Gait recognition could provide biometric cryptography to protect wearable data transmission

Aug 21, 2018 | Chris Burt

Categories Biometric R&D | Biometrics News

The limited computational power of body sensor network (BSN) devices such as wearables makes it difficult to secure the sensitive data they collect, but a team of researchers from the Imperial College London have devised a biometric cryptosystem (BCS) approach which generates keys that pass NIST and Dieharder tests with high efficiency.

With BSNs used to monitor the health of elderly patients or those affected by chronic conditions, BSNs could be attractive targets for hackers.

The researchers’ paper on “An Artificial Neural Network Framework for Gait Based Biometrics” describes the use of physiological and behavioral traits, including face, iris, fingerprint, electrocardiogram (ECG) and photoplethysmography (PPG) to secure wireless communications based on gait signal energy variations and an artificial neural network (ANN).

“State-of-the-art biometrics/wearable security often uses electrocardiogram (ECG), the electrical activity of the heart, but its skin-attached electrodes greatly limit its applications,” Yingnan Sun, the lead author of the paper told TechXplore. “We felt it was necessary to explore a new kind of biometrics that is both easy to collect and non-invasive, and gait, the way people walk, came to mind.”

The system relies on inertial sensors, such as accelerometers, but those are inexpensive, and are aready embedded in almost all wearables, according to TechXplore.

“We found that the use of the proposed ANN framework can significantly increase correlations between gait signals captured by different wearable sensors, resulting in a huge improvement in the performance of the security scheme,” Sun said. “This newly proposed security framework is 68.75 percent more efficient than our previous work, generating a 128-bit key within only 12 seconds of walking.”

Gait recognition technology has been maturing, with a team of scientists from two universities unveiling an accurate method of gait recognition using floor sensors and high-resolution cameras earlier this year as an example.