computer system is a major security breach to many personal and corporate system users. Due to this, data loss and data modification can occur which may cause serious problem in the real world. The user authentication mechanisms applied to the computer systems must be very secure. For achieving high level security, transparent authentication mechanisms must be implemented. The traditional password based authentication mechanism as well as biometric(physiological) authentication mechanism use one time authentication, that is at point of entry only. If an intruder have access in the system, then it may access the sensitive data and misuse it. Thus, a transparent authentication of a user is mandatory. Considering the challenges posed by behavioral biometrics like: low repeatability and high redundancy, we are proposing here a multimodal behavioural biometric based authentication technique using keystroke dynamics and mouse dynamics. We have shown a comparison of unibiometrics with multimodal biometric authentication and found, in average case single biometrics achieved 20–22% of equal error rate(EER) where as using multimodal biometric achieved 8% EER.