Trapped Ion Quantum Networking and Telecommunications Coexisting on One Fiber

AI summary

Overview

Research area: Quantum networking with trapped-ion quantum memories, combined with classical optical telecommunications (coexistence of quantum and classical signals on shared deployed fiber).

Technical level: Intermediate to Advanced. The core idea is accessible, but full appreciation requires familiarity with ion-photon entanglement, polarization encoding, wavelength-division multiplexing, and coherent optical communications.

Scope: A demonstration of ion-photon entanglement distributed over a deployed 2.8 km fiber loop that simultaneously carries 400 Gbps Ethernet and 5G radio-over-fiber traffic, with all control and polarization-stabilization signaling on the same fiber.

What This Paper Is About

Long-distance, memory-based quantum networking has so far relied on dark fibers, because ordinary telecom traffic creates detector background noise that can swamp single-photon signals. That restriction excludes most fibers already installed in the field. This paper shows experimentally that entanglement between a trapped ⁸⁸Sr⁺ ion and a 1092 nm photon can be distributed over a deployed fiber loop while Ethernet and 5G traffic and all necessary control signals travel on that same fiber.

Key Contributions

1. First demonstration of memory-photon entanglement coexisting with live telecom traffic on one deployed fiber. Ion-photon entanglement from a ⁸⁸Sr⁺ ion is distributed over a deployed 2.8 km fiber loop while Ethernet and 5G signals traverse it, with fidelity maintained over more than 12 hours of operation.

2. A wavelength strategy that avoids frequency conversion. The 1092 nm transition lies outside the spontaneous-Raman-scattering noise band associated with telecom signals, so photons can be multiplexed with telecom traffic without conversion. The tradeoff is higher fiber loss (0.7 dB/km at 1092 nm versus 0.18 dB/km at the optimal telecom wavelength), which the authors characterize as giving a usable range of several kilometers.

3. Complete co-propagation of classical control and stabilization on the same fiber. All distributed control signaling runs between 1470 nm and 1610 nm using TTL-over-fiber modules and coarse wavelength-division multiplexing; polarization drift is sensed at 1530 nm during entanglement distribution and corrected at 1092 nm using a piezoelectric fiber squeezer.

4. Quantified classical link performance during quantum operation. The 400 Gbps Ethernet link at 1549.32 nm remains error-free after forward error correction, and the 5G 16-QAM signal at 1542.93 nm achieves an average EVM of approximately 9%, below the 3GPP limit of 12.5%.

Main Findings

• Entanglement fidelity: Averaged over 13 hours, 46 minutes, and 9 seconds of operation, fidelity bounds were 92.4(4)% ≤ F ≤ 96.1(4)%. Uncertainties were estimated by nonparametric bootstrapping with replacement, with confidence intervals incorporating 68% of bootstrapped outcomes. Fidelity was computed on a rolling basis from ZZ and XX basis measurements.

• Entanglement rate: 32,500 photon detection events were recorded, corresponding to 39.3 per minute. The authors attribute this rate to the 5.6% branching ratio of the 1092 nm transition, the finite numerical aperture of the collection optics (0.6), and photon detection wait time.

• Timing margin: The total delay between ion excitation and reception of SNSPD clicks back at the transmitter node is 28.335 µs, well below the ion qubit decoherence time of >700 µs. The detection window is 15 ns, and the loop restarts after 50 unsuccessful attempts.

• Polarization drift correction: Fifteen corrections were executed during the experiment; three were triggered by real fiber drifts and the remainder were scheduled hourly corrections. Average correction duration was 24.7 seconds, total time in correction mode was 6 minutes and 11 seconds, and entanglement distribution occupied 99.3% of experiment time (13 hours, 39 minutes, and 58 seconds).

• Drift trigger threshold: A trigger threshold of 1.5° in polarization drift distance was used, chosen to be well below the level at which significant infidelity arises while above polarimeter measurement noise.

• Background suppression: Telecom signals contributed a total optical power of 1.3 mW on the fiber (measured just before the receiver node's first WDM), and none were shuttered during photon detection windows. Three filtering layers—the fused fiber WDM (>10 dB attenuation), custom interference filters with optical density >10, and spatial separation from chromatic dispersion in the Wollaston prism—reduced telecom-induced background counts to dark count level.

• Classical link quality: The Ethernet transmitter used a Wistron Galileo Flex-T platform with Lumentum CFP2-DCO coherent pluggable modules running the NEC Network Operating System, encoding 16-QAM at 1549.32 nm with two polarization channels. The pre-FEC bit error rate remained low enough that forward error correction reduced residual BER to zero. The 5G link used a USRP B210 and an Optilab LT-12-E-M 12 GHz electro-absorption modulator, with reception on an Optilab PR-12-B-M 12 GHz photodiode receiver and a second USRP B210.

• Fiber loss budget: The total loss of the fiber component chain before the polarization state analyzer was 10.6 +0.3/−0.2 dB, including the transmitter MEMS switch (2.58 dB), quantum transmitter WDM (0.74 dB), 2.8 km fiber loop (3.32 dB), fiber squeezer (1.74 dB), quantum receiver WDM (0.45 dB), and receiver MEMS switch (2.60 dB).

• Fiber stability characteristics: A 24-hour Stokes-parameter measurement of a classical 1092 nm beam showed slow drift from day/night temperature change plus faster noise on the minute timescale. The fast noise did not appear on 1550 nm light, indicating wavelength dependence, and APC fiber tips did not eliminate it. The authors attribute it to the multi-mode nature of 1092 nm light in SMF-28, since 1092 nm lies below the single-mode cutoff and permits both LP01 and LP11 modes.

• Additional error sources: Magnetic field variations in the lab caused qubit splitting instability on the scale of 1.5 kHz, imprinting on ion X-basis measurements and contributing to small fidelity fluctuations.

Methodology in Plain English

The experiment used four nodes. A quantum transmitter and a quantum receiver sit in the Duke Quantum Center, linked by a single deployed 2.8 km fiber loop running underground. A telecom transmitter and receiver on Duke's West Campus connect to the quantum nodes through deployed 6.2 km fibers, so Ethernet and 5G traffic enters the quantum fiber loop, travels it, and is demultiplexed at the far end for decoding.

At the quantum transmitter, a single ⁸⁸Sr⁺ ion is held in a four-rod RF quadrupole trap, with a 4.3 G magnetic field defining the quantization axis. Exciting the ion to |5P₁/₂, −1/2⟩ leads to decay along the 1092 nm transition to 4D₃/₂ with 5.6% branching ratio. Light collected along the magnetic field carries σ⁺ and σ⁻ components that are mapped to H and V polarizations in fiber, producing the state (√3/2)|0⟩|H⟩ + (1/2)|1⟩|V⟩.

Photons pass through a MEMS switch, are merged with telecom-band light (1470–1610 nm) in a fused fiber WDM, traverse the deployed loop, and are separated again at the receiver. A polarization state analyzer (quarter, half, quarter waveplates, a Wollaston prism, and two SNSPDs) measures the photon, while the ion is read out on any detection event. Readout uses 1004 nm Raman beams for basis rotation (0 to 76.3 µs), shelving with 1004 nm σ⁻ and 408 nm σ⁺ beams for 180 µs, and fluorescence detection with 422 nm and 1092 nm beams for 13 ms.

Successful detection is signaled back to the transmitter by converting SNSPD TTL pulses to 1590 nm and 1610 nm optical pulses sent over the same fiber. Drift is monitored continuously by 1530 nm sensing beams read on a receiver polarimeter; when drift exceeds threshold, both nodes switch into a correction mode in which 1092 nm H- and D-polarized beams are alternated every 1.5 seconds and a gradient-descent algorithm drives a four-channel piezoelectric fiber squeezer until both polarizations return to target.

Why This Matters

Impact on research: The work removes the assumption that memory-based quantum links require dedicated dark fiber, showing that the relevant detector background can be pushed to dark-count level through wavelength choice and filtering. It also demonstrates that the full stack of classical support functions—heralding, synchronization, polarization sensing, and feedback—can share the quantum fiber rather than requiring separate infrastructure.

Real-world applications:

• City-scale quantum links between buildings using existing municipal fiber that already carries Ethernet and 5G.
• Quantum-secured or entanglement-based communication between data centers connected by coherent 400 Gbps interconnects.
• 5G fronthaul deployments, where analog radio-over-fiber links could share fiber with quantum channels.
• Fiber-sensing and infrastructure-monitoring services co-existing with quantum channels on the same strand.

Industry relevance: The result suggests network operators could offer quantum link capacity on lit fiber rather than reserving dark fiber, using standard components such as fused fiber WDMs, MEMS switches, interference filters, and commercial TTL-over-fiber modules. The authors note that significant additional classical bandwidth remains available for multiplexing more traffic before infidelity arises.

Future Directions

• Raising the entanglement rate. The 39.3 per minute rate is limited by the 5.6% branching ratio, the finite numerical aperture, and photon detection wait time; cavity integration and ion multiplexing are proposed to boost it.
• Handling less stable fibers. The deployed loop sits in a stable underground tunnel. Aerial fibers can present polarization perturbations on faster timescales and greater strength, which the current correction scheme would handle at the cost of reduced quantum link uptime. Time-bin encoding is proposed as an alternative, and the authors note that the WDMs, MEMS switches, and filters used are agnostic to photon basis.
• Extending coexistence to more telecom traffic. Since background counts reached dark-count level, the paper argues there is room to add more classical channels before fidelity degrades.
• Stabilizing remaining error sources. Magnetic field stabilization techniques are suggested to reduce the 1.5 kHz qubit splitting instability that imprints on X-basis measurements, and the non-unitary polarization fluctuations from multi-mode propagation at 1092 nm remain a fidelity contributor.

Target Audience

Quantum networking researchers and experimentalists working on trapped-ion or memory-based entanglement distribution; optical communications engineers interested in quantum-classical coexistence and wavelength planning; network operators and infrastructure planners evaluating whether quantum links can run on lit fiber; and graduate students in quantum information or photonics who want a concrete system-level example of integrating quantum hardware with deployed telecom infrastructure.

Authors’ abstract

Research into long-distance quantum memory-based networking to date has exclusively used dark fibers. This avoids the detector background from telecommunications (telecom) traffic, but as a result excludes many fibers deployed in the field. If memory-photon entanglement and telecom signals coexist on one fiber, the entire classical fiber infrastructure becomes available for quantum links. We present the first experimental demonstration of such coexistence. Ion-photon entanglement using 1092 nm photons emitted by a Strontium-88 ion is distributed over a deployed 2.8 km fiber loop which carries Ethernet and 5G traffic. All classical control signals required to coordinate the quantum transmitter and receiver systems co-propagate on the same fiber. These include fiber sensing for polarization stabilization. Our results demonstrate that memory-based quantum networks can be realized on active classical network infrastructure.