GPU-Accelerated Optimisation of Symmetric Bidirectional Ultrawideband Coherent Transmission Under Launch Power Constraints

Eric Sillekens; Mindaugas Jarmolovi\v{c}ius; Polina Bayvel; Robert I. Killey

Under an 18-dBm fibre power limit, reversing O-band direction in bidirectional OESCL-band coherent links raises aggregate capacity by up to 24.5 percent; unconstrained, the gain vanishes.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-15 02:18 UTC pith:XJJKDV2E

load-bearing objection Clear numerical finding of 24.5% capacity gain from O-band reversal under an 18 dBm power limit in bidirectional multi-band fibre, but the abstract alone leaves the result unanchored to any validation. the 4 major comments →

arxiv 2607.12921 v1 pith:XJJKDV2E submitted 2026-07-14 eess.SP

GPU-Accelerated Optimisation of Symmetric Bidirectional Ultrawideband Coherent Transmission Under Launch Power Constraints

Mindaugas Jarmolovi\v{c}ius , Eric Sillekens , Polina Bayvel , Robert I. Killey This is my paper

classification eess.SP

keywords bidirectional transmissionOESCL-bandcoherent optical communicationsRaman amplificationlaunch-power optimisationGPU accelerationfibre power constraintsmulti-band systems

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper asks whether the direction of the O-band relative to the E, S, C and L bands can be used as a free design choice to raise capacity when total launch power is limited. The authors couple a GPU-accelerated boundary-value Raman solver to a launch-power optimiser and evaluate three-span bidirectional OESCL coherent transmission. Without any power ceiling the co-propagating and counter-propagating O-band layouts deliver essentially the same capacity. Once total fibre power is capped at 18 dBm, however, reversing the O-band direction yields up to 24.5 percent higher aggregate capacity. The result matters because real multi-band systems are already power-constrained by eye-safety, amplifier and nonlinear-impairment limits; a purely directional change that recovers nearly a quarter of the lost capacity is therefore a practical lever.

Core claim

Under a total fibre launch-power limit of 18 dBm, three-span bidirectional OESCL-band coherent transmission with the O-band propagating in the reverse direction of the E, S, C and L bands increases aggregate capacity by as much as 24.5 percent relative to the co-propagating configuration; without the power constraint no capacity gain is observed.

What carries the argument

A GPU-accelerated boundary-value Raman solver that evaluates the multi-band power evolution under bidirectional launch, coupled to a launch-power optimiser that maximises aggregate capacity subject to the total fibre power constraint.

Load-bearing premise

That the Raman solver together with the nonlinear-interference and transceiver models used for capacity evaluation accurately capture the dominant impairments of real bidirectional multi-band fibre under the stated power limit.

What would settle it

A laboratory three-span OESCL bidirectional transmission experiment under an 18-dBm total power cap that measures whether the reversed-O-band layout delivers a capacity advantage consistent with the reported 24.5 percent figure.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

System designers facing tight total-power budgets can treat O-band direction as a free binary design variable that recovers capacity without extra spectrum or amplifiers.
Power-constrained multi-band links become more attractive once the reversed-O-band layout is adopted, because the same fibre and amplifiers support substantially higher throughput.
Unconstrained systems gain nothing from direction reversal, so the optimisation effort can be concentrated on power-limited regimes.
The same GPU-accelerated solver can be reused to re-optimise launch powers for other span counts or power ceilings without rewriting the Raman model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the capacity gain is driven by reduced Raman tilt and inter-band nonlinear interference under power sharing, similar directional benefits may appear when the S or E band is reversed instead of the O band.
The 18-dBm threshold suggests a practical operating point near eye-safety or amplifier-output limits; mapping the gain versus power ceiling would identify the regime where direction reversal is most valuable.
Because the gain vanishes without a power constraint, any future multi-band standard that already enforces total-power limits should explicitly list O-band direction as a configuration option.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

Clear numerical finding of 24.5% capacity gain from O-band reversal under an 18 dBm power limit in bidirectional multi-band fibre, but the abstract alone leaves the result unanchored to any validation.

read the letter

The one thing to know is that this abstract reports a concrete optimisation result: under an 18 dBm total fibre power limit, reversing the O-band direction in 3-span bidirectional OESCL coherent transmission yields up to 24.5% higher aggregate capacity than the co-propagating case; without the power constraint the gain disappears. That interaction of direction and power budget is the actual new number.

What they do well is combine a GPU-accelerated boundary-value Raman solver with launch-power optimisation under a realistic total-power constraint. Bidirectional multi-band Raman transfer and power-limited capacity are both known topics, but the specific optimised comparison and the zero-gain unconstrained baseline are useful for system designers thinking about OESCL links. The claim is non-circular on its face: capacity is an output of the search, not an input.

The soft spot is exactly what the stress-test flags and it is not minor. Everything rests on an unvalidated numerical stack (Raman BVP plus unspecified NLI and transceiver models). The abstract gives no fibre parameters, span lengths, modulation formats, error bars, measurement cross-checks or independent-code comparison. In multi-band fibre the dominant physics is direction-dependent inter-band Raman transfer, so any systematic bias in the solver or NLI model scales almost linearly into the 24.5% figure. Because the unconstrained case shows zero gain, the reported benefit is precisely that model-constraint interaction. With only the abstract we cannot judge how solid the models are.

This is for optical-communications engineers working on multi-band or bidirectional system design who need quantitative guidance under power limits. It is not foundational, but the number is actionable if it survives scrutiny. I would send the full paper to a serious referee; the result is important enough inside the subfield and formally grounded enough as a numerical study to deserve that time, even if heavy revision on validation is likely. Based on the abstract alone I would not cite it yet and would only bring it to reading group if the full methods appear.

Referee Report

4 major / 3 minor

Summary. The manuscript presents a numerical optimisation study of bidirectional OESCL-band coherent transmission under total fibre launch-power constraints. Using a GPU-accelerated boundary-value Raman solver and launch-power optimisation, the authors report that, for a three-span link subject to an 18 dBm total fibre power limit, reversing the O-band propagation direction relative to the remaining bands increases aggregate capacity by up to 24.5 % compared with a co-propagating configuration. In the unconstrained case the same directional reversal yields no capacity gain.

Significance. If the numerical result is confirmed under realistic models and validated conditions, the work would be of practical interest for multi-band optical transport design: it identifies band-direction configuration as a useful degree of freedom once total power is limited by fibre or amplifier constraints. The GPU acceleration of a bidirectional Raman boundary-value problem is a potentially reusable methodological contribution. The capacity figure is presented as the output of a forward search rather than a quantity forced by construction, which is a methodological strength provided the underlying Raman, NLI and transceiver models are accurate and externally anchored.

major comments (4)

The load-bearing 24.5 % capacity gain (abstract) is stated without fibre parameters (loss, Raman gain spectrum, A_eff), span length, WDM channel plan, modulation formats, FEC overhead or transceiver SNR models. Without these the result cannot be reproduced or compared with independent work; they are essential for any claim of this magnitude.
Inter-band Raman power transfer is the dominant, strongly direction-dependent physics. The abstract supplies no measurement validation, independent-code cross-check or residual-error analysis of the GPU-accelerated BVP Raman solver. Any systematic bias in bidirectional multi-band power evolution maps almost linearly onto the reported capacity difference and therefore onto the central claim.
The conversion from Raman power profiles to aggregate capacity relies on unspecified nonlinear-interference and transceiver models. Because the unconstrained case shows zero gain, the 24.5 % benefit is precisely the interaction of the 18 dBm power limit with those models; their formulation (ISRS-aware GN, residual bidirectional NLI, transceiver limits) is load-bearing and must be stated and justified.
The co-propagating baseline, the precise definition of the total fibre power limit (per-span or end-to-end, including or excluding ASE), and the optimisation algorithm itself are not described. It is therefore unclear whether 24.5 % is a robust optimum or a single-point numerical outcome.

minor comments (3)

The abstract should state at least headline fibre type, span length and channel plan so that the 24.5 % figure can be contextualised by a reader.
“OESCL-band” should be expanded or referenced on first use for non-specialist readers of an eess.SP venue.
Error bars, sensitivity ranges or a brief statement of numerical convergence around the 24.5 % figure would strengthen the abstract claim.

Circularity Check

0 steps flagged

No circularity: capacity gain is numerical output of constrained optimisation, not forced by definition or self-citation.

full rationale

Only the abstract is available. It describes a GPU-accelerated boundary-value Raman solver used for launch-power optimisation of bidirectional OESCL-band transmission under an external total-fibre power limit. The reported 24.5% capacity increase (and the zero-gain unconstrained case) are presented as numerical results of that search, not as quantities obtained by fitting a parameter that is then re-labelled a prediction, nor by invoking a uniqueness theorem or ansatz from the authors' prior work. No equations, self-citations, or definitional identities appear in the supplied text that would allow a reduction of the claimed gain to its inputs by construction. Model-validation risk is a correctness concern, not circularity. Score 0 with empty steps is therefore the warranted finding.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

Abstract-only review. Free parameters, modelling axioms and any invented entities cannot be enumerated from the text; the ledger therefore records only the high-level modelling assumptions that must be present for the claimed optimisation to be meaningful.

free parameters (1)

total fibre launch-power limit = 18 dBm
The 18 dBm figure is an externally imposed constraint that defines the regime in which the 24.5 percent gain appears; its precise numerical value is chosen by the authors for the study.

axioms (2)

domain assumption Steady-state Raman power-transfer equations accurately describe bidirectional multi-band power evolution in the fibre under study.
The entire optimisation rests on a boundary-value Raman solver; if the Raman model is incomplete the capacity numbers are unreliable.
domain assumption Nonlinear interference and transceiver noise models used for capacity evaluation are sufficiently accurate under the optimised power profiles.
Capacity is computed from the solved power profiles; any systematic error in the NLI or transceiver model scales the reported gain.

pith-pipeline@v1.1.0-grok45 · 5969 in / 2161 out tokens · 17089 ms · 2026-07-15T02:18:40.877100+00:00 · methodology

0 comments

read the original abstract

We optimise bidirectional OESCL-band coherent transmission under total fibre power constraints using a GPU-accelerated boundary-value Raman solver and launch-power optimisation. While no gain is observed without power constraints, for an 18-dBm limit 3-span transmission reversing the O-band direction increases aggregate capacity by up to 24.5%

GPU-Accelerated Optimisation of Symmetric Bidirectional Ultrawideband Coherent Transmission Under Launch Power Constraints

Core claim

What carries the argument

Load-bearing premise

What would settle it

If this is right

Where Pith is reading between the lines

discussion (0)