Biometric Tech Leads Target Security, Inclusivity

The Department of Homeland Security (DHS) and National Science Foundation (NSF) are making biometric technology more robust to enhance its performance in densely populated and austere environments. Researchers are focusing on three key areas as they build out the tech’s capabilities: security, inclusivity and connectivity.

“There are reasons for the use of biometrics within the department, but we are also responsible for helping to screen and inspect people crossing the borders, entering secure facilities, traveling and flying, to help make sure that the right person is going through the right process,” Arun Vemury, director of the Biometric and Identity Technology Center at DHS, told GovCIO Media & Research.

Bad actors are leveraging new attack modes to expose the gap between biometric unlocking, a security feature that uses unique biological characteristics to verify an individual’s identity for device access, and authenticating non-human identities (NHI).

Dr. Stephanie Schuckers, director of the NSF’s Center for Identification Technology Research (CITeR), is conducting research to improve security of biometric technologies.

In the age of digital transformation and the demand for streamlined processes, users often grant access to various third-party applications. This frequently leads to NHIs gaining access to sensitive data, such as application programming interfaces (APIs). If these third-party applications are compromised, malicious actors can exploit them to move laterally within an organization or infiltrate other connected organizations.

“That’s a key piece of biometrics, not just that it matches, which is obviously one key piece, but also that you’ve captured from that real life person at the time of making the decision,” said Schuckers.

The research community is also focusing on making biometric identification systems more inclusive. Vemury highlighted current technologies might not work well for everyone, especially technologies that are based on gender, race, age or skin tone.

Researchers have historically used the Fitzpatrick skin test, a five-point scale used to assign skin tones. It was later discovered that the Fitzpatrick skin test was developed to determine ultraviolet radiation sensitivity for caucasians.

“A lot of systems can perform poorly if people have very light-colored skin and maybe the face photos are overexposed, or people have very dark skin tones and their faces are underexposed,” said Vemury. “At the end of the day, the point of these things is to have a scale to measure, as well as encourage people to directly measure.”

Schuckers and Vemury said they are seeing improvements in facial recognition technologies as AI gains traction, adding that AI may be applied to iris and fingerprint recognition technologies in the future.

“We need some supplementary technologies to help fill in the gaps. Getting a good photo from everybody is not trivial or [something] that’s true for all biometric modalities,” said Vemury. “There are some people who have mobility issues or other challenges, and it’s hard for them to provide a fingerprint through those scanners. … How do we get better samples, regardless of people is their demographic category characteristics?”

Environmental constraints also come into play when identifying people. For example, collecting biometrics at sea with the Coast Guard would need different features than the biometric technology used at the border with Customs and Border Protection.

“Making these technologies more robust to environmental and use case situations, there’s a lot of interest in potentially new modalities … especially as we start getting to more things like smartphone based or remote applications,” said Vemury.