Designing A Novel Biometric Authentication System Based on ECG Signals Using Deep Learning

Abstract

Biometric authentication systems play a crucial role in securing sensitive information and resources. This paper proposes a novel approach to biometric authentication by leveraging Electrocardiogram (ECG) signals. The proposed system employs a comprehensive methodology that includes signal processing, feature extraction, wavelet decomposition; QRS wave detection, internal calculation, wave modelling, distance and deviation calculation, averaging threshold, and an Artificial Neural Network (ANN) classifier. The entire system is implemented and evaluated using the MATLAB tool. The process begins with the acquisition of input ECG signals, followed by pre-processing to enhance signal quality. Feature extraction is then performed to capture the unique characteristics of the ECG waveform. Wavelet decomposition is employed to analyse the signal in both time and frequency domains. QRS wave detection identifies the specific components crucial for biometric authentication. Internally, the system calculates various parameters and models the ECG waves to establish a robust representation. Distance and deviation calculations further refine the feature set. An averaging threshold is applied to enhance the system's resilience to noise and variability. The final classification is accomplished through an ANN classifier trained on the extracted features. The proposed system outputs the authentication result, displaying whether the individual is identified as female or male. The system's performance is evaluated through extensive testing using a dataset of ECG signals. Results indicate high accuracy, demonstrating the effectiveness of the proposed biometric authentication system. This research contributes to the field of biometrics by introducing a novel approach based on ECG signals, offering a secure and reliable means of personal identification. The integration of deep learning techniques enhances the system's ability to adapt to variations in ECG patterns, making it suitable for real-world applications in security and access control.