Sequential Multi-Factor Authentication for Attendance Using LBPH and Geolocation

Abstract

Conventional attendance systems remain vulnerable to proxy attendance, location manipulation, and inefficient administrative processes, while many existing intelligent attendance solutions rely on computationally intensive deep learning models that are less suitable for lightweight web-based implementations. This study addresses these limitations by proposing a web-based smart attendance system integrating GPS-based geofencing and Local Binary Pattern Histogram (LBPH) within a Sequential Multi-Factor Authentication (MFA) framework. The proposed framework also incorporates GPS spoofing detection based on abnormal geolocation properties and IP geolocation consistency to improve resistance against location manipulation. The authentication mechanism performs geofencing validation before facial verification to improve security and avoid unnecessary biometric processing. The system was developed using the Laravel framework integrated with Python-based facial recognition and evaluated through black-box functional testing. Experimental evaluation involving 120 attendance scenarios achieved an overall accuracy of 96.67% while effectively detecting GPS spoofing and presentation attacks with low false acceptance and false rejection rates. The implementation results demonstrated that all major functional modules operated according to the predefined requirements, while the confidence threshold of 40 provided reliable facial verification under moderate environmental variations. These findings indicate that the proposed framework provides a practical, secure, and lightweight solution for web-based attendance management and offers an effective alternative to computationally intensive deep learning-based attendance systems.