Photonic implementation of physically unclonable functions

• Autor(a) principal: Silvério, Tiago Filipe Santos
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/32114
• topic: Optical physically unclonable functions; Cryptographic systems; Optical coherent speckle pattern; Internet of things; Anticounterfeiting; Organic-inorganic hybrids

This dissertation aimed to study and develop optical Physically Unclonable Functions, which are physical devices characterized by having random intrinsic variations, thus being eligible towards high security systems due to their unclonability, uniqueness and randomness. With the rapid expansion of technologies such as Internet of Things and the concerns around counterfeited goods, secure and resilient cryptographic systems are in high demand. Moreover the development of digital ecosystems, mobile applications towards transactions now require fast and reliable algorithms to generate secure cryptographic keys. The statistical nature of speckle-based imaging creates an opportunity for these cryptographic key generators to arise. In the scope of this work, three different tokens were implemented as physically unclonable devices: tracing paper, plastic optical fiber and an organic-inorganic hybrid. These objects were subjected to a visible light laser stimulus and produced a speckle pattern which was then used to retrieve the cryptographic key associated to each of the materials. The methodology deployed in this work features the use of a Discrete Cosine Transform to enable a low-cost and semi-compact 128-bit key encryption channel. Furthermore, the authentication protocol required the analysis of multiple responses from different samples, establishing an optimal decision threshold level that maximized the robustness and minimized the fallibility of the system. The attained 128-bit encryption system performed, across all the samples, bellow the error probability detection limit of 10-12, showing its potential as a cryptographic key generator.