Decoy-state quantum key distribution over 227 km with a frequency-converted telecom single-photon source

TL;DR

The paper tackles the challenge of enabling secure QKD over long distances with imperfect single-photon sources by integrating a decoy-state QKD protocol with a frequency-converted quantum-dot SPS operating in telecom wavelengths. It introduces an updated security framework for sub-Poissonian sources, and demonstrates a two-decoy-state implementation that bounds vacuum and single-photon yields and phase error under finite statistics using Chernoff-type analyses. Experimentally, it showcases a frequency-converted InGaAs QD emitting at 1550 nm with $g^{(2)}(0)\approx$0.015–0.016 and reports positive secret-key rates at $\sim$227 km ($\approx$44 dB loss) under asymptotic conditions, with finite-key results confirming practical operation at realistic acquisition times. The work broadens the utility of SPSs in quantum networks, relaxing single-photon purity requirements and paving the way for robust, long-distance QKD in deployed fibre networks, including potential extensions to more advanced protocols such as MDI and twin-field QKD.

Abstract

We implement a decoy-state quantum key distribution scheme using a telecom C-band single-emitter source. The decoy states are created by varying the optical excitation of the quantum emitter to modulate the photon number distribution. We provide an analysis of our scheme based on existing security proofs, allowing the calculation of secret key rates including finite key effects. This enables us to demonstrate, with a realistic single-photon source, positive secret key rates using our scheme over 227 km of optical fiber, equivalent to a loss tolerance one order of magnitude greater than non-decoy schemes. This work broadens the scope of single-photon sources in future quantum networks by enabling long-distance QKD with realistic levels of single-photon purity.

Figures (3)

• Figure 1: QKD Experimental Setup. A Ti:Sapphire laser generates optical excitation pulses with a repetition rate of 80 MHz and the excitation pulse area is controlled by a VOA. The QD is resonantly excited producing single photons at 942 nm that are converted to 1550 nm in a PPLN waveguide using a CW seed laser at 2.4 $\mu$m. The 1550 nm photons are extracted using shortpass, longpass and bandpass spectral filters (SP2050, LP1400, BP1550). A half-wave plate (HWP) prepares polarisation-encoded photons before coupling into single mode (SMF28) fibre and sent to the receiver. A passive receiver performs BB84 measurements using an in-fibre 50:50 beamsplitter (FBS) with fibre polarisation controllers (FPC) and polarising beamsplitters (PBS) to project into X and Z bases, before detection with superconducting nanowire single-photon detectors (SNSPDs).
• Figure 2: Source characterisation. (a) The measured detection count rates (black data, left-hand axis) and $g^{(2)}(0)$ (orange data, right-hand axis) of the resonantly excited QD emission after QFC from 942 nm to 1550 nm as the excitation pulse area is varied from $[0, 2\pi]$. Solid lines are generated from a simplified two-level model, see Appendix \ref{['sec:qd-model']} for details. Error bars are estimated assuming Poissonian statistics. (b) Coincidence measurements for the signal state ($\delta^{(1)}$) with $g^{(2)}(0)$ of $0.0159\pm 0.0002$ and decoy state ($\delta^{(2)}$) with $g^{(2)}(0)$ of $0.0155\pm 0.0005$.
• Figure 3: Comparison of Non-Decoy and Decoy protocols. AKR and FKR ($10^{11}$ signals sent) as a function of channel loss for the Non-Decoy BB84 protocol (red) and Decoy protocol (green). Key rate estimates from the experimental data (filled green circles) follow our system model. AKR for a source with the same $\langle n \rangle$ as our source and $g^{(2)}(0)=0$ is shown to achieve positive key rates at only slightly greater distances than our source with the Decoy protocol. The unfilled green circle is the experimentally measured AKR when using a smaller gating window of 250 ps.