An embedded system for on field testing of human identification using ECG biometric

Abstract

In this paper a complete system for on field testing of the human identification using Electrocardiograms (ECG) biometric is proposed. The enrollment and test procedures are realized in software, while the recognition is implemented in real time on an embedded platform. It uses the wearable Vitalsens wireless sensor with ECG electrodes placed on the chest of the person to be identified, the ECG sensors communicate via Bluetooth with the LM058 Bluetooth adapter connected to the RS232 interface of the RC10 Field Programmable Gate Array (FPGA) prototyping board. A new human identification method based on the fiducial independent feature extraction from ECG signals is implemented on the low power Spartan 3L FPGA chip available on the board. The Principal Component Analysis (PCA) is exploited to select the main significant features. The projected ECG signals on the principal components are then compared by using the Euclidian distance metric. By occupying just the 45% of logic resources and 75% of the BRAM blocks, the embedded system reaches an identification accuracy of 90%.