Skip to content
Quantum.world

Research

Inter-city 18.9 km free-space QKD over a dense urban corridor

Field-deploys decoy-state BB84 across 18.9 km of dense Barcelona with room-temperature detectors and no adaptive optics, a realistic urban free-space QKD datapoint.

Inter-city 18.9 km free-space QKD over a dense urban corridor
arXiv
2611.16774
Published
2026-11-12
Authors
Ignacio H. López Grande, Nicolas Linale, Giacomo Paganini, Achraf El Bouazzaoui, Adrián S. Blanca, Ilektra Karakosta-Amarantidou, Manu Canals i Codina, Adam Vallés, Andrea Gorni, Jorge Fuenzalida, Evelyn A. Ortega, Valerio Pruneri

AI summary

Overview

Research area: Quantum communications — specifically terrestrial free-space quantum key distribution (QKD) in a metropolitan environment.

Technical level: Intermediate. Readers should be comfortable with BB84, decoy states, QBER, and secret key rate, but the paper is written as a system-engineering and field-deployment report rather than a security-proof paper.

Scope: An experimental demonstration of a polarization-encoded decoy-state BB84 system operating over a single, direct 18.9 km free-space optical link between Castelldefels and Barcelona, entirely inside the optical-emission restriction zone of Barcelona–El Prat Airport.

What This Paper Is About

Long-distance free-space QKD has mostly been demonstrated over rural, lake, or mountaintop observatory paths, while urban links have generally been limited to a few kilometers, or to multi-link spans that add up to longer distances without being one continuous channel. Dense cities are harder: turbulence, strict pointing and tracking requirements, background light, and airport airspace rules that forbid high-power laser beacons all add complexity. The authors build and field-test a comparatively simple, aviation-compliant QKD system that sends single photons 18.9 km directly across a densely populated corridor, and they measure the channel and estimate the finite-size secret key rate.

Key Contributions

  1. A single direct 18.9 km urban QKD link. Unlike prior metropolitan results that sum two shorter interfering links (for example 19.2 km as 7.7 + 11.5 km, and 14.2 km as 7.1 + 7.1 km), this is one continuous horizontal channel crossing four municipalities with population densities from roughly 1.5×10³ to 1.3×10⁴ inhabitants/km².

  2. An aviation-compliant optical architecture. Both terminals and the whole path lie inside the Laser-beam Critical Flight Zone of Barcelona–El Prat Airport, where maximum permitted irradiance is 5 µW/cm². Pointing and tracking therefore use low-power, non-collimated LED beacons (633 and 500 nm) instead of high-power laser beacons, with angular accuracy on the order of one arcsecond at roughly 1 Hz update rate.

  3. A deployment-oriented transmitter plus independent synchronization. A fiber-coupled, single-emitter transmitter at 785 nm and 100 MHz repetition rate derives all BB84 polarization states and decoy intensities from one DFB laser, using a Sagnac-loop intensity modulator, PM-fiber polarization modulation, and an RFSoC FPGA with 16-bit DACs. Synchronization uses two GPS-disciplined rubidium clocks and time taggers, avoiding any extra optical synchronization channel.

  4. Room-temperature detection without adaptive optics. The receiver uses a 406 mm Ritchey–Chrétien telescope, a fast steering mirror for tip–tilt correction referenced to a low-power 633 nm beacon, and commercial silicon SPADs at approximately 60% detection efficiency, with overall receiver loss below 7 dB.

Main Findings

  • Channel attenuation: During the 40-minute field test the average channel attenuation was approximately 19 dB (abstract), with a mean inferred value of 19.7 dB from detected signal counts after correcting for polarization-analysis module insertion loss and detector efficiency. A continuous-wave 785 nm measurement taken before the session over 5 minutes gave 18.5 dB.
  • QBER: The detection-weighted mean QBER over the 40-minute test was 1.31%, with per-state values of 0.99% (D), 1.79% (A), 1.35% (R), and 1.07% (L). A 20 s detection-weighted moving average stayed stable for all polarization states.
  • Secret key rate: Using the finite-size analysis of Lim et al., with target blocks of n_key = 2×10⁶ sifted detections in the D/A key basis, ε_sec = 10⁻¹⁰, ε_cor = 10⁻¹⁵, and error-correction efficiency f_EC = 1.16, the average finite-size secret key rate was 11.17 kbit/s, with an average acquisition time of 116.4 s per block.
  • Detection gains: Q_signal = 0.97×10⁻³, Q_decoy = 5.67×10⁻⁴, and Q_vacuum = 1.40×10⁻⁵.
  • Transmitter calibration: Pulses had approximately 360 ps FWHM; the intensity stage achieved more than 30 dB extinction ratio over 200 voltage values spanning one V_π. Operating mean photon numbers were μ_s = 0.45 for signal and μ_d = 0.22 for decoy, with vacuum at maximum attenuation. The mean intrinsic QBER of the transmitter was 0.16%.
  • 15-hour laboratory stability: Under protocol-representative operation, mean photon numbers were 0.445, 0.222, and 5.23×10⁻⁴ photons per pulse for signal, decoy, and vacuum, and the QBER averaged (0.22 ± 0.02)%.
  • Fixed-pattern system test: At 1.5 photons per pulse over 380 s, the mean QBER was 0.82%, with D 0.99%, A 0.40%, R 0.91%, and L 1.00%. Reconstructed pulses were about 500 ps FWHM, motivating a 1 ns valid-count gating window.
  • Polarization analysis: The polarization-analysis module achieved extinction ratios above 23 dB in both bases, corresponding to a crosstalk contribution to QBER below about 0.5%.
  • Long-term CW link monitoring: Over more than 50 minutes, mean attenuation was 16.5 dB, and the 5-minute moving average varied only between 15.1 and 18.7 dB.
  • Reproducibility across campaigns: Over a three-month operating period spanning multiple campaigns, measured attenuation ranged from 16 to 22 dB and QBER stayed between 1% and 3%.
  • Deployment effort: Installation and alignment took approximately 5 working days in total across two stages, separated by a deliberate 12-month gap, with alignment performed at night.

Methodology in Plain English

The team split the deployment into two stages to reduce risk. First, they installed telescopes, motorized alt-azimuth mounts, and LED-based pointing and tracking at both sites, and characterized the optical channel with classical light. Only once the attenuation looked compatible with QKD did they add the tip–tilt stabilization, polarization-analysis module, and single-photon detectors at the receiver and connect the quantum transmitter at the transmitter site.

At the transmitter, a single laser produces short pulses, an intensity stage sets the signal, decoy, and vacuum levels, and a polarization stage sets each of the four BB84 states by controlling the relative phase between the two axes of a polarization-maintaining fiber. The prepared states are coupled into single-mode fiber and launched through a 75 mm collimation lens next to a 203 mm tracking telescope.

At the receiver, a 406 mm telescope collects both the quantum signal and a weak 633 nm reference beacon. A fast steering mirror, driven by the spot position of that beacon on a fast camera, corrects beam wander before the light reaches the polarization-analysis module, so no higher-order adaptive optics are needed. Timing is handled independently: each site has a rubidium clock disciplined by GPS, giving initial synchronization on the order of 10 ns, which post-processing cross-correlation refines enough to identify events within a 1 ns coincidence window.

Before going to the field, the transmitter was calibrated in the lab with fixed states, then run continuously for 15 hours in back-to-back configuration to confirm that one calibration holds. In the field, the team first sent a fixed D/A/R/L pattern at elevated photon number to isolate state-preparation and alignment errors, then ran the actual protocol emulation by repeatedly sending a 256-symbol pseudorandom sequence for 40 minutes.

Why This Matters

The result shows that useful QKD rates can be obtained across a real city without cryogenic detectors, without adaptive optics, and without high-power alignment lasers — the three things that most complicate urban deployment. It also shows that a fiber-coupled quantum payload can be dropped onto an existing free-space terminal without hardware modification, because synchronization is handled by GPS rather than an optical channel.

Real-world applications:

  • Hybrid fiber–free-space networks: The receiver at Collserola Tower, a telecom hub with direct fiber connectivity, illustrates how free-space links can bridge geographically separated fiber nodes through trusted nodes.
  • Airport-adjacent and emission-restricted sites: The low-power LED beacon design is directly relevant wherever laser beacons are regulated or prohibited.
  • Rapid-deployable or backup links: Free-space paths avoid the accidental-cut and deliberate-disruption vulnerabilities of dedicated fiber.
  • Testbed for other quantum protocols: The measured stability and polarization-transmission fidelity are compatible with protocols beyond decoy-state BB84.

Industry relevance: The architecture emphasizes modularity, commercial silicon SPADs, room-temperature operation, and a transmitter derived from a deployment-oriented design lineage — all signals aimed at practical integration rather than laboratory record-chasing.

Future Directions

  1. Full long-term channel characterization. The authors state that statistics over months and seasons, the correlation between residual pointing errors and QBER, and continuous unattended operation over days including daytime are still required.
  2. Automated polarization alignment. Transmitter–receiver polarization alignment was performed manually in this work; the authors note the same procedure can be automated, as demonstrated in prior work they cite.
  3. Improved receiver fiber coupling. The paper notes that fiber coupling loss "can be improved with careful alignment," which would raise the key rate.
  4. Simplified electronics. Because the polarization sweep used only 20% of the available DAC dynamic range, the authors suggest lower-resolution DACs could be used in a future integrated implementation with adjusted amplification gain.

Target Audience

Experimental quantum communication researchers and engineers building QKD hardware; free-space optical communications groups interested in urban propagation and pointing/tracking under emission constraints; network architects planning hybrid fiber–free-space quantum networks; and readers who want a concrete, numbers-heavy account of what a realistic metropolitan QKD deployment looks like rather than a record-distance claim.

Authors’ abstract

Terrestrial free-space quantum key distribution (QKD) provides a flexible complement to fiber-based networks, but its deployment over long horizontal paths in dense metropolitan environments remains challenging, as atmospheric turbulence, demanding pointing and tracking requirements, and operational constraints can substantially increase system complexity. We experimentally demonstrate the operation of a decoy-state BB84 system over an 18.9~km free-space link across a…

The full abstract on arXiv
Read the original paper