🔬 optics

Fiber Link Stabilization with a Multicore Fiber Amplifier

This paper demonstrates that utilizing separate cores of a multicore erbium-doped fiber amplifier for noise cancellation enables ultrastable optical frequency transfer with fractional instability as low as $1.4\times10^{-18}$ over 40 km, establishing multicore fiber networks as a viable platform for future precision time and frequency distribution.

Original authors: Yifan Liu, Takuma Nakamura, Daniel J. Elson, Yuta Wakayama, Charles A. McLemore, Tetsuya Hayashi, Mikael Mazur, Nicolas Fontaine, Franklyn Quinlan, Nazanin Hoghooghi

Published 2026-02-26

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yifan Liu, Takuma Nakamura, Daniel J. Elson, Yuta Wakayama, Charles A. McLemore, Tetsuya Hayashi, Mikael Mazur, Nicolas Fontaine, Franklyn Quinlan, Nazanin Hoghooghi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to send a message across the country using a fiber optic cable. But this isn't just any message; it's a "perfect" signal, like a super-precise atomic clock ticking away. The problem is that the Earth is messy. The ground shifts, the temperature changes, and the wind blows. All of these things stretch and squeeze the fiber cable, causing the light signal to get out of sync. It's like trying to walk a tightrope while someone is shaking the rope up and down.

To fix this, scientists usually use a trick called noise cancellation. They send a signal out, bounce it back, and compare the two. If the signal got "wobbly" on the way out, they can calculate exactly how much and adjust the signal on the way back to cancel out the wobble.

However, there's a catch. Most fiber optic cables used for the internet are designed for one-way traffic. They have "one-way valves" (isolators) to prevent signals from bouncing back and causing chaos. To do noise cancellation, you need two-way traffic, which usually requires building custom, expensive equipment that doesn't exist in standard networks.

The Big Idea: The "Multi-Lane Highway"

This paper introduces a clever solution using a special type of cable called Multicore Fiber (MCF).

Think of a standard fiber optic cable as a single-lane road. You can only drive one way at a time.
Now, imagine a multicore fiber as a highway with 7 lanes all bundled inside the same protective tube.

Here is the magic: Because all 7 lanes are wrapped tightly together in the same tube, they experience the exact same bumps, temperature changes, and vibrations. If the road gets hot, all 7 lanes expand equally. If a truck drives by, all 7 lanes shake at the same time.

The Experiment: The "Magic Amplifier"

The researchers wanted to know: Can we use this 7-lane highway to send a super-stable signal, and can we amplify it (make it stronger) without ruining the perfect timing?

They built a special Multicore Amplifier (like a gas station for light) that has 4 separate pumps, one for each lane.

The Outgoing Trip: They sent their perfect signal down Lane 1.
The Return Trip: They took a tiny bit of that signal and sent it back down Lane 2.
The Comparison: Because Lane 1 and Lane 2 are neighbors in the same tube, they shook in perfect unison. When the scientists compared the two, the "noise" (the shaking) canceled out almost perfectly, leaving only the pure signal.

The Results: A Super-Stable Connection

They tested this in two ways:

Just the Amplifier: They tested the "gas station" alone. It was incredibly stable, like a metronome that never skips a beat.
The Long Road: They connected the amplifier to a 40-kilometer (25-mile) long spool of this 7-lane fiber.

The Outcome:
Even after traveling 40 kilometers and getting amplified, the signal was still incredibly stable.

The Analogy: Imagine you are trying to keep a pendulum swinging perfectly for 1,000 seconds. A normal clock might drift by a tiny fraction. This system was so stable that if it were a clock, it would only lose or gain one second every 300 million years.

Why Does This Matter?

Currently, to send these super-precise signals, scientists have to build custom, expensive, one-off networks. This is like building a private, dedicated railway just to move a single, delicate egg.

This paper proves that we can use existing, future-ready fiber networks (the 7-lane highways) to do this.

No more custom valves: We can use the lanes going in opposite directions to cancel out noise.
Scalable: We can use standard telecom equipment (amplifiers) to boost the signal over long distances.
The Future: This means that in the future, the same fiber cables that carry your Netflix stream and your video calls could also carry the most precise time and frequency signals in the world, helping us build better GPS, detect earthquakes, and synchronize global atomic clocks.

The One Small Hiccup

The researchers noted that the connection points where they switched from a single-lane cable to the 7-lane cable added a tiny bit of "jitter" (like a bumpy patch on the highway). If they can smooth out those connections in the future, the system will be even better.

In short: They turned a standard "one-way" internet problem into a "multi-lane" solution, proving that the future of the internet can also be the future of ultra-precise science.

1. Problem Statement

The global telecommunications infrastructure relies heavily on single-core (SC) optical fibers. While these networks are ideal for data transfer, repurposing them for ultrastable optical frequency transfer (e.g., for comparing optical atomic clocks or geodesy) presents significant challenges:

Unidirectional Nature: Telecom links typically use isolators to prevent back-reflections, making them unidirectional. Ultrastable frequency transfer requires bidirectional operation to measure and compensate for environmental phase noise (Doppler cancellation).
Amplifier Limitations: Existing bidirectional links using SC fibers require custom amplifiers. However, back-reflections in these setups force severe gain curtailment, limiting the link length and performance.
Multicore Fiber (MCF) Potential: Weakly-coupled MCFs offer a solution where multiple cores share the same cladding, leading to highly correlated phase fluctuations. This allows one core to carry the signal forward and another to return it, enabling noise cancellation while maintaining unidirectional propagation in each core.
The Gap: While MCF links have been demonstrated, it was unknown whether Multicore Erbium-Doped Fiber Amplifiers (MC-EDFAs) could be integrated into these links without destroying the critical phase correlation between cores required for high-fidelity frequency transfer.

2. Methodology

The authors designed an experiment to test the compatibility of a commercial MC-EDFA with a noise-canceled MCF link.

Hardware Setup:
- Amplifier: A 4-core, core-pumped EDFA (980 nm pumping) designed for telecom. Each core is pumped independently via side-polished couplers. The amplifier includes MCF isolators to ensure unidirectional light propagation within the device.
- Fiber Link: A 40 km spool of weakly-coupled 7-core fiber.
- Interface: Fan-in/Fan-out (FI/FO) devices were used to couple single-mode fibers (SC) to the MCF cores.
- Stabilization Scheme: A standard Doppler-cancellation loop was used. Light from a cavity-stabilized 1550 nm laser was sent through one core (Forward), amplified, and a portion was routed back through a second core (Return). The return signal was mixed with the local reference to generate an error signal, which corrected the phase via an Acousto-Optic Modulator (AOM).
Experimental Configurations:
1. Amplifier-Only Test: The MC-EDFA was stabilized alone to measure its intrinsic phase noise and core-to-core correlation.
2. Integrated Link Test: The 40 km 7-core MCF was added to the amplifier to simulate a long-haul link.
3. Control Tests: Measurements were taken without the amplifier (MCF only) and with non-common SC fibers (to isolate the noise contribution of the FI/FOs and patch cords).
Characterization:
- Pump Noise: The team quantified how 980 nm pump power fluctuations (which are uncorrelated between cores) translate to phase noise on the 1550 nm signal.
- Metrics: Performance was evaluated using Modified Allan Deviation (MDEV) to assess fractional frequency instability over averaging times ranging from 1 second to 10,000 seconds.

3. Key Contributions

First Demonstration of MC-EDFA in Stabilized Links: This is the first study to prove that separate cores of a multicore amplifier can maintain sufficient phase correlation to support ultrastable frequency transfer.
Quantification of Pump Noise Impact: The authors measured the conversion factor between pump power fluctuations and signal phase shifts ( $\approx 3 \times 10^{-3}$ rad/mW). They demonstrated that under standard telecom operating conditions, this noise source is negligible ( $< 10^{-20}$ instability) and does not require a common pump laser.
Identification of Bottlenecks: The study isolated the primary sources of instability in current MCF setups, identifying non-common Single-Core fibers (used in FI/FOs and patch cords) and inter-core correlation limits as the dominant noise sources, rather than the amplifier itself.

4. Key Results

Amplifier-Only Performance:
- The stabilized MC-EDFA alone achieved a fractional frequency instability of $5 \times 10^{-19}$ at 1000 s averaging time.
- At 10,000 s, instability reached $1.5 \times 10^{-19}$ .
- This confirms the amplifier itself introduces negligible noise compared to the fiber link.
Integrated 40 km Link Performance:
- When integrated with the 40 km 7-core fiber, the system achieved a fractional frequency instability of $1.4 \times 10^{-18}$ at 1000 s.
- The free-running (unstable) link had an instability of $7 \times 10^{-15}$ , demonstrating the effectiveness of the noise cancellation.
Noise Limitations:
- Short-term (1s – 300s): Stability is limited by non-common SC fibers (the ~1.5m to 2m of patch cords and FI/FO pigtails connecting the components).
- Long-term (>1000s): Stability is limited by the inter-core correlation of the 40 km MCF spool. The authors note that a deployed fiber (buried/undersea) would likely show even better correlation than the lab spool due to more uniform environmental exposure.
- Pump Noise: The uncorrelated pump power fluctuations were found to be well below the noise floor of the system.

5. Significance and Future Outlook

Infrastructure Compatibility: The results prove that essential telecom components (MC-EDFAs) do not pose a barrier to ultra-stable frequency transfer. This paves the way for integrating precision time and frequency distribution into future MCF communication networks.
Scalability: The demonstrated gain (4 dB net over 40 km) shows that MCF links can be extended with amplification, overcoming the loss limitations of passive fiber links.
Path Forward: The authors suggest that future performance can be significantly improved by:
1. Developing bidirectional MC-EDFAs where neighboring cores operate in opposite directions (eliminating the need for FI/FOs and reducing non-common fiber noise).
2. Integrating isolators directly into the MCF amplifier package for individual cores.
3. Minimizing non-common SC fiber lengths in the link architecture.

In conclusion, this work establishes MCF networks as a viable and promising platform for the next generation of global time and frequency distribution, bridging the gap between high-speed data communication and precision metrology.