Researchers from Nanjing University of Posts and Telecommunications developed an audiovisual presentation paradigm to record the EEG signals of subjects
Increasing development in smart sensing technologies, wireless communications, and mobile intelligent computing has led to significant uses of Internet of Things (IoT). However, increasing adoption of IoT also poses security issues. Integration of novel IoT terminals and machine learning has facilitated development of smart access control and identity authentication. The approach finds application in smart home, intelligent building, or safeguards for the purpose of security enhancement. However, conventional IoT applications rely on user passwords, device PINs, and Radio-frequency (RF) cards for identity authentication. Smart biometric authentication technologies such as fingerprint recognition suffer from several challenges such as replication of biological features. Several studies have therefore, focused on IoT devices based on electroencephalography (EEG) signals.
Now, a team of researchers from Nanjing University of Posts and Telecommunications combined visual and auditory stimuli to develop an audiovisual paradigm for an IoT application scenario. The research focused on development of smart entrance guard for a smart home or building. The EEG-based identity authentication system integrates visual and auditory presentations. The stimulus source that consists self and non-ego face images and voice can stimulate the subject to produce unique brainwave activity. To attain artifact removal and to construct optimal feature subset, several effective methods were selected. The team selected base learners from logistic regression, naïve Bayes, and BP neural network, which were later combined with a robust learner.
The team conducted experiments to validate the reliability, effectiveness, and feasibility of the identity authentication system. According to the team, the characteristics of the EEG-based identity authentication system cannot be forged or replaced. Therefore, the system is suitable for those with normal brain activity but has impaired sensory functions. In further research, the team plans to focus on attaining higher classification accuracy along with higher reliability. The team also plans to integrate the method with other biometric technologies to offer ultra-effective and highly secure authentication system. The research was published in the journal MDPI Sensors on April 9, 2019.
Subscribe to our newsletter to get notification about new updates,information, discount, etc..