As adoption of biometric authentication increases, it’s important to understand the security methods used to protect biometric data, writes GREG SARRAIL, VP of Solutions Business Development at Biometrics at HID Global.
Biometric solutions are rapidly becoming the new standard for providing secure and convenient identity verification for consumers and corporations. In recent years, biometric technologies have been adopted to enhance security on mobile devices, secure access to facilities and even validate individual identity within the banking industry. When faced with new technology, many people question the security of the solution. Where does the biometric data reside? Is it protected? Can it be easily accessed? If the data is compromised, can it be used maliciously?
Protect and/or render useless
Biometric fingerprint data is the information that is obtained by capturing unique features from an individual fingerprint image. There are several ways to protect this information to ensure that it cannot be openly accessed and used for fraudulent means. During user authentication, the biometric data collected by the sensor must match the information that was captured during enrolment and is stored on a back-end system. Most biometric systems use templates, mathematical representations of biometric data, rather than a raw image of a fingerprint. Templates are much smaller than full images, which decreases the time required to provide a match, minimizes storage requirements and protects user privacy because a fingerprint image cannot be reconstructed from a template. Some systems provide an additional layer of security by encrypting the transport tunnel and even the templates themselves to ensure the data is protected as it moves from the sensor to the back-end system.
Additional security methods can be deployed which are more dependent on the specific use case. For example, in an ATM setting, a user’s biometric information can be augmented before it is stored in a uniform way. This security practice is called “salting” and is done by combining the individual’s PIN and the fingerprint data prior to being stored. When verifying the biometric information, the same PIN is used with the same salting algorithm to provide a match. The advantage of this approach is that the back-end database does not contain an image of a fingerprint or even a standard template, but rather the combined “salted” template. This approach increases both the security and privacy of a system.
An alternate approach is to eliminate the back-end database altogether by placing the secured biometric information on a card that is carried by the user. The new South African National ID, for example, is an identity card that securely stores an individual’s unique biometric fingerprint information that was captured during the enrolment process and was written to the card. This card is then presented at the time of verification. After the individual places a finger on the sensor the information is matched locally against the data stored on the card. No database must be queried; the transaction simply confirms that the identity of the user matches the identity stored on the card. This approach reduces the reliance on the back-end database and external transmission security.
Biometrics is the measurement of physiological characteristics; characteristics that are unique to each individual. Facial characteristics are plainly available — this is how people recognize each other, after all — and fingerprints are left behind at every restaurant, subway rail or door that we touch. A secure system must ensure that an individual, and only that individual, can use his or her own biometric data to authenticate. Thus, it is not enough to simply match biometric characteristics against enrolled data, since access to your fingerprint information isn’t protected. A secure fingerprint system will evaluate whether the finger being presented is real or simply a falsified representation of actual fingerprint data. This capability is called liveness detection and it provides an important way to secure biometric information. Liveness detection reduces the ability for a fraudster to use a fake finger or replay stolen biometric data since the data is useless without a live finger. Whichever combination of security methods are used to secure your identity, the ultimate goal is to render biometric data useless if a perpetrator were to access it.
Verify, not identify
In the non-criminal setting, biometrics is typically used to verify an individual and not to identify an individual. To verify a person’s identity the goal is to confirm with the highest level of assurance that the person is who he or she claims to be. Commercial applications often use demographic information, account numbers, card numbers or digital certificates in addition to the fingerprint data to determine a match.
Criminal systems typically don’t have any other information aside from the fingerprint, or partial fingerprint, and therefore must determine an identity with only the biometric data. This process utilizes a large back-end database to compare individual unique features of a fingerprint and to find probable matches among a stored database of fingerprint templates. This process is time intensive and expensive and is not often used in a commercial setting.
Biometric security systems are as unique as fingerprints. Yet, good biometric systems combine the use of fingerprint templates with liveness detection to validate the identity of the right individual. Successful biometric systems are designed in accordance with the specific use case and with the desired results in mind: secure, convenient and reliable authentication that properly verifies the right individuals and rejects the wrong.