Abstract
The security guarantees of classical crypto-systems rely on computational conjectures, such as the hardness of prime factorisation or discrete logarithm calculation. The efficiency of Shor's algorithm for factoring and computing discrete logarithm makes some of these crypto-systems insecure once a quantum computer is available. I will first discuss in this talk a performance analysis of various quantum computing architectures to show what it actually takes to run Shor's algorithm.
In their seminal work, Bennett and Brassard on the one hand and Ekert on the other hand, proposed quantum cryptography protocols that are provably secure, i.e. the security does not rely on computational assumptions. Their quantum protocols allow two parties, Alice and Bob, who are connected by a quantum channel to share a secret random bit string -a key- that can subsequently be used to encrypt messages. In these quantum key distributions, the claim that an adversary, Eve, who may fully control the quantum channel, gains no information about the key relies on the assumption that Alice and Bob possess perfectly characterized devices to generate, manipulate and measure quantum states according to a precise recipe. This assumption is difficult to verify in practice. Numerous ingenious experiments successfully attack quantum key distribution protocols which did not fulfil perfectly this assumption. In the second part of my talk, I will discuss quantum key distribution protocols which are immune to these vulnerabilities, i.e. for which the security guarantees are device-independent. I will show in particular the first results of our efforts to provide the theoretical groundwork needed to make the first experimental demonstration of device-independent quantum key distribution.
