Generic methodology to create true randomness in integrated circuits.

The focus of this project is the study and design of sources of randomness in integrated circuits. True randomness is not the result of a deterministic process and thus cannot be computed but instead relies on uncontrollable physical effects in semiconductor circuits and processing. Traditionally this is called noise and in most designs it is an unwanted effect. We distinguish two types of noise. First there is the electrical noise caused by small current and voltage fluctuations such as thermal noise. It varies over time during operation of the circuit and it is used as entropy source for True Random Number Generators (TRNGs). Secondly, random process variations in deep submicron technologies result in variations fixed over time (or slowly varying due to ageing or other effects). It can be used as chip unique fingerprints, the so-called Physically Unclonable Functions (PUFs). In analog and digital circuit design classes we learned different techniques to cope with the unwanted effects of the different noise sources. In this project, we aim at investigating novel circuits, associated models and generic design techniques to precisely measure or to amplify the noise sources for security purposes.