08-02-2021 | | By Robin Mitchell
Researchers from North Carolina State University have recently demonstrated a proof-of-concept 3D fingerprint scanner that utilises ultrasound to image blood vessels. What challenges do typical fingerprint scanners face, how does this new technology work, and what advantages can it bring?
Fingerprint scanners provide an easy-to-use authentication system as there is no need to remember a PIN or password. Fingerprint scanners are also extremely convenient. They are very fast, are difficult to reproduce without seeing an image of the fingerprint, and unique to every person on the planet.
As a result, fingerprint scanners are being integrated into a wide range of different technologies, including citizen ID systems, smartphones for quick unlocking, and authentication for access areas. However, while biometrics is convenient, it is not entirely the safest method for authentication and can be fooled.
To start, the scanner looks for a fingerprint pattern, and these can be easily lifted off common objects (something which CSI has been doing for decades). A clean pattern can then be transferred to a silicon mould or even a clean piece of plastic that most sensors will recognise.
Personally, I have fooled fingerprint sensors on smartphone devices by simply activating the touch-sensitive fingerprint reader on the very edge of the sensor. If a fingerprint impression remains on the screen, the phone will recognise the impression.
Thus, fingerprint scanners may be able to use a unique pattern to every person on the planet. Still, these patterns do not require confirmation of living tissue (i.e. the pattern is from a valid living finger) and open to potential attacks.
Recently, researchers from the North Carolina State University announced that they had demonstrated a proof-of-concept for a 3D fingerprint scanner. Unlike traditional scanners, the developed system can take a 3D image of the finger and the blood vessels underneath.
The system utilises the same technology found in imaging fetuses; ultrasound. High-frequency pulses of ultrasound are sent into a finger, and capillaries and blood vessels create distortions in the reflected waveform.
The technology demonstrated by the researchers shows image resolutions of 500 x 500 dpi but is currently limited in several ways. To start, the researchers only tested the concept on an artificial finger, which included limited patterns in the fingerprint and a single blood vessel filled with bovine blood. Secondly, the image took one hour to generate, and thirdly only 40% of the blood vessel could be imaged.
The slow imaging time comes from using a single ultrasound transducer, and a finished product would utilise a phased array of transducers. The inability to fully see blood vessels came from the valleys in the fingerprint pattern between ridges. This occurs because ultrasound cannot penetrate the air gap between the transducer and artificial skin well. Thus reflections are virtually undetectable.
While the proof-of-concept demonstrated by the team is in its early days it already demonstrates some advantages over traditional scanners.
To start, 3D technology allows for a system to create a 3D map of a finger instead of a 2D map. This means that fingerprints cannot be spoofed onto the sensor with a simple 2D image, and the sensor will only recognise a 3D model of a finger.
The use of 3D technology also requires that a finger that is attached to the body is physically present. One of the keynotes from this announcement is that the artificial finger had blood flowing through the blood vessel. If a finger were forcibly removed and placed onto the sensor, the blood vessel's pressure would be low (if any pressure at all), and the sensor would detect this.
The 3D fingerprint technology researchers now need to continue their development by changing the read time from 1 hour to 1 second and being able to ignore valleys in a fingerprint pattern so that the correct finger can be inferred. It might be best if the researchers combined their ultrasonic technology with traditional readers so that a traditional 2D reader only operates if it detects flowing blood underneath the fingerprint.