J. Eli Bourassa

We identify a time-dependent passive source side-channel in common measurement-device-independent quantum key distribution implementations that rely on Faraday mirrors for stable phase modulation. We model the time-dependence of the side channel and use this information in conjunction with a recently developed numerical security proof technique based on semidefinite programming to quantify the impact on the secure key rate of the protocol. We explore the sensitivity of security to the parameters of the side channel and the choice of model for the signal.

The security of measurement device-independent quantum key distribution (MDI QKD) relies on a thorough characterization of one's optical source output, especially any noise in the state preparation process. Here, we provide an extension of the loss-tolerant protocol [Phys. Rev. A 90, 052314 (2014)], a leading proof technique for analyzing the security of QKD, to MDI QKD protocols that employ mixed signal states. We first reframe the core of the proof technique, noting its generalization to treat d-dimensional signal encodings. Concentrating on the qubit signal state case, we find that the mixed states can be interpreted as providing Alice and Bob with a virtual shield system they can employ to reduce Eve's knowledge of the secret key. We then introduce a simple semidefinite programming method for optimizing the virtual twisting operations they can perform on the shield system to yield a higher key rate, along with an example calculation of fundamentally achievable key rates in the case of random polarization modulation error.