device-independent quantum key distribution

DI-QKD derives security from observed input-output correlations rather than trusting a detailed model of the transmitters and detectors. A Bell-inequality violation is used to certify that the correlations cannot be reproduced by an ordinary local hidden-variable device.

No, it still needs assumptions such as secure laboratories, trusted randomness, authenticated classical communication, and no unintended information leakage. Device-independent means substantially fewer internal-device assumptions, not assumption-free cryptography.

Measurement-device-independent QKD removes detector side-channel attacks by allowing an untrusted measurement station, but it still models the state-preparation devices. DI-QKD seeks stronger certification from Bell correlations and is experimentally much more demanding.

If too many trials are lost, an adversarial device could selectively report results and imitate a Bell violation. DI-QKD therefore requires high-efficiency detection, low noise, and strict treatment of all trials, which makes long-distance practical systems difficult.