PPG-KeyGen: Using Photoplethysmogram for Key Generation in Wearable Devices

摘要

In this paper, two different photoplethysmogram (PPG) based cryptographic key generation approaches are proposed. The goal is to improve the security of Body Area Networks (BAN) through robust key generation where keys are generated in a plug-n-play manner without requiring key pre-distribution solutions. The Inter-beat Interval (IBI) feature of PPG underlays both of the proposed approaches. The first approach is realized by using Galois linear feedback shift register and consecutive IBI sequences. An IBI is measured from two consecutive peak points of an individual’s PPG signal where they are the tallest and most conspicuous peaks in a PPG signal. The second approach is realized by utilizing the Advanced Encryption Standard (AES) algorithm and IBI as the seed generator for the AES algorithm. The efficiency of the proposed approaches is evaluated using real PPG signals obtained from the PPG-DaLiA dataset of University of California, Irvine. The security analyses of the generated keys are carried out in terms of randomness, distinctiveness, and temporal variance as well as using the NIST benchmark. The analyses show that our key generation approaches provide a better randomness level in comparison to existing approaches relying only on singleton IBI sequences. In addition, our calculations reveal that the average Hamming Distance (HD) between the cryptographic keys generated from PPG signal of two different subjects using the IBI, IBI-Galois, and IBI-AES approaches are: 47.56%, 48.34%, and 49.67%, respectively. Also, our experiments show that the average HD between the cryptographic keys which are generated via PPG signal of the same subject at different time intervals are 47.18%, 48.73%, and 49.51%, respectively.