Physically Uncloneable Functions in the Stand-Alone and Universally Composable Framework

In this thesis, we investigate the possibility of basing cryptographic primitives on Physically Uncloneable Functions (PUF). A PUF is a piece of hardware that can be seen as a source of randomness. When a PUF is evaluated on a physical stimulus, it answers with a noisy output. PUFs are unpredictable such that even if a chosen stimulus is given, it should be infeasible to predict the corresponding output without physically evaluating the PUF. Furthermore, PUFs are uncloneable, which means that even if all components of the system are known, it is computational infeasible to model their behavior. In the course of this dissertation, we discuss PUFs in the context of their implementation, their mathematical description, as well as their usage as a cryptographic primitive and in cryptographic protocols. We first give an overview of the most prominent PUF constructions in order to derive subsequently an appropriate mathematical PUF model. It turns out that this is a non- trivial task, because it is not certain which common security properties are generally necessary and achievable due to the numerous PUF implementations. Next, we consider PUFs in security applications. Due to the properties of PUFs, these hardware tokens are good to build authentication protocols that rely on challenge/response pairs. If the number of potential PUF-based challenge/response pairs is large enough, an adversary cannot measure all PUF responses. Therefore, the at- tacker will most likely not be able to answer the challenge of the issuing party even if he had physical access to the PUF for a short time. However, we show that some of the previously suggested protocols are not fully secure in the attacker model where the adversary has physical control of the PUF and the corresponding reader during a short time. Finally, we analyze PUFs in the universally composable (UC) framework for the first time. Although hardware tokens have been considered before in the UC framework, designing PUF-based protocols is fundamentally different from other hardware token approaches. One reason is that the manufacturer of the PUF creates a physical object that outputs pseudorandom values, but where no specific code is running. In fact, the functional behavior of the PUF is unpredictable even for the PUF creator. Thus, only the party in possession of the PUF has full access to the secrets. After formalizing PUFs in the UC framework, we derive efficient UC-secure protocols for basic tasks like oblivious transfer, commitments, and key exchange