The use of Biometrics keeps growing. Every day, we use biometric recognition to unlock our phones or to have access to places such as the gym or the office, so we rely on what security manufacturers offer when protecting our privileges and private life. Moreover, an error in a biometric system can mean that a person can have access to an unintended property, critical infrastructure or cross a border. Thus, there is a growing interest on ensuring that biometric systems work correctly on two fronts: our personal information (smartphones, personal computers) and national security (borders, critical infrastructures).
Given that nowadays we store increasing sensitive data on our mobile devices (documents, photos, bank accounts, etc.), it is crucial to know how secure the protection of the phone really is. Most new smartphones include an embedded fingerprint sensor due to its improved comfort, speed and, as manufacturers claim, security. In the last decades, many studies and tests have shown that it is possible to steal a person’s fingerprint and reproduce it, with the intention of impersonating them. This has become a bigger problem as the adoption of fingerprint sensor cell phones have become mainstream.
For the case of border control and critical infrastructures, biometric recognition eases the task of person identification and black-list checking. Although the performance rates for verification and identification have dropped in the last decades, protection against vulnerabilities is still under heavy development. There have been cases in the past where fake fingers have been used to surpass the security of such entities.
The first necessary step for overcoming these issues is to have a common ground for performing security evaluations. This way, different systems’ abilities to detect and reject fake fingerprints can be measured and compared against each other. This is achieved by standardization and the corresponding certification of biometric systems. The new software and hardware presentation attack detection techniques shall undergo tests that follow such standards.
The aim of this Thesis is two-fold: evaluating commercial fingerprint biometric systems against presentation attacks (fake fingers) and developing a new presentation attack detection method for overcoming these attacks. Moreover, through this process, several contributions were proposed and accepted in international ISO standards.
On the first matter, a few questions are meant to be answered: it is well known that it is possible to hack a smartphone using fake fingers made of Play-Doh and other easy-to-obtain materials but, to what extent? Is this true for all users or only for specialists with deep knowledge on Biometrics? Does it matter who the person doing the attack is, or are all attackers the same when they have the same base knowledge? Are smartphone fingerprint sensors as reliable as desktop sensors? What is the easiest way of stealing a fingerprint from someone? To answer these, five experiments were performed on several desktop and smartphone fingerprint readers, including many different attackers and fingerprint readers. As a general result, all smartphone capture devices could be successfully hacked by inexperienced people with no background in Biometrics. All of the evaluations followed the pertinent standards, ISO/IEC 30107 Parts 3 and 4 and Common Criteria and an analysis of the attack potential was carried out. Moreover, the knowledge gathered during this process served to make methodological contributions to the above-mentioned standards.
Once some expertise had been gathered on attacking fingerprint sensors, it was decided to develop a new method to detect fake fingerprints. The aim was to find a low-cost and efficient system to solve this issue. As a result, a new optical system was used to capture fingerprints and classify them into real or fake samples. The system was tested by performing an evaluation using 5 different fake finger materials, obtaining much lower error rates than those reported in the state of the art at the moment this Thesis was written.
The contributions of this Thesis include: • Improvements on the presentation attack detection evaluation methodology.
• Contributions to ISO/IEC 30107 - Biometric presentation attack detection - Part 3: Testing and reporting and Part 4: Profile for evaluation of mobile devices.
• Presentation attack detection evaluations on commercial desktop and smartphone fingerprint sensors following ISO/IEC 30107-3 and 4.
• A new low-cost and efficient optical presentation attack detection mechanism and an evaluation on the said system.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados