On the performance of helper data template protection schemes

The use of biometrics looks promising as it is already being applied in elec- tronic passports, ePassports, on a global scale. Because the biometric data has to be stored as a reference template on either a central or personal storage de- vice, its wide-spread use introduces new security and privacy risks such as (i) identity fraud, (ii) cross-matching, (iii) irrevocability and (iv) leaking sensitive medical information. Mitigating these risks is essential to obtain the accep- tance from the subjects of the biometric systems and therefore facilitating the successful implementation on a large-scale basis. A solution to mitigate these risks is to use template protection techniques. The required protection properties of the stored reference template according to ISO guidelines are (i) irreversibility, (ii) renewability and (iii) unlinkability. A known template protection scheme is the helper data system (HDS). The fun- damental principle of the HDS is to bind a key with the biometric sample with use of helper data and cryptography, as such that the key can be reproduced or released given another biometric sample of the same subject. The identity check is then performed in a secure way by comparing the hash of the key. Hence, the size of the key determines the amount of protection. This thesis extensively investigates the HDS system, namely (i) the the- oretical classication performance, (ii) the maximum key size, (iii) the irre- versibility and unlinkability properties, and (iv) the optimal multi-sample and multi-algorithm fusion method. The theoretical classication performance of the biometric system is deter- mined by assuming that the features extracted from the biometric sample are Gaussian distributed. With this assumption we investigate the in uence of the bit extraction scheme on the classication performance. With use of the the- oretical framework, the maximum size of the key is determined by assuming the error-correcting code to operate on Shannon's bound. We also show three vulnerabilities of HDS that aect the irreversibility and unlinkability property and propose solutions. Finally, we study the optimal level of applying multi- sample and multi-algorithm fusion with the HDS at either feature-, score-, or decision-level.