Fingerprint ID is almost ubiquitous now, but they remain the phone’s biggest security vulnerability. Fake fingers already exist, but researchers at Michigan State University have created a more advanced version that can be used to test fingerprint scanners and make them harder to hack. A technical report detailing the process has been submitted to arXiv, a repository of non-peer-reviewed papers. The paper will also appear later in the journal IEEE Transactions on Information Forensics and Security.
It’s not too difficult to create fake fingers to hack a phone. Researchers at CITERhave done this using 3D-printed molds developed from an image. At the CCC conference in 2014, a security researcher called Starbug used a high-res photo of the German defense minister’s hand to create a working model.
Study author Anil Jain has long worked on these biometric issues. Last year, police asked his lab to re-create a dead man’s fingertips to unlock his phone, and the team succeeded on the second attempt. For this new paper, Jain’s team used a 3D printer to create a mold for the realistic fake finger. The finger itself is made of different types of silicone and pigments.
To understand why the new finger is better, let’s look at how the technology works. Fingerprint readers use different methods of ID. Some are optical scanners that basically take a photo of your fingertip and see if it matches the one stored. Some use the conductivity of skin and electrical currents to create the image instead of light; these are called capacitive scanners and can create a clearer picture. Another method uses ultrasound. In these, an ultrasonic pulse presses against your finger. Depending on the specific ridges of your fingertips, some of the pulse is absorbed and some goes back to the sensor. The sensor then uses mechanical stress to detect the details of your fingertips. Ultrasound sensors see deeper than the other two, which focus on the surface level, and are more expensive and thus rarely used.
The lab’s new finger has the optical, electrical, and mechanical properties of a real finger — as opposed to most fake fingers, which only have a couple of these properties. They can be used to test the accuracy of optical versus capacitive sensors. Those are the two most common kinds, but sometimes their readings don’t match up. The idea is that the new finger can help designers make their readers more accurate and sensitive.
Next, Jain and his team will design their own fingerprint reader to test how good these fake fingers are. Meanwhile, the rest of us are obsessing over the new iPhone’s facial recognition technology. Who knows what Jain’s team will make from that?
COMMENTS