Quantum cryptography with untrusted devices

Quantum mechanics allows us to accomplish certain cryptographic tasks that are impossible classically, such as generating a secret key between two participants using only a public channel. However, these quantum protocols usually assume that the participants are using devices that work exactly according to their specifications. In practice, such perfection is unfortunately very difficult to achieve. Even worse: the devices might be supplied by a malicious manufacturer that is itself trying to break the protocol. To mitigate this problem, protocols whose correctness can be checked only from the devices' observable behavior have been developed. However, the security of these "device independent" protocols is generally much harder to prove. In this talk, I will present a new method that allows us to bound the amount of randomness produced by an n-step quantum process which can be used to give nearly optimal security bounds for several device independent quantum protocols.

 

This is joint work with Rotem Arnon-Friedman, Omar Fawzi, Renato Renner and Thomas Vidick