Influence of the temperature on true random number generators

Today TRNGs are used in many different applications. The quality of their randomness is determined by these applications: for example those with security requirements need very good random numbers while simulations have fewer constraints on their properties. It is therefore necessary to investigate their robustness when under stress, being due to extreme conditions of utilization or deliberates attacks. Many TRNG designs exist and we decided to investigate two randomness sources and two post-processors that are commonly found in the literature. These TRNGs were implemented into a chip and put under test with variations of their temperature. The behavior of the randomness sources and the efficiency of the post-processors are evaluated thanks to several standard statistical tests presented in the literature.