Like any identification system designed to safeguard sensitive information (passwords, PIN cards, etc.), a biometric fingerprint system can also get an attack. To avoid this, the developers of biometric systems try to detect all possible fraud attempts. This field of research is known as anti-spoofing.
Silicone fingerprint. Max Braun with CC License.
Despite their apparent high level of security, they present some points where they can be attacked. From fake silicone fingerprints to cutting a person’s finger, several methods have been used to circumvent safety. Simultaneously, those who are dedicated to anti-spoofing have developed techniques to defend the systems from these attacks.
The most common attacks can be classified into the following categories:
- Attack to the scanner / sensor. The attacker may try to destroy the scanner, present a latent fingerprint or an image of the fingerprint it intends to impersonate. Rubber or silicone fingerprints can also be made to mimic another fingerprint.
- Attack on communication between the scanner and the characteristics extractor. This communication can be intercepted and it would be possible to enter other information to replace the one sent by the scanner.
- Attack to the communication between the extractor of characteristics and the matcher. Similar to the previous attack, a hacker can try sending information entered in the communication channel between the characteristics extractor and the matcher.
- Attack to the matcher. A program can impersonate the matcher as a Trojan and send a false score or a YES or NO decision to the authentication application.
- Attack to the system database. An attacker can seize the information of the users by accessing the database or modifying it. This attack can occur at any time, such as the recognition phase or even while the user is registering.
- Attack on the communication between the matcher and the database. The communication between the matcher and the database can be interfered to extract the information that is sent. In this way, the user’s information can be obtained to reproduce it other times.
- Attack between the matcher and the application requesting verification. The information that circulates through the channel between the application that requests a verification and the matcher can be extracted to replicate it at another time simulating a successful or failed authentication according to the will of the attacker.
To prevent attacks, manufacturers are developing various life detection techniques to prevent synthetic fingers or fingers from deceased people from being used to obtain fraudulent identification in a biometric fingerprint recognition system.
The most widespread life detection methods in fingerprint recognition systems are:
- Ultrasounds: through ultrasounds, information can be obtained from the inner layers of the skin or even from the blood flow. This method is not viable because its automation is complex and usually requires a post-analysis by an expert.
- Flexibility: starting from the fact that the pressure of the finger on the surface of the scanner is not homogeneous but decreases from the center to the edges. If the user moves the finger while in contact with the scanner, the elasticity of the skin can be determined. Anyway it is possible to find materials whose elastic properties are similar to those of the skin.
- Electrical measurements: the conductivity or impedance of the skin can be measured to determine if a real fingerprint is actually present in the scanner. These parameters are very variable depending on the skin conditions, so their modeling is complex. No reliable results have been achieved even with these techniques. Another measurable electrical parameter is the electrocardiogram. Its results are more reliable, although it is possible to create very fine silicone fingerprints that, attached to a real finger, cause the pulse to be detected in the attacker’s finger.
- Measurements of light: the transluminance of the finger can be measured, detecting the light that passes through the finger or the reflection of light on the finger by means of a transmitter and a light receiver. Using a photoplethysmography, the cardiac pulse can be measured by detecting the variation in the volume of blood circulating through the finger.
- Oximetry: oximetry is a measure of the amount of oxygen in the blood. It is done by means of infrared LEDs that are located at the end of the finger. In general it is a good technique for the detection of life, although its main problem is that the measurement requires several cardiac pulses so it is quite slow.
- Infrared images (temperatures): it is a method of detecting life from an infrared image in which the changes in the body temperature can be appreciated. The method requires a camera, which makes the system more expensive.
To finish, remember that from Umanick we think that the best defense is a good attack. Therefore, the implementation of multimodal systems is recommended. It is much safer to use different techniques to identify the subject, for example the use of the fingerprint and the iris.