Toward Improving Electrocardiogram (ECG) Biometric Verification using Mobile Sensors: A Two-Stage Classifier Approach

Area of Science:

• Biometrics and Signal Processing
• Cybersecurity and Data Security
• Mobile Health Technology

Background:

• Electrocardiogram (ECG) signals from mobile devices present opportunities for biometric identity recognition.
• Enhanced data security is crucial for remote access control systems.
• Current biometric systems may lack robustness in diverse conditions.

Purpose of the Study:

• To develop and evaluate a novel two-stage classifier for robust ECG-based biometric identification using mobile devices.
• To improve the effectiveness and accuracy of biometric recognition systems leveraging mobile ECG data.
• To assess the algorithm's performance across a diverse dataset including various health conditions.

Main Methods:

• A two-stage classification algorithm combining Random Forest and Wavelet Distance Measure was developed.
• A probabilistic threshold schema was integrated to optimize classification decisions.
• The algorithm was validated on a mixed dataset comprising 184 subjects.

Main Results:

• The proposed two-stage classifier achieved a subject verification accuracy of 99.52%.
• This performance surpasses the accuracy of Random Forest alone (98.33%) and Wavelet Distance Measure alone (96.31%).
• The algorithm demonstrated superior effectiveness and robustness in biometric identification.

Conclusions:

• The novel two-stage classifier significantly enhances ECG-based biometric identification accuracy and reliability.
• The proposed method is practical for applications requiring high data security, such as cloud security, cybersecurity, and remote healthcare.
• Mobile ECG biometrics offer a promising solution for secure remote authentication.