Robuuste Willekeurigheidsgeneratie op Ingebedde Apparaten

Cryptography plays a crucial role in securing communications and data in our modern society. High-quality randomness is essential for cryptographic systems, as it ensures security by being the only piece of information that a malicious party lacks to understand the entire system. Embedded devices, however, face significant challenges in generating this randomness due to their limited resources and minimal external input. As a result, these devices must rely heavily on on-chip generated randomness, making the design of TRNGs a critical area of research. This dissertation makes several theoretical contributions to the field of TRNGs. Firstly, it proposes a more accurate entropy model that applies to a broader range of noise types. Our findings suggest that more noise types than previously recognized have potential utility in TRNGs, although further research is necessary to determine their precise entropy contributions. Additionally, this work offers a refined time-based characterization of various noise types on an ASIC platform, revealing that under practical operating conditions, a different noise type dominates compared to earlier assumptions. On the practical side, this dissertation enhances TRNG design through increased control over key parameters, allowing designers to achieve optimal operating conditions with significantly less effort. The robustness of TRNGs against varying operating conditions has also been improved, making the output usable across a broader range of environments. Moreover, the research provides deeper insights into the trade-offs involved in performance optimization, facilitating a more streamlined design process compared to previous iterative approaches.