8. Enhanced Biometric Security and Privacy Using ECG on the Zynq SoC

Electrocardiogram (ECG) waveforms hold valuable and critical information that can be used in connected health where privacy can be as important as vitality. A secure connected health solution for human identification is presented in this chapter. The recognition of patients uses ECG biometric data streaming while the ECG signals are encrypted and decrypted using the advanced encryption standard (AES). Three different ECG databases representing 2372 samples are used. The Xilinx ZC702 Zynq based platform is used for the hardware implementation of the proposed system. High level synthesis is used to develop and implement different IP-cores corresponding to various block of the system including the AES cipher, AES decipher, and recognition blocks. In addition, various ECG identification algorithms [i.e., principal component analysis along with Euclidian distance, k-nearest neighbors, and extended nearest neighbor] are implemented and evaluated before hardware implementation. Finally, hardware implementation results have shown that the real-time requirements have been met. Furthermore, the presented solution outperforms current field programmable gate array based systems in terms of processing time, power consumption, and hardware resources usage. Using the most optimized hardware implementation, a single ECG signal can be processed in 10.71 ms while the system uses 30 % of all available resources on the chip and consumes only 107 mW. Moreover, the classification accuracy is between 94 % and 100 % depending on the classifier and on the dataset used.