Design and development of a user authentication system using electroencephalography (EEG) signals based on picture gesture authentication (PGA) environments
Loading...
Date
2024-02-28
Authors
Βεργίνης, Δημήτριος
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Nowadays, we all rely on authentication systems to gain access to secured information, using our credentials to bypass the barriers of digital security. However, challenges persist in addressing issues like shoulder surfing and impersonation attacks, particularly in the context of graphical passwords. In this study, we present a Multi-Factor Authentication (MFA) system built upon Picture Gesture Authentication (PGA) schemes, leveraging Electroencephalogram (EEG) as a biometric factor to mitigate risks associated with impersonation and shoulder surfing attacks. Our experimentation involved 30 participants, and we proposed two distinct methods for preprocessing the biometric data. Features were extracted using Power Spectral Density, Time domain statistics, Sample Entropy, and Discrete Wavelet Transform. Subsequently, we employed this data to train machine learning models, namely SVM, NN, Random Forest Classifiers, and Decision Trees. Our efforts resulted in an impressive average accuracy of 98.16±3.29%, providing a robust foundation for further exploration in this research domain.
Description
Keywords
Cybersecurity, Electroengephalography, User authentication, Machine learning