Towards Quantum Networks: Characterizing Raman Noise over Metropolitan-scale Fiber Network

Technical Summary: Characterizing Raman Noise over Metropolitan-scale Fiber Networks

Problem Statement
The coexistence of quantum and classical signals within existing optical fiber infrastructures is a critical prerequisite for the scalability of large-scale quantum networks. However, this coexistence faces significant challenges due to noise sources, particularly spontaneous Raman scattering (SpRS). When high-power classical signals propagate alongside weak quantum signals in the same fiber, SpRS can scatter photons from the classical band into the quantum channel band. This introduces ground noise that degrades the fidelity of entangled states and limits the performance of quantum protocols. While Raman noise has been extensively studied in controlled laboratory settings, there is a lack of empirical data regarding its behavior in real-world, deployed metropolitan fiber networks using commercial equipment. Furthermore, previous studies often relied on spatial separation (multi-core fibers) or temporal multiplexing to mitigate noise, rather than analyzing simultaneous co-propagation in standard single-mode fiber.

Methodology
The authors conducted an experimental characterization of Raman scattering over a 7 km metropolitan fiber loop belonging to the University of Naples Federico II's National Quantum Internet.it testbed. The setup involved a loop-back architecture connecting the Monte Sant'Angelo (MSA) campus to the Engineering Faculty (EF) campus.

• Signal Configuration: The study investigated the co-propagation of a classical signal in the O-band (1310 nm) and a quantum signal in the C-band (1525–1565 nm). This configuration was chosen because the C-band offers the lowest attenuation for quantum transmission, while the O-band is standard for classical short-to-medium range communications.
• Sources: Two types of classical sources were utilized to compare controlled versus real-world conditions:
  1. A narrowband laser source (1310 nm, ~-4.6 dBm).
  2. A commercial SFP transmitter (1310 nm, ~-0.8 dBm) to simulate realistic deployment scenarios.
• Measurement Setup: The classical O-band signal was filtered and launched into the fiber loop. Raman-scattered photons generated in the C-band were spectrally selected using a tunable narrowband filter (25 GHz bandwidth, 500 pm tuning steps) and detected by a superconducting nanowire single-photon detector (SNSPD) with ~80% system detection efficiency.
• Loss Characterization: Insertion losses were measured for the fiber segments (SMF-28 and SMR fibers) to account for attenuation differences between the 1310 nm and 1550 nm bands.
• Comparison: Results were normalized based on source power and fiber length, and compared against reference measurements taken in a noise-free laboratory environment using a 5 km fiber spool.

Key Results

• Consistency with Laboratory Models: After normalizing for power and length, the Raman spectral profiles measured in the metropolitan loop showed strong agreement with those obtained in controlled laboratory conditions. This confirms that theoretical models derived from lab experiments accurately predict the dominant Raman scattering mechanisms in real-world deployments.
• Localized Spectral Anomalies: Unlike the smooth spectral trends observed in the lab, the metropolitan fiber measurements revealed distinct, localized spectral peaks (increasing photon counts by up to an order of magnitude) at specific wavelengths. These peaks were also present in background measurements (no launched signal), indicating they originate from imperfections or damaged sections within the deployed fiber loop rather than the Raman scattering process itself. The authors hypothesize these may be linked to inter-fiber crosstalk within the shared cable jacket.
• Dominance of Raman Noise: The study demonstrates that Raman-induced noise constitutes a dominant contribution to the quantum signal-to-noise ratio (SNR) in realistic deployments, often exceeding background and detector noise.
• Channel Allocation: The data allows for the identification of specific spectral regions within the C-band that are less affected by Raman noise, providing a basis for optimal channel selection.

Significance and Claims
The paper claims to be the first experimental demonstration of quantum networking where the quantum signal is allocated to the C-band while classical propagation occurs in the O-band, contrasting with previous reverse configurations. The primary significance of this work lies in bridging the gap between controlled laboratory research and deployed telecommunication environments.

The authors assert that their findings offer practical guidelines for the deployment of quantum communication systems over existing fiber networks. Specifically, they conclude that:

1. Careful selection of operating frequencies is required when co-propagating C-band quantum and O-band classical signals, as Raman scattering significantly impacts transmission quality.
2. Standard SNR definitions for quantum systems must be revised to explicitly account for Raman-induced noise in addition to conventional background noise.
3. Identifying spectral regions with minimal Raman noise is essential for allocating robust quantum channels in metropolitan-scale networks.

The work supports the development of robust and scalable quantum infrastructures by validating that while Raman behavior is predictable, real-world fiber imperfections introduce unique spectral anomalies that must be considered in network planning.