⚛️ quantum physics

Optimization of C-band quantum traffic coexisting with O-band classical traffic: preliminary results

This study experimentally investigates spontaneous Raman scattering noise in a co-propagating C-band quantum and O-band classical traffic setup using commercial transceivers, leading to a robust, source-independent model that accurately predicts noise levels to identify optimal C-band channels for quantum signal transmission in standard urban fiber infrastructures.

Original authors: Laura d'Avossa, Elena Montella, Marco Grillo, Angela Sara Cacciapuoti, Marcello Caleffi

Published 2026-07-30

📖 3 min read🧠 Deep dive

CC BY 4.0

Original authors: Laura d'Avossa, Elena Montella, Marco Grillo, Angela Sara Cacciapuoti, Marcello Caleffi

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 the internet as a massive, bustling highway system where data travels as pulses of light. For decades, this highway has been dominated by "classical" traffic—your emails, videos, and cat memes—zipping along at incredible speeds. But scientists are now building a new kind of vehicle for the future: the quantum car. These cars carry "entanglement," a spooky connection between particles that allows for unhackable communication and super-powerful computing. The problem? Quantum cars are incredibly fragile. If they share the road with the heavy, noisy trucks of classical traffic, the vibrations can knock the quantum passengers right out of their seats, destroying the delicate information they carry.

To make this work, we need to figure out exactly how to share the road without a crash. The key concept here is "noise." In the world of light, noise isn't just static on a radio; it's stray photons (tiny packets of light) bouncing off the fiber walls and crashing into our quantum signals. One major source of this noise is something called "Spontaneous Raman Scattering." Think of it like a game of billiards: when a fast-moving cue ball (the classical light) hits the table, it doesn't just stop; it transfers some energy to the felt, creating a ripple that spreads out in all directions. Some of these ripples end up right where the quantum ball is trying to roll. The big question is: where on the highway is the road smoothest? Which lane should the quantum cars take to avoid the worst of the ripples?

This paper dives into that exact question. The researchers set up a real-world experiment to see what happens when they run a loud, classical signal in one part of the light spectrum (the "O-band," around 1310 nm) and try to listen for a whisper-quiet quantum signal in another part (the "C-band," around 1550 nm) on the same fiber optic cable. Unlike previous studies that used fancy, quiet lasers or separated the signals in time or space, these scientists used standard, off-the-shelf equipment—the kind you might find in a real city network. They wanted to see the messy, real-world truth.

What they found is a map of the "noise landscape." By sending classical light through fiber cables of different lengths (up to 5 kilometers) and measuring the stray noise that leaked into the quantum band, they discovered that the noise isn't spread out evenly. It's bumpy. There is a specific "quiet zone" where the noise is at its lowest. Using their data, they built a mathematical model—a sort of noise prediction engine—that can tell you exactly how much noise to expect based on how powerful your classical signal is and how long the fiber cable is.

The most exciting part of their discovery is that this model works no matter what kind of light source you use. Whether they used a standard commercial laser or a more precise one, the model predicted the noise levels with high accuracy. They identified a "sweet spot" around 1535 nm (specifically channel 44) where the noise from the O-band traffic is at its minimum. This suggests that if we want to build a future internet where quantum and classical data share the same fiber, we should park our quantum signals in this quiet lane. The paper doesn't claim to have solved the entire problem of quantum networking, but it provides a robust, tested tool that engineers can use to design these shared networks, ensuring the quantum cars can drive safely alongside the classical trucks.

Technical Summary: Optimization of C-band Quantum Traffic Coexisting with O-band Classical Traffic

Problem Statement
The deployment of large-scale quantum networks requires leveraging existing telecommunications fiber infrastructure rather than dedicated "dark" fibers, which are impractical and prohibitively expensive. A critical challenge in this context is the coexistence of low-power quantum signals with high-power classical traffic within the same single-core, single-mode fiber. The primary impairment in such co-propagation scenarios is Spontaneous Raman Scattering (SpRS). In this process, photons from the classical signal are scattered across a broad spectrum, generating noise that can fall within the spectral band of the co-propagating quantum signal. While theoretical models for Raman scattering exist, they often fail to capture the fine spectral structure observed in standard fibers under realistic conditions. Furthermore, previous experimental studies have frequently relied on narrow-linewidth laboratory lasers or spatial/temporal multiplexing architectures, which do not reflect the constraints of deployed urban fiber infrastructures using commercial off-the-shelf components.

Methodology
This work experimentally investigates C-band SpRS noise generated by O-band classical traffic within a standard single-mode fiber environment. The study is conducted using the National Quantum Internet.it testbed at the University of Naples Federico II.

Experimental Setup: The researchers employed commercial SFP optical transceivers (specifically Finisar and Huawei modules) operating at 1310 nm (O-band) as the classical source. Unlike previous studies, no spatial or temporal multiplexing was used; signals co-propagated simultaneously.
Filtering and Isolation: To ensure signal purity, the setup utilized cascaded WDM filters to isolate the O-band source and suppress out-of-band components. At the receiver, a second WDM cascade isolated the C-band, followed by a narrowband tunable filter (25 GHz bandwidth) to scan ITU channels 15 through 63.
Detection: SpRS photons were detected using a Superconducting Nanowire Single-Photon Detector (SNSPD) operating at 2.2 K, achieving a detection efficiency of ~80% and a dark count rate of ~100 cps.
Variables: Experiments were conducted over fiber spools of varying lengths (100 m, 500 m, and 5 km) to simulate different urban scenarios. The study also compared results between commercial SFPs and a narrow-linewidth laser to assess source dependency.

Key Contributions and Modeling
The primary contribution of this work is the derivation of a compact, predictive model for SpRS noise that is independent of the specific optical source used.

The Model: The authors propose a model for the SpRS count rate ( $C_{SpRS}$ ) as a function of source power, fiber length, and wavelength:
$C_{SpRS}(S_{src}, \lambda, L) = S_{src} A(\lambda) \frac{e^{-\alpha_c L} - e^{-\alpha_o L}}{\alpha_o - \alpha_c}$
Here, $S_{src}$ is a source-dependent scale factor, and the attenuation term accounts for the effective interaction length based on O-band ( $\alpha_o$ ) and C-band ( $\alpha_c$ ) fiber losses.
Spectral Profile ( $A(\lambda)$ ): Unlike generalized theoretical models, the Raman scattering coefficient $A(\lambda)$ is modeled as a fourth-degree polynomial derived directly from experimental data. This approach captures the specific spectral variations and fine structures observed in the testbed.
Validation: The model was validated against experimental data from three distinct sources (two commercial SFPs and one laser) across multiple fiber lengths. The relative error between the model and experimental data was found to be low (e.g., 2.7% to 5.1% for 500 m and 5 km lengths), confirming the model's robustness and source-agnostic nature.

Results

Optimal Channel Identification: The experimental data and the derived model consistently identify a spectral minimum in SpRS noise around 1535 nm (ITU Channel 44). This suggests that Channel 44 is the most suitable C-band channel for quantum signal transmission when co-propagating with 1310 nm classical traffic.
Filtering Impact: The study demonstrated that cascading WDM components is essential for isolating the signal of interest. While increasing the number of WDMs introduces insertion loss, it significantly reduces count fluctuations and effectively suppresses residual out-of-band noise.
Source Independence: The model successfully predicted SpRS behavior for different hardware manufacturers (Finisar vs. Huawei) and different source types (commercial SFP vs. narrow-linewidth laser), proving that the underlying physical behavior is captured regardless of the specific emitter's internal architecture.
Length Dependence: The model accurately describes noise scaling with fiber length. While deviations were slightly more pronounced at very short lengths (100 m) where Raman counts are less significant compared to background noise, the global trend and the identification of the spectral minimum remained reliable.

Significance
The paper posits that these results pave the way for a parameter-robust description of Raman scattering applicable to diverse fiber-based systems. By providing a predictive model that does not require direct spectral measurements for every new configuration, the work offers a practical tool for designing and optimizing quantum-classical coexistence. Specifically, it enables the identification of optimal C-band channels for quantum allocation that are least affected by SpRS noise from O-band classical traffic. This is a fundamental step toward enabling the large-scale deployment of quantum networks over existing urban fiber infrastructures without the need for dedicated dark fibers.

More like this