When someone accesses their mobile device by scanning their face or finger, they are using biometric authentication (BA).
Dr. Vibhuti Gupta, assistant professor of computer science and data science, recently worked on a study to pursue a new biometric authentication classifier method for smart phone devices. The study is part of an important area of mHealth research, data security and privacy.
“Without biometric authentication, unauthorized people can attack biomedical devices, disrupt their functioning or otherwise lead to adverse outcomes,” says Dr. Gupta.
Biometric authentication is the process by which physiological measurements can identify a specific person. BA captures a person’s photoplethysmography (PPG) signals. PPG is an inexpensive and unobtrusive method for estimating oxygen saturation levels in blood and other physiological parameters such as respiration rate and heart rate.
Dr. Gupta and his colleagues developed a smartphone, PPG-based, BA system solution that analyzes the PPG signal obtained by a subject’s smartphone camera and authenticates the subject based on the PPG analysis results. The experimental results of the EBT-based BA algorithm achieved acceptable false positive levels of 98 percent and false negative levels of 95 percent. The equal error rates measurements also accomplished acceptable values of 2.42 and 5.9 percent.
Dr. Gupta worked with Dr. Bengie L. Ortiz, Dr. Jo Woon Chong, Dr. Monay Shoushan, Dr. Kwanghee Jung, and Dr. Tim Dallas, each from Texas Tech University, on the study. He helped set up the experiment and analyzed the results. He also built a deep learning model that will be compared to this study in the future.
The study is a natural fit for Dr. Gupta whose research interests include mHealth wearable sensor data analytics.
“I pursued this study to initiate collaboration with the SmartBioMed which is a recognized lab for biomedical signal processing, wearables and mHealth at Texas Tech University,” says Dr. Gupta. “They are also working on smartphone and wearables research and are developing AI/ML methods to analyze data captured from biomedical devices. This paper is a small part of it.” The paper, A Biometric Authentication Technique Using Smartphone Fingertip Photoplethysmography Signals, is available online.