Apparaît dans la collection : 2016 - T1 - WS5 - Secrecy and privacy theme

Unlike many other fields in computer science, randomness is essential for cryptography: secrets must have uncertainty to the attacker, and many cryptographic algorithms must be randomized (e. g. , two stateless encryptions of the same message must look different). Traditionally, one assumes the existence of perfect randomness. However, this assumption is often unrealistic. In this talk I will survey what is know about basing cryptography of various (realistic) sources of randomness. We will ask the following questions: 1) Does Cryptography need nearly perfect (“extractable”) sources of randomness, or is entropy sufficient? 2) What if the secret key is imperfect but “local” (or public) perfect randomness is available? As we will see, the answer to the first question is largely negative, while the second questions leads to many positive answers, some of which found many applications beyond cryptography.