Ileana Buhan

Cryptographic keys from noisy data, theory and applications

Biometric security systems that verify a persons identity by scanning fingers, hands, eye or face are becoming more and more common. As a result biometrics is one of the fastest growing industries. Applications for biometrics range from homeland security physical access to various facilities and health and social services. Utilizing biometrics for personal authentication is more convenient and than current methods such as passwords or PINs. Another important advantage of biometric authentication is that it links events to a user and is becoming socially acceptable and inexpensive. Biometric authentication requires comparing a registered or enrolled biometric sample against a newly captured biometric sample. However, biometric authentication is not perfect and the output of a biometric authentication system can be subject to errors due to imperfections of the classification algorithm, poor quality of biometric samples, or an intruder who has tampered with the biometric authentication systems. Although biometric authentication is intended primarily to enhance security, storing biometric information in a database introduces new security and privacy risks, which increase if the database is connected to a network. This is the case in most practical situations. This thesis looks at security aspects of biometric authentication and proposes solutions to mitigate the risk of an attacker who tries to misuse biometric information or who bypasses modules of biometric systems to achieve his malicious goals. Our contribution is threefold. Firstly we propose 3W-tree, an analysis tool used to identify critical attack scenarios for a biometric system. We apply the 3W-tree design tool to the SmartGun biometric recognition system with the purpose of identifying critical security issues. Secondly, we explore the challenges of secure template protection, which are both theoretical and practical and we put forward solutions to part of the issues. Thirdly, we present a practical solution to the secure template transfer, which should allow transfer of the biometric traits between two biometrically enabled devices when no security infrastructure is available and the users are no security experts.

Secure pairing with biometrics

Secure pairing enables two devices that share no prior context with each other to agree upon a security association, which they can use to protect their subsequent communication. Secure pairing offers guarantees of the association partner identity and it should be resistant to eavesdropping and to a man-in the middle attack. We propose the SAfE pairing system, a user friendly solution to this problem. Details are presented along with a discussion of the security features, experimental validation with two types of biometric data (face recognition and hand grip pressure pattern) and a usability analysis for face recognition biometric pairing.

A quantitative analysis of indistinguishability for a continuous domain biometric cryptosystem

Biometric information is regarded as highly sensitive information and therefore encryption techniques for biometric information are needed to address security and privacy requirements of biometric information. Most security analyses for these encryption techniques focus on the scenario of one user enrolled in a single biometric system. In practice, biometric systems are deployed at different places and the scenario of one user enrolled in many biometric systems is closer to reality. In this scenario, cross-matching (tracking users enrolled in multiple databases) becomes an important privacy threat. To prevent such cross-matching, various methods to create renewable and indistinguishable biometric references have been published. In this paper, we investigate the indistinguishability or the protection against cross-matching of a continuous-domain biometric cryptosystem, the QIM. In particular our contributions are as follows. Firstly, we present a technique, which allows an adversary to decide whether two protected biometric reference data come from the same person or not. Secondly, we quantify the probability of success of an adversary who plays the indistinguishability game and thirdly, we compare the probability of success of an adversary to the authentication performance of the biometric system for the MCYT fingerprint database. The results indicate that although biometric cryptosystems represent a step in the direction of privacy enhancement, we are not there yet.

Analytical template protection performance and maximum key size given a Gaussian-modeled biometric source

Template protection techniques are used within biometric systems in order to protect the stored biometric template against privacy and security threats. A great portion of template protection techniques are based on extracting a key from or binding a key to a biometric sample. The achieved protection depends on the size of the key and its closeness to being random. In the literature it can be observed that there is a large variation on the reported key lengths at similar classification performance of the same template protection system, even when based on the same biometric modality and database. In this work we determine the analytical relationship between the system performance and the theoretical maximum key size given a biometric source modeled by parallel Gaussian channels. We consider the case where the source capacity is evenly distributed across all channels and the channels are independent. We also determine the effect of the parameters such as the source capacity, the number of enrolment and verification samples, and the operating point selection on the maximum key size. We show that a trade-off exists between the privacy protection of the biometric system and its convenience for its users.

Embedding Renewable Cryptographic Keys into Continuous Noisy Data

Fuzzy extractor is a powerful but theoretical tool to extract uniform strings from discrete noisy data. Before it can be used in practice, many concerns need to be addressed in advance, such as making the extracted strings renewable and dealing with continuous noisy data. We propose a primitive fuzzy embedderas a practical replacement for fuzzy extractor. Fuzzy embedder naturally supports renewability because it allows a randomly chosen string to be embedded. Fuzzy embedder takes continuous noisy data as input and its performance directly links to the property of the input data. We give a general construction for fuzzy embedder based on the technique of Quantization Index Modulation ( QIM ) and derive the performance result in relation to that of the underlying QIM . In addition, we show that quantization in 2-dimensional space is optimal from the perspective of the length of the embedded string. We also present a concrete construction for fuzzy embedder in 2-dimensional space and compare its performance with that obtained by the 4-square tiling method of Linnartz, et al.[13].

Maximum Key Size and Classification Performance of Fuzzy Commitment for Gaussian Modeled Biometric Sources

Template protection techniques are used within biometric systems in order to protect the stored biometric template against privacy and security threats. A great portion of template protection techniques are based on extracting a key from, or binding a key to the binary vector derived from the biometric sample. The size of the key plays an important role, as the achieved privacy and security mainly depend on the entropy of the key. In the literature, it can be observed that there is a large variation on the reported key lengths at similar classification performance of the same template protection system, even when based on the same biometric modality and database. In this work, we determine the analytical relationship between the classification performance of the fuzzy commitment scheme and the theoretical maximum key size given as input a Gaussian biometric source. We show the effect of the system parameters such as the biometric source capacity, the number of feature components, the number of enrolment and verification samples, and the target performance on the maximum key size. Furthermore, we provide an analysis of the effect of feature interdependencies on the estimated maximum key size and classification performance. Both the theoretical analysis, as well as an experimental evaluation using the MCYT fingerprint database showed that feature interdependencies have a large impact on performance and key size estimates. This property can explain the large deviation in reported key sizes in literature.

A Survey of the Security and Privacy Measures for Anonymous Biometric Authentication Systems

The challenge in applying the known information theoretical measures for biometric authentication systems is that on one hand these measures are defined in a specific context and on the other hand there are several constructions known for the protection of biometric information. The goal of this work is to organize and conceptualize the existing knowledge in the area of security of biometrics and build a bridge between the formal model of cryptography and the practical view of the signal processing area. It is the scope of this paper to build and present the framework where results from both cryptography and signal processing can be integrated.

Feeling is believing: a secure template exchange protocol

We use grip pattern based biometrics as a secure side channel to achieve pre-authentication in a protocol that sets up a secure channel between two hand held devices. The protocol efficiently calculates a shared secret key from biometric data. The protocol is used in an application where grip pattern based biometrics is used to control access to police hand guns.

Embedding renewable cryptographic keys into noisy data

A fuzzy extractor is a powerful but theoretical tool that can be used to extract uniform strings from (discrete) noisy sources. However, when using a fuzzy extractor in practice, extra features are needed, such as the renewability of the extracted strings and the ability to use the fuzzy extractor directly on continuous input data instead of discrete data. Our contribution is threefold. Firstly, we propose a fuzzy embedder as a generalization of the fuzzy extractor. A fuzzy embedder naturally supports renewability, as it allows a string to be embedded instead of extracted. It also supports direct analysis of quantization effects, as it makes no limiting assumptions about the nature of the input source. Secondly, we give a general construction for fuzzy embedders based on the technique of quantization index modulation (QIM). We show that the performance measures of a QIM, as proposed by the watermarking community, translate directly to the security properties of the corresponding fuzzy embedder. Finally, we show that from the perspective of the length of the embedded string, quantization in two dimensions is optimal. We present two practical constructions for a fuzzy embedder in two-dimensional space. The first construction is optimal from reliability perspective, and the second construction is optimal in the length of the embedded string.

Evaluation of a template protection approach to integrate fingerprint biometrics in a PIN-based payment infrastructure

Biometric authentication has a great potential to improve the security, reduce cost, and enhance the customer convenience of payment systems. Despite these benefits, biometric authentication has not yet been adopted by large-scale point-of-sale and automated teller machine systems. This paper aims at providing a better understanding of the benefits and limitations associated with the integration of biometrics in a PIN-based payment authentication system. Based on a review of the market drivers and deployment hurdles, a method is proposed in which biometrics can be seamlessly integrated in a PIN-based authentication infrastructure. By binding a fixed binary, renewable string to a noisy biometric sample, the data privacy and interoperability between issuing and acquiring banks can improve considerably compared to conventional biometric approaches. The biometric system security, cost aspects, and customer convenience are subsequently compared to PIN by means of simulations using fingerprints. The results indicate that the biometric authentication performance is not negatively influenced by the incorporation of key binding and release processes, and that the security expressed as guessing entropy of the biometric key is virtually identical to the current PIN. The data also suggest that for the fingerprint database under test, the claimed benefits for cost reduction, improved security and customer convenience do not convincingly materialize when compared to PIN. This result can in part explain why large-scale biometric payment systems are virtually non-existent in Europe and the United States, and suggests that other biometric modalities than fingerprints may be more appropriate for payment systems.