Yousefi, F (2022) Brain Signal as a New Biometric Authentication Method. Doctoral thesis, Liverpool John Moores University.

Preview

Text 2022yousefiphd.pdf - Published Version Available under License Creative Commons Attribution Non-commercial. Download (5MB) | Preview

Abstract

Security authentication defines the process of verifying the identity of a person. For many years, authentication technology has played a crucial role in terms of data security. Existing typical biometric authentication technologies have some limitations. Progress in technology has produced several specific devices with ability to reproduce human biometrics that are currently used, mainly because they are mostly visible and touchable. Therefore, a new biometric method is required to overcome the limitations of current biometric authentication systems. Human brain signals are one of the biometric characteristics that has recently been utilised in diverse Brain-Computer Interface (BCI) applications. Brain-based authentication is achieving more popularity among researchers where research studies have reported considerable accuracy using different BCI methods for authentication purposes. However, there are some limitations in terms of usability, time efficiency, and most importantly the permanency of the method through time. The proposed research suggests two different brain-based authentication methods using picturising patterns and deep breathing patterns as brain states. This process starts with signal acquisition using the above-mentioned brain patterns as user strategies to capture the raw EEG data from the participants, followed by the pre-processing stage for data cleansing and standardisation. For noise removal, filtering techniques and the Independent Component Analysis (ICA) algorithm are utilised to make data ready for the feature extraction. Fast Fourier Transform (FFT), Power Spectrum Density (PSD), and Discrete Wavelet Transform (DWT) are used for the feature extraction from the time-frequency domain data representation. The extracted features are then forwarded to classification methods that include Linear Discriminant Analysis (LDA), Support Vector Machines (SVM), and Artificial Neural Network (ANN). The proposed picturising and deep breathing methods indicate 88% and 91% accuracy respectively when evaluated over pure unseen data. The results show that the picturising pattern method improves the security level according to invisibility and changeability of the brain-ID using images; and deep breathing patterns method improves the usability and permanency of the system because this pattern can be used anywhere and at any time.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	brain-computer interface, biometric authentication, EEG, signal acquisition, signal processing
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science T Technology > T Technology (General)
Divisions:	Computer Science & Mathematics
SWORD Depositor:	A Symplectic
Date Deposited:	23 Aug 2022 08:36
Last Modified:	01 Sep 2023 00:50
DOI or ID number:	10.24377/LJMU.t.00017427
Supervisors:	Kolivand, H
URI:	https://researchonline.ljmu.ac.uk/id/eprint/17427

View Item