Physically Unclonable Functions: Principle, Advantages and Limitations

Abstract

The Physically Unclonable Function (PUF) has become an inescapable security primitive, as it generates a fingerprint unique to each device, and is natively robust against reverse engineering attacks. It allows the user to devise authentication protocols or protect specific software or hardware blocks. However, it natively suffers from a lack of reliability, is sensible to modeling and physical attacks, and its entropy must ensure a good uniquenesses. Hence, it requires specific method and processing blocks to make it robust steady and unique. This talk gives an overview of PUF principles, its main applications, the main architecture types and the ways to correct its inherent defects to provide a trusted primitive.