Fused multimodal biometric authentication for an open distance learning environment online examination

Abstract

The recent development of digital sensors such as cameras and scanners for capturing images and sound analysis systems for voice verifications has contributed significantly to authentication mechanisms in many applications. The validation of students' identity during examinations in an Open-Distance Electronic Learning (ODeL) environment continues to pose challenges. As a result, much attention has not been given to the implementation of authentication techniques in this area to prevent a third party from dictating answers at an unnoticeable location to an agent during an examination. This paper proposes a fused multimodal biometric authentication model based on facial recognition, voice recognition, and keystroke dynamics. An experiment conducted on Moodle Learning Management System (LMS) on a client-server platform for 220 diverse students and 25 imposter students indicated prediction accuracy of 90% with a False Acceptance Rate (FAR) of 1.11% and False Rejection Rate (FRR) of 3.8%. The results show significant improvement in authenticating students and preventing the dictator's assistance in examination in the ODeL platform. The results will be useful for assessing quality management systems in ODeL universities and any organisation conducting virtual platform examinations.