A Survey on Silicon PUFs
Article : Articles dans des revues internationales ou nationales avec comité de lecture
Integrated Circuits (ICs) and electronic devices have become an integral part of daily human life
(mobile, home, car, etc.). However, specific security measures should be taken to protect the communicated
information to and from these devices. However, the existing conventional security primitives
require large amounts of memory capacity, processing power, and energy resources that contradict the
specific nature of devices. On the other hand, they store secret keys on the devices for future use, making
them vulnerable to physical attacks. A new concept, known as Physically Unclonable Functions
(PUFs), has been recently investigated to mitigate this problem. A PUF is a hardware-specific security
primitive uses the randomness found in the disorder of physical media caused by the manufacturing
variation process to provide cryptographic functionalities. Consequently, PUFs are inexpensive to
fabricate, prohibitively challenging to duplicate, admit no compact mathematical representation, and
are intrinsically tamper-resistant. This manuscript gives a complete survey of PUFs as a promising
research field in security with a wide application, especially with connected devices. First, we motivate
our contribution by comparing it with the existing surveys about PUFs. Then we provide the
needed background to understand PUF architectures and applications by covering: the variability and
randomness concepts, their classes, and properties. Then, we survey the existing initiatives related to
silicon PUFs in terms of implementation and design used to extract unique secret information from
the physical characteristics of an integrated circuit. In addition, we compare the surveyed works in
terms of performance and security. Furthermore, we classify and compare the existing silicon PUF
applications and use cases. Before concluding, we give the principal metrics used to evaluate the
PUFs’ performance and present some related attacks. Finally, we talk about the current limitations of
silicon PUF architectures and applications, and we look at and talk about research opportunities and
major trends.