Cancelable HD-sEMG-Based Biometrics for Cross-Application Discrepant Personal Identification.

Area of Science:

• Biometrics and Health Informatics
• Signal Processing and Machine Learning
• Cybersecurity in Healthcare

Background:

• Body sensor network (BSN)-based health informatics raises significant information security concerns.
• Traditional biometrics suffer from noncancelability and cross-application invariance, posing risks if templates are compromised.
• Existing biometric modalities are vulnerable to permanent compromise and cross-account credential theft.

Purpose of the Study:

• To propose a cancelable and cross-application discrepant biometric approach for personal identification using high-density surface electromyogram (HD-sEMG).
• To enhance security in body sensor network (BSN) health informatics by addressing limitations of traditional biometrics.
• To evaluate the feasibility of HD-sEMG signals as unique biometric tokens for multiple accounts per user.

Main Methods:

• Utilized high-density surface electromyogram (HD-sEMG) signals from the right dorsal hand during isometric contractions.
• Employed different finger muscles and isometric contractions to generate distinct biometric tokens for multiple accounts.
• Collected training and testing data 9 days apart to assess signal variation and system robustness.

Main Results:

• Achieved a promising identification accuracy of 85.8% for 44 identities (22 subjects across 2 accounts).
• Demonstrated high accuracy for differentiating between accounts of the same user, indicating cancelability and cross-application discrepancy.
• Confirmed the effectiveness of HD-sEMG biometrics despite signal variations over a 9-day period.

Conclusions:

• The proposed HD-sEMG-based biometric approach offers a promising solution for secure personal identification in BSN health informatics.
• The system exhibits effective cancelability and cross-application discrepancy, mitigating risks associated with compromised biometric templates.
• This study represents the first application of HD-sEMG for personal identification, considering inter-day signal variability.