Supervisors info:
Δημήτριος Συβρίδης, Καθηγητής, Τμήμα Πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
Summary:
Physical unclonable function (PUF), which is the main subject of this dissertation, is defined as a device that takes advantage of the inherent randomness introduced during construction to give a physical entity a unique "fingerprint. Also, authentication confirms the identity of the parties during communications. Apart from that, PUF concepts for electronic applications are many and can be easily integrated and digitized. In addition, a flexible and optical authentication system based on physical indeterminate functions (PUF) is analyzed, while a visual hierarchical certification scheme based on the principle of interference and the hash function was presented. Finally, the machine learning framework is analyzed which includes elements for detecting attacks that can utilize supervised, semi-supervised and unsupervised learning to detect attacks and, where appropriate, determine their type and intensity.
Keywords:
PUF, extensions, authentication, encryption, machine learning