Curator's Take
This article tackles a key bottleneck for free‑space continuous‑variable QKD by showing that simply using an average loss figure underestimates the true performance when single‑mode fiber receivers collect turbulent light. By deriving an effective coupling efficiency and exposing a protocol‑specific reference‑noise penalty in pilot‑assisted schemes, the work bridges realistic atmospheric modeling with practical key‑rate predictions for ground‑to‑satellite links. The findings suggest that GG02 can still be evaluated with a scalar coupling model, but more advanced transmitted‑reference protocols will need additional monitoring of signal‑pilot phase mismatches to maintain security margins.
— Mark Eatherly
Summary
We study free-space continuous-variable quantum key distribution (CV-QKD) with single-mode fiber (SMF) reception under atmospheric turbulence. The optical channel is modeled by split-step propagation through random phase screens, followed by finite-aperture collection and projection onto the guided receiving mode. We first examine the standard GG02 setting and ask which receiver-side observable is sufficient for effective key-rate prediction. We show that a mean-loss description is generally too optimistic, whereas a scalar effective law for the SMF coupling efficiency provides an accurate downstream Gaussian-channel description within the effective model considered here. We then extend the optical model to a pilot-assisted architecture in which the signal and pilot propagate through correlated but non-identical turbulent realizations generated by a frozen-flow construction. In this case, the signal coupling law alone is no longer sufficient: signal--pilot phase mismatch and loss of post-coupling coherence produce an additional protocol-dependent reference-noise penalty. The results distinguish two regimes: a scalar coupling description is largely adequate for GG02, while transmitted-reference architectures require an additional differential reference observable beyond the signal coupling statistics.