Secure biometric systems based on bio-signals and DNA encryption of optical spectrograms

Area of Science:

• Cybersecurity
• Biometric Systems
• Signal Processing

Background:

• Biometrics are increasingly vital for identity verification, especially in Internet of Things (IoT) security.
• Existing biometric systems require enhanced protection against sophisticated hacking attempts.
• Template protection methods like cryptosystems, cancelable biometrics, and hybrid systems are crucial.

Purpose of the Study:

• To propose two novel bio-signal-based biometric authentication systems: one unimodal and one multimodal.
• To implement deoxyribonucleic acid (DNA) encryption for secure template protection.
• To evaluate system performance using standard metrics and assess robustness against noise.

Main Methods:

• Developed a unimodal authentication system using optical spectrograms of bio-signals.
• Created a multimodal system integrating bio-signals via singular value decomposition (SVD).
• Applied deoxyribonucleic acid (DNA) encryption to biometric templates for enhanced security.

Main Results:

• Achieved high accuracy and efficiency, with Equal Error Rate (EER) near 0 and Area Under the ROC Curve (AROC) near 1.
• Obtained excellent performance metrics: False Accept Rate (FAR) of 1.6×10⁻⁸, False Reject Rate (FRR) of 9.05×10⁻⁶, and Decidability (D) of 29.34.
• Demonstrated stable system performance and robustness in the presence of varying noise levels.

Conclusions:

• The proposed unimodal and multimodal biometric authentication systems offer significant advancements in security and accuracy.
• DNA encryption provides a robust method for protecting sensitive biometric templates.
• These systems are highly effective for securing IoT devices and other applications requiring reliable identity verification.