We consider private communication over bosonic Gaussian channels via the most general adaptive protocols based on local operations and two-way classical communication. These protocols include all the possible strategies allowed by quantum mechanics where two remote parties have local quantum computers but do not share prior quantum entanglement. In this context, Pirandola-Laurenza-Ottaviani-Banchi (PLOB) [Nat. Commun. 8, 15043 (2017)] established weak converse upper bounds for the secret key capacity of these channels. These bounds were computed by combining teleportation stretching, able to simplify any adaptive protocol into a block form, and the channel’s relative entropy of entanglement, so that data-processing properties allow one to write simple single-letter quantities. Here we discuss an extension of these bounds to repeater-assisted quantum communications. Then, using an energy-constrained version of the diamond norm and the Braunstein-Kimble teleportation protocol, we can rigorously show the strong converse property of the bounds discovered by PLOB. Our analysis provides a full mathematical justification of recent claims appeared in the literature.
Consider two bosonic modes which are prepared in one of two possible Gaussian states with the same local energy: either a tensor-product thermal state (with zero correlations) or a separable Gaussian state with maximal correlations (with both classical and quantum correlations, the latter being quantified by quantum discord). For the discrimination of these states, we compare the optimal joint coherent measurement with the best local measurement based on single-mode Gaussian detections. We show how the coherent measurement always strictly outperforms the local detection strategy for both single- and multi-copy discrimination. This means that using local Gaussian measurements (assisted by classical communication) is strictly suboptimal in detecting discord. A better performance may only be achieved by either using non Gaussian measurements (non linear optics) or coherent non-local measurements.
We consider the secret key capacity of the thermal loss channel, which is modeled by a beam splitter mixing an input signal mode with an environmental thermal mode. This capacity is the maximum value of secret bits that two remote parties can generate by means of the most general adaptive protocols assisted by unlimited and two-way classical communication. To date, only upper and lower bounds are known. The present work improves the lower bound by resorting to Gaussian protocols based on suitable trusted-noise detectors.
We consider two remote parties connected to a relay by two quantum channels. To generate a secret key, they transmit coherent states to the relay, where the states are subject to a continuous-variable (CV) Bell detection. We study the ideal case where Alice's channel is lossless, i.e., the relay is locally in her lab and the Bell detection is perfomed with unit efficiency. This configuration allows us to explore the optimal performances achievable by CV measurement-device-independent quantum key distribution. This corresponds to the limit of a trusted local relay, where the detection loss can be re-scaled. Our theoretical analysis is confirmed by an experimental simulation where 10<sup>-4</sup> secret bits per use can potentially be distributed at 170km assuming ideal reconciliation.
We consider two bosonic Gaussian channels whose thermal noise is strong enough to break bipartite entanglement. In this scenario, we discuss how the presence of separable correlations between the two channels is able to restore the broken entanglement. This reactivation occurs not only in a scheme of direct distribution, where a third party (Charlie) broadcasts entangled states to remote parties (Alice and Bob), but also in a configuration of indirect distribution which is based on entanglement swapping. In both schemes, the amount of entanglement remotely activated can be large enough to be distilled by one-way distillation protocols.
Recently, we have shown the advantages of two-way quantum communications in continuous variable quantum
cryptography. Thanks to this new approach, two honest users can achieve a non-trivial security enhancement as
long as the Gaussian interactions of an eavesdropper are independent and identical. In this work, we consider asymmetric strategies where the Gaussian interactions can be different and classically correlated. For several attacks of this kind, we prove that the enhancement of security still holds when the two-way protocols are used in direct reconciliation.