The fingerprint recognition system may suffer attacks at different points during the authentication process. The following figure shows the possible points. In each of these point the data may be altered and forced an authentication of unregistered user.
- The most common attacks occur by the use of fake fingerprint during the capture of image. A fake fingerprint are build from latent fingerprint left at touched items such as glasses, doorknobs, glossy paper, etc. Using this fingerprint are build three-dimensional molds of rubber membrane, glue, or gelatin.
- During the transmission of the image to the feature extractor may occur interception of the channel, and consequently, the fingerprint image may be stolen and later, used for fake fingerprint construction or for directly access to feature extractor by bypassing the scanner.
- The feature extractor may be substitute by a Trojan horse, which bypass the feature extractor and generate artificial template and submit to the matcher.
- The transmission channel between the feature extractor and matching may also be intercepted and the fingerprint feature may be stored for the later use.
- In the matching module may occur the same problem as in the feature extractor. The presence of Trojan horse may produce always the desired result independent of the input fingerprint.
- The database may also suffer attack of Trojan horse, by which can be created artificial record and submit to the matching module.
- The record of legitimate user may be stolen by intercepting the communication channel between the database and matching.
- Finally, the channel between the matching module and the application requesting verification is also susceptible for possible attacks.
All these attacks are similar to those presented in token and knowledge based authentication (password) system, except the case of attack using fake fingerprints which is particular of fingerprint recognition system. In this subsection are presented some recommendation to countermeasure the possible attacks, in especial the attacks by the use of fake fingerprint.
Independently how the fingerprint was stolen, the fingerprint scanners should be able to reject the fake fingerprints. However detecting the aliveness of a finger it is not an easy task.
The main problem relies on how to differentiate a live finger from that one made of some synthetic material. There has been proposed some ideas to deal with this problem, which consist in using the thermal, electric and optical properties of the material presented to the fingerprint scanner. By using the temperature information, for example, it is expected that the fake finger made of silicone rubber is about 2 degree cooler than a live finger, however, due the temperature variation of the environment and the possibility of artificial heating the fake finger, the thermal measurements are not very reliable.
The conductivity is another measure that could be explored, however, the conductivity of a live finger varies a lot depending of weather condition such as humidity and temperature. The optical properties such absorption, reflection, scattering and refraction, in the human skin are different than many other synthetic material. However, it is not difficult to find materials
that have optical properties close to those of a live finger.
As we can see, there no exists a reliable characteristic that could reject all fake fingerprints. Therefore, it is important to take special attention during the design and development of a secure fingerprint system. Here are listed some considerations in order to improve the security.
- Enroll and use multiple finger for single authentication
- Change occasionally the patterns by using multiple fingerprint
- Use device that detect better the aliveness of the finger
- For more secure system, include password verification
- Occasionally re-enroll the fingerprint
- Allow identification to occur only from a certain fingerprint scanner
- Reduce the sensitivity to reduce the possibility of false positive
- Control the physical access to fingerprint scanner where possible
- Make regular maintenance of devices for heavy usage environments