PUF-PRFs: A New Tamper-resilient Cryptographic Primitive

With the proliferation of physical attacks that may compromise even the theoretically strongest cryptographic schemes, the need for affordable physical protection of cryptographic devices becomes more visible by each day. In this context, Physically Unclonable Functions (PUFs), a promising new technology, provide a low cost technique to realize tamper-resilient storage for secret keys in integrated circuits (ICs). However, PUFs possess some unusual properties that set them apart from ordinary hardware components: their responses are noisy and tend to change when PUFs are manipulated through external influences. These properties have limited the applications of PUFs so far to only physically protecting stored key material. This raises the question as to what extent PUFs can be used to construct other cryptographic schemes. In this paper, we take the first step towards answering this question and place PUFs in the core of a pseudorandom function (PRF) construction. PRFs are one of the most important cryptographic primitives used to design various cryptographic schemes such as stream or block ciphers. We first give a theoretical model for PUFs and justify it by real-life PUF-implementations. Then, we show how to use PUFs to construct tamper-resilient PRFs, termed as PUF-PRFs. However, for several reasons that we outline in this work, PUF-PRFs cannot directly replace PRFs. Nevertheless, we show that PUF-PRFs represent a new cryptographic primitive with very useful properties: they are inherently resilient to tampering attacks and allow for generating (almost) uniformly distributed values.