Capacity and SKR tradeoff in coexisting classical and CV-QKD metropolitan-reach optical links

Cagla Ozkan; Chigo Okonkwo; Cristian Antonelli; Joao dos Reis Frazao; Kadir Gumus; Lucas Alves Zischler

REVIEW 4 major objections 2 minor 12 references

Capacity and SKR tradeoff in coexisting classical and CV-QKD metropolitan-reach optical links

T0 review · 4 major / 2 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that in metropolitan DWDM, placing a CV-QKD channel at the band edge with a 100–150 GHz guardband yields a 108% secret-key-rate improvement at −1.5 dBm/ch while incurring only 3.4% classical capacity loss, versus 6.8% for

desk verdict A useful, clearly written guardband design study for metropolitan CV-QKD/classical DWDM coexistence, but the headline numbers all come from an unvalidated cited model, so the quantitative claims are not yet secured. read the letter →

arxiv 2512.14408 v2 pith:QZWQ6LZA submitted 2025-12-16 quant-ph

Cagla Ozkan , Lucas Alves Zischler , Kadir Gumus , Joao dos Reis Frazao , Cristian Antonelli , Chigo Okonkwo This is my paper · ORCID

classification quant-ph

keywords CV-QKDquantum-classicalcoexistenceDWDMguardbandoptimizationfour-wavemixingspontaneousRamanscatteringsecretkeyratemetropolitanopticalnetwork

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper aims to establish a design rule for letting weak continuous-variable quantum key distribution signals share a metropolitan dense wavelength-division multiplexing fiber with strong classical traffic. It argues that the dominant noise source switches with launch power: spontaneous Raman scattering dominates at low power, while four-wave mixing dominates as per-channel power rises above roughly −5 to 0 dBm. Because FWM falls off sharply with frequency separation but Raman does not, guardbands only help in the FWM-dominated regime. The paper's central quantitative claim is that band-edge placement with a 100–150 GHz guardband recovers 108% more secret key rate than no guardband at −1.5 dBm/ch, at half the capacity penalty of a symmetric band-center guardband. If correct, this gives network operators a simple, parameter-based rule for coexisting quantum and classical channels without expensive iterative simulations.

What carries the argument

The load-bearing object is a coupled power-evolution equation (their eq. 3) that accumulates interference power in the quantum channel from SpRS and from degenerate and non-degenerate FWM along the fiber, weighting each product by its phase-matching efficiency. This model converts fiber parameters (loss, dispersion β₂, nonlinearity γ, Raman gain) directly into excess noise and thus into secret-key rate via the SKR formula. Its work is to show that FWM's cubic power scaling and sharp phase-matching frequency dependence produce the power-regime transition that makes guardband size a decisive, regime-dependent knob.

What would settle it

Measure excess noise or secret-key rate on a 10 km metropolitan fiber with an 88-channel DWDM load at −1.5 dBm/ch, comparing no guardband to a 150 GHz guardband around a band-edge quantum channel. If the observed SKR improvement is not near 108%, or if counter-propagation improves more than co-propagation, the model's phase-matching/FWM assumptions are wrong.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is a power-regime-dependent guardband strategy for metropolitan CV-QKD/classical coexistence. Using a cited coupled-power model that tracks four-wave mixing and spontaneous Raman scattering along the fiber, the authors sweep guardband size, launch power, and distance on an 88-channel 50 GHz DWDM grid. They find that at high per-channel power (0.5 dBm/ch), FWM is so strong that no guardband yields zero SKR, and a three-channel (150 GHz) guardband restores 38 Mbit/s at 10 km; at low power (−4.5 dBm/ch), Raman dominates and guardbands add nothing; in between (−1.5 dBm/ch), widening the guardband from 0 to 3 channels raises SKR by 108%. They also show that

Load-bearing premise

The quantitative results all flow from a cited coupled-power coexistence model that the paper neither re-derives nor validates experimentally; if that model misestimates four-wave mixing or Raman noise, the 108% gain, the 38 Mbit/s high-power rate, and the 3.4% capacity-loss figure would all move.

Editorial extensions

If this is right

At per-channel powers around −1.5 dBm, co-propagating CV-QKD and classical channels can more than double key rate by reserving a 100–150 GHz guardband, at a cost of 3–4% of the 88-channel capacity.
At high loads (≈0.5 dBm/ch), coexistence is not possible without a guardband; a three-channel guardband is the minimum that allows any nonzero secure key at metropolitan reach.
At low loads (−4.5 dBm/ch), guardbands are pure capacity waste: SKR stays near 195–205 Mbit/s regardless of guardband size.
Counter-propagating quantum-classical traffic suppresses FWM phase matching, so guardbands yield only 2.6% SKR improvement there, versus 108% for co-propagation.
The FWM-vs-SpRS crossover at −5 to 0 dBm per channel gives a design rule: guardband allocation should be adaptive rather than fixed.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the model's FWM phase-matching term is correct, the optimal guardband at a given power should shrink as total channel count or dispersion changes; one can test this by sweeping loading fraction on the same fiber, since FWM products scale with the number of interacting channel triplets.
The 108% gain at −1.5 dBm/ch is computed for a homodyne CV-QKD with model parameters; extending to LLO or high-rate systems with different reconciliation efficiencies might shift the crossover power, so the design rule likely needs recalibration per hardware.
The capacity-loss proxy ΔC=N_GB/88 treats dropped channels as lost capacity, but in a flexible-grid network those guardband slots could still carry lower-rate or non-interfering services; the true economic tradeoff may be more favorable than 6.8%.
A direct implication the authors leave implicit: at high power, the three-channel guardband enables not just operation but a specific SKR (38 Mbit/s), which may be inadequate for continuous key generation under realistic overheads; linking SKR to practical key rates including finite-size effects is a natural next step.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The one thing to know: this paper is a clean parameter sweep that turns a known coexistence model into a practical design rule, and the rule itself—band-edge placement with 100–150 GHz unilateral guardbands, adapted to the FWM/SpRS power regime—is plausible and well explained. The soft spot is equally clear: every quantitative headline, including the 108% SKR gain and the 3.4% vs 6.8% capacity-loss comparison, is a direct output of Eq. (3), which is taken from ref. [7] rather than derived, benchmarked, or experimentally validated here. That makes the paper a useful engineering extrapolation, not a demonstrated result.

What the paper does well: it separates the two interference mechanisms cleanly, explains why FWM dominates at high power with cubic scaling and strong frequency dependence while SpRS is broadband and guardband-insensitive, and then translates that physics into an explicit, easy-to-apply recommendation: at per-channel powers above about −1.5 dBm, leave 2–3 channels of guardband at the band edge; below about −4.5 dBm, guardbands cost capacity without buying SKR. The band-edge versus band-center capacity-loss asymmetry (half the penalty for the same SKR gain) is a sensible and likely robust insight. The presentation is straightforward and the equations shown are standard.

The soft spots, in proportion: first, and most important, there is no independent validation. The paper cites ref. [8] for split-step Fourier simulations but does not compare against that method, and it provides no code, data, or error bars. A few dB of error in predicted interference power would shift the FWM/SpRS crossover and nonlinearly change the reported improvement percentages. Second, the capacity-loss proxy ΔC = N_GB/88 assumes dropped channels are pure loss; it ignores the option of reusing that spectrum at reduced rate, which would change the tradeoff numbers. Third, the abstract and conclusion use “demonstrate” and “validate” for what is actually a sweep over a cited model; that language overstates the evidence. These are addressable weaknesses rather than fatal flaws, and the paper does honestly attribute the model to ref. [7], including author overlap.

Bottom line: if you work on QKD deployment or spectrum management, this is worth reading and citing with a caveat. It deserves a serious referee—the qualitative design rule is solid enough to warrant engagement—but I would push for release of the simulation code and either a direct validation against split-step or a sensitivity analysis before the quantitative claims are taken as established.

Referee Report

4 major / 2 minor

Summary. The paper proposes an analytical design rule for placing a CV-QKD quantum channel alongside classical DWDM channels in metropolitan links. Using the coupled-power coexistence model of ref. [7] (Eq. 3), the authors sweep guardband size (0–10 channels), per-channel launch power (−4.5 to 0.5 dBm/ch), and distance (0–30 km) on an 88-channel, 50 GHz grid. They report that placing the quantum channel at the band edge with 2–3 channel (100–150 GHz) guardbands yields a 108% SKR improvement at −1.5 dBm/ch relative to no guardband, enables 38 Mbit/s at 0.5 dBm/ch where SKR is otherwise zero, and incurs 3.4% classical capacity loss versus 6.8% for band-center placement. The paper attributes these results to a power-regime transition from SpRS-dominated to FWM-dominated interference, and presents Fig. 1 as support. No code, data, or independent validation of the model is provided.

Significance. If the underlying model of ref. [7] is accurate, the paper offers a practically useful design guideline: place the quantum channel at the band edge and adapt the guardband to the launch power regime, thereby suppressing FWM noise without unnecessarily sacrificing classical spectrum. The systematic parameter sweep and the clear separation of FWM and SpRS contributions are strengths. However, every quantitative headline—the 108% gain, the 38 Mbit/s high-power result, and the 3.4% versus 6.8% capacity-loss comparison—is a direct output of the cited model, which is not derived, experimentally validated, or benchmarked here. The qualitative design rule is plausible, but the numerical claims are not yet established by this manuscript alone.

major comments (4)

[Section 2, Eq. (3)] The central quantitative claims (108% SKR improvement, 38 Mbit/s at 0.5 dBm/ch, 3.4% versus 6.8% capacity loss) are all generated by the coexistence model of ref. [7], reproduced as Eq. (3). The present paper does not derive this model, does not compare it against split-step Fourier simulations (refs. [8,12]), and provides no experimental data. Since the FWM-to-SpRS crossover is power-law driven, a few dB error in the predicted interference power could shift the crossover and change the headline percentages nonlinearly. The authors should either provide an independent validation of Eq. (3) for the specific metropolitan scenario, e.g., against split-step Fourier simulations, or clearly state that the quantitative results are inherited from ref. [7] and supply a sensitivity analysis. Without this, the numerical conclusions are not reproducible from the manuscript.
[Section 3, Fig. 1(c) and capacity-loss metric] The capacity-loss comparison uses the proxy ΔC = N_GB/88 for band-edge and 2N_GB/88 for band-center, i.e., it counts only fully dropped channels. This ignores the possibility of using the guardband spectrum for lower-rate transmission, flexible-grid channels, or other services. Thus the specific claim that band-edge incurs 3.4% loss versus 6.8% for band-center is an artifact of this counting rule. The authors should justify this proxy or use a more realistic spectral-efficiency/capacity model that accounts for partial or adaptable use of the guardband.
[Section 3, Fig. 1(b) and text] The '108% SKR improvement' is based on a single intermediate power (−1.5 dBm/ch) and a single guardband step (0 to 3 channels). The claimed transition 'between −5 and 0 dBm/ch' is inferred from only three power values (−4.5, −1.5, 0.5 dBm/ch) and no uncertainty or sensitivity analysis is given. The absolute SKR values at −1.5 dBm/ch are not reported, and no error bars or parameter tolerances are provided. The authors should present more power steps, absolute SKR numbers, and a sensitivity analysis (e.g., over γ, β2, gR) to substantiate the 'power-regime-dependent' guardband rule.
[Section 2, Eq. (3) and Section 1] The FWM efficiency ρ_ihkl and SpRS efficiency η_ih are not defined in the manuscript; the reader cannot reproduce Eq. (3) or the numerical results without consulting ref. [7]. Because the paper claims to provide a framework that 'eliminates costly iterative simulations' and enables deployment decisions 'directly from fiber parameters,' the model should be specified sufficiently, or at least the key efficiency functions and their parameter dependencies should be summarized in an appendix. This is also a reproducibility concern for a modeling paper.

minor comments (2)

[Section 3, system parameters] The list of fiber parameters gives α=0.2 dB/km, while Eq. (2) uses T=e^{-αL}. If α is in dB/km, the transmittance should be 10^{-αL/10}; if α is the linear attenuation coefficient, the stated numerical value is misleading. Please clarify the units consistently.
[References and minor text] There are minor typographical issues, e.g., 'Y . Mao' in ref. [2] and the redundant 'local local oscillator (LLO)' in the introduction. The reference list should be checked for consistent formatting.

Circularity Check

1 steps flagged · score 4.0 of 10

Headline numbers are generated solely by Eq. (3) from the authors' own ref. [7], which is cited but not independently validated; the qualitative scaling rule is independently motivated.

self citation load bearing [Sec. 1 (Introduction) and Sec. 2, Eq. (3); Sec. 3 sweeps; ref. [7]]
"We employ the coexistence model [7] to perform parameter sweeps of guardband size, classical power, and distance. ... To assess the impact of distinctive nonlinear sources, we employ the comprehensive coexistence model from [7], which simultaneously evaluates FWM and SpRS interference through coupled power evolution equations along the fiber."

The headline quantities (108% SKR gain, 38 Mbit/s at 0.5 dBm/ch, 3.4% vs 6.8% capacity loss) are outputs of Eq. (3), which is quoted from ref. [7] rather than derived or independently validated here. Ref. [7] shares authors (L.A. Zischler, C. Antonelli) with the present paper, and Sec. 3's 'To validate these predictions' sweeps the same model rather than testing it against split-step Fourier simulations, experiments, or released code/data. Thus the quantitative predictions reduce to a model asserted via a self-citation; only the qualitative FWM-vs-SpRS scaling rule is independently motivated. This is load-bearing self-citation, not a fit or definitional identity, so the score is moderate.

full rationale

This is not a case of a fitted parameter being renamed as a prediction: the paper performs no fitting, and Eq. (3) is a physically motivated ODE model with parameters from the literature. There is also no definitional identity between input and output of the form X is defined in terms of Y. The circularity burden is specifically the load-bearing self-citation: all quantitative claims—the 108% SKR improvement, the 38 Mbit/s high-power reach, and the 3.4%-vs-6.8% capacity-loss comparison—are outputs of Eq. (3), taken from ref. [7], whose authors include two of the present authors. The paper provides no independent benchmark (split-step Fourier simulation, experiment, or code/data release) to show that Eq. (3) is accurate; the sentence 'To validate these predictions' merely introduces parameter sweeps of the same model. Under the supplied rubric, a cited result counts as independent support only if machine-checked, code-reproduced, or externally falsifiable outside fitted values; none of those conditions is met here. I therefore score 4 rather than 0-2. I do not score 6 because the reduction is not 'by construction' in the mathematical sense: Eq. (3) is a substantive physical model, not a restatement of the conclusions, and the qualitative design rule (FWM cubic vs SpRS linear, band-edge reduction of phase-matching) is independently supported by standard nonlinear fiber optics cited in Sec. 2. The weakness is missing external validation for the model, which is a load-bearing self-citation rather than a definitional circularity.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No parameters are fitted in this paper; all inputs are from cited literature. The main burden is the unvalidated self-cited model [7] that generates every quantitative result, plus several domain assumptions about capacity loss and transceiver parameters.

assumptions (6)

domain assumption The coexistence model of ref [7] (coupled power evolution eq. 3) accurately predicts FWM and SpRS interference for the 88-channel DWDM grid.
All SKR and capacity numbers are outputs of this model; the paper does not independently validate it, and two of its authors are among the present authors.
standard math The SKR formula SKR=βI_AB-χ_BE for Gaussian-modulated CV-QKD under collective attacks with reverse reconciliation is valid.
Taken from ref [9]; standard CV-QKD security result, not derived here.
domain assumption Excess noise from nonlinear interference is given by ξ_B=P_int/(T h f B_s).
From ref [4]; maps integrated interference power in the quantum band into shot-noise units.
domain assumption The chosen CV-QKD parameters (V_A=8, η_b=0.6, β=0.95, V_el=0.01) are representative of experimental systems.
Cited to ref [11]; results are not tested across a range of transceiver parameters.
domain assumption Classical capacity loss equals the fraction of DWDM channels removed: N_GB/88 at the band edge and 2N_GB/88 at band center.
Assumes each 50-GHz channel carries equal capacity and cannot be repurposed; may overstate loss if guardband channels could carry reduced-rate or filtered signals.
standard math FWM phase-matching decays as 1/(1+(ΔβL_eff)^2) with Δβ∝β2Δf^2 and SpRS is broadband (~40 THz) and linear in pump power.
From Agrawal's Nonlinear Fiber Optics; standard physics used to explain the power-regime transition.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Capacity and SKR tradeoff in coexisting classical and CV-QKD metropolitan-reach optical links." pith.science (2026). https://pith.science/paper/QZWQ6LZA

@misc{pith2026251214408,
  author       = {Pith},
  title        = {Pith review of: Capacity and SKR tradeoff in coexisting classical and CV-QKD metropolitan-reach optical links},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QZWQ6LZA}},
  note         = {Machine review of arXiv:2512.14408}
}

read the original abstract

We demonstrate power-regime-dependent guardband optimization for quantum-classical coexistence in metropolitan DWDM. Quantum channel at band-edge with 100-150 GHz guardbands achieves 108% SKR improvement at -1.5 dBm/ch, incurring 3.4% capacity loss versus 6.8% for band-center.

Figures

Figures reproduced from arXiv: 2512.14408 by the authors.

**Figure 1.** (a) separates FWM and SpRS contributions at −4.5 dBm/ch as a function of classical channel position across the 88-channel DWDM grid to identify dominant effects. FWM-only shows 15% lower SKR at band-center 0 2 4 6 8 10 100 101 102 20 40 60 80 0 50 100 150 FWM-Only SpRS-Only FWM+SpRS Band-Edge Band-Edge (a) DWDM Channel Number SKR at 10 km (Mbit/s) 108% gain (b) Guardband Size (Channels) SKR at 10 km (Mbit/s) 0 2 4 6… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 1 linked inside Pith

[7]

Accurate and Effective Model for Coexistence

L. A. Zischleret al., “Accurate and Effective Model for Coexistence...” arXiv preprint arXiv:2510.02807 (2025)

arXiv 2025
[8]

Continuous variable QKD in flexible optical

S. Moreoloet al., “Continuous variable QKD in flexible optical...” J. Opt. Commun. Netw.17, B71–B82 (2025)

2025
[1]

A quantum access network,

F. Berndet al., “A quantum access network,” Nature501, 69–72 (2013)

2013
[2]

Integrating quantum key distribution with classical communications in

Y . Maoet al., “Integrating quantum key distribution with classical communications in...” Opt. Express26, 6010 (2018)

2018
[3]

Coexistence of continuous-variable quantum key

A.A.E. Hajomeret al., “Coexistence of continuous-variable quantum key...” arXiv preprint arXiv:2502.17388 (2025)

arXiv 2025
[4]

Feasibility of quantum key distribution through a dense wavelength

B. Qiet al., “Feasibility of quantum key distribution through a dense wavelength...” New J. Phys.12, 103042 (2010)

2010
[5]

Coexistence of continuous variable QKD with intense DWDM

R. Kumaret al., “Coexistence of continuous variable QKD with intense DWDM...” New J. Phys.17, 043027 (2015)

2015
[6]

Impact of Four-Wave-Mixing Noise from Dense Wavelength-

S. Duet al., “Impact of Four-Wave-Mixing Noise from Dense Wavelength-...” Phys. Rev. Appl.14, 024013 (2020)

2020

Show all 12 references

[9]

Improved composable key rates for CV-QKD,

S. Pirandola and P. Papanastasiou, “Improved composable key rates for CV-QKD,” Phys. Rev. Res.6, 023321 (2024)

2024
[10]

G. P. Agrawal,Nonlinear Fiber Optics(Academic Press, 2007), 5th ed

2007
[11]

Co-propagation of Classical and Continuous-variable

J. dos Reis Fraz ˜ao, “Co-propagation of Classical and Continuous-variable...” OFC pp. Th1C–5 (2024)

2024
[12]

SDN-enabled CV-QKD and classical channels

M. Iqbalet al., “SDN-enabled CV-QKD and classical channels...” J. Opt. Commun. Netw.17, B28–B37 (2025)

2025

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [7]

Accurate and Effective Model for Coexistence

L. A. Zischleret al., “Accurate and Effective Model for Coexistence...” arXiv preprint arXiv:2510.02807 (2025)

arXiv 2025

[2] [8]

Continuous variable QKD in flexible optical

S. Moreoloet al., “Continuous variable QKD in flexible optical...” J. Opt. Commun. Netw.17, B71–B82 (2025)

2025

[3] [1]

A quantum access network,

F. Berndet al., “A quantum access network,” Nature501, 69–72 (2013)

2013

[4] [2]

Integrating quantum key distribution with classical communications in

Y . Maoet al., “Integrating quantum key distribution with classical communications in...” Opt. Express26, 6010 (2018)

2018

[5] [3]

Coexistence of continuous-variable quantum key

A.A.E. Hajomeret al., “Coexistence of continuous-variable quantum key...” arXiv preprint arXiv:2502.17388 (2025)

arXiv 2025

[6] [4]

Feasibility of quantum key distribution through a dense wavelength

B. Qiet al., “Feasibility of quantum key distribution through a dense wavelength...” New J. Phys.12, 103042 (2010)

2010

[7] [5]

Coexistence of continuous variable QKD with intense DWDM

R. Kumaret al., “Coexistence of continuous variable QKD with intense DWDM...” New J. Phys.17, 043027 (2015)

2015

[8] [6]

Impact of Four-Wave-Mixing Noise from Dense Wavelength-

S. Duet al., “Impact of Four-Wave-Mixing Noise from Dense Wavelength-...” Phys. Rev. Appl.14, 024013 (2020)

2020