{"id":"2fa381c9-4d14-4b3e-9cd4-2ca334988257","arxiv_id":"2606.23527","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":11,"one_line_summary":"At 64 GBaud, a modeled transceiver back-to-back SNR ceiling near 25 dB makes 1024-QAM infeasible on both standard and hollow-core fiber, leaving hollow-core's advantage in reach at 256-QAM and below.","lead":"This paper builds a unified signal-to-noise budget for very high-speed optical links over hollow-core fiber, combining fiber, amplifier, gas, and transceiver impairments. It concludes that at 64 GBaud the transceiver, not the fiber, caps the achievable modulation format at 256-QAM, so hollow-core fiber should be valued for reach and baud rate rather than for ultra-high-order QAM.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unmeasured transceiver calibration is load-bearing: a ~1-bit ENOB shift at 64 GBaud would flip 1024-QAM from infeasible to feasible.","rationale":"I read the paper as a system-modeling contribution whose novel synthesis is a reciprocal SNR budget placing fiber and transceiver impairments on the same footing. That synthesis is internally consistent and the paper is unusually explicit about its calibration assumptions. The most load-bearing point is exactly the one the reader identified: the 64-GBaud 'infeasible' statement inherits all uncertainty from an unmeasured, hand-calibrated transceiver ceiling, and the uncertainty is quantitatively relevant because a ~1 dB converter/ENOB shift at 64 GBaud, combined with the analog and nonlinearity floors, corresponds to roughly a 1-bit ENOB difference and could move 1024-QAM across the feasibility threshold. I agree with the reader rather than partially because this is not just one of several weaknesses—the headline numerical conclusion is built on it. I would not move the verdict: the paper's qualitative transceiver-limited picture has independent support (published 20 GBaud HCF PS-1024-QAM and 60 GBaud PS-256-QAM fall near the modeled boundaries; the model's low-baud consistency is informative), and the paper explicitly disclaims the absolute calibration. The reader's CONDITIONAL verdict already captures the need for a measurement or scoped wording; my stress-test sharpens the specific quantitative test but does not change the verdict. A direct 64-GBaud B2B measurement with 1024-QAM, or at least an inferred ENOB(64), would settle whether the abstract's strong 'infeasible' language is warranted.","tokens_in":25529,"tokens_out":9923,"duration_ms":101418,"concrete_test":"Measure the back-to-back SNR of a state-of-the-art 64-GBaud coherent transceiver at the 1024-QAM operating point: ideal optical loopback with no fiber, 50 kHz ECL Tx/LO, high-end DAC/ADC, and the same receiver DSP and pilot-aided CPE as the paper's Section III-C model. If the achieved pre-FEC SNR per polarization exceeds ≈29.0 dB (1024-QAM threshold minus 1.53 dB PCS credit plus 1.0 dB margin), the central 'infeasible at 64 GBaud' conclusion fails. If the measured SNR is below ≈26 dB, the model's ceiling is confirmed. If no 1024-QAM transceiver mode is available, measure with a probe format and use the model's PAPR/ENOB relation to infer ENOB(64); check whether ENOB(64) is ≥6.0 bits, the value needed to make 1024-QAM feasible under the paper's own parallel-noise combination.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—1024-QAM and above are infeasible at 64 GBaud on either fiber—rests on the transceiver B2B ceiling of Section III-B, Eqs. (6)–(8): ENOB0=5.5 at 32 GBaud, 0.6 bit/octave roll-off, S_afe=38 dB, S_nl=42 dB. Section III-B explicitly states these are calibration parameters and that no measured B2B curve is used. At 64 GBaud the modeled ENOB is 4.9 bits and SNR_TRx≈25.3 dB. The net 1024-QAM requirement is ≈29.0 dB (threshold 29.6 dB − PCS credit 1.53 dB + margin 1.0 dB), leaving a ~3.7 dB gap. That gap corresponds to needing ENOB(64)≈6.0 bits instead of 4.9—a ~1-bit calibration shift at 64 GBaud, not an enormous amount in the reported converter survey spread. If a real best-in-class 64-GBaud transceiver delivers >29 dB B2B SNR, 1024-QAM becomes feasible at short/medium reach on HCF and the abstract's 'infeasible on either fiber regardless of fiber quality' is false. The low-baud external checks (Fan 20 GBaud, Wang 60 GBaud) and the Varughese ~10 dB penalty anchor near 100 GBaud do not pin down the 64-GBaud crossover to ±1 bit: the Fan check exercises the low-baud limb, and the Varughese anchor is itself model-based. The qualitative statement that the transceiver—not the fiber—becomes the bottleneck is more robust; the precise 1024-QAM infeasibility boundary is not yet anchored to direct measurement. This is acknowledged in Section V-F as a calibration freedom, but the abstract presents the boundary as a physical result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a per-channel effective-SNR budget for coherent WDM transmission over hollow-core fiber (HCF) and standard SMF, combining ASE, Kerr NLI, inter-modal interference (IMI), pigtail NLI, CO2 gas absorption, a parameterized transceiver back-to-back (B2B) SNR ceiling, and SNR-equivalent penalties for phase noise, EEPN, jitter, PDL, and line-system effects. The central quantitative claim is that at 64 GBaud with 75 GHz spacing, the modeled transceiver B2B ceiling (~25.3 dB for the 'high-performance' class) falls below the net requirement for 1024-QAM (~29.0 dB), making 1024-QAM and higher infeasible on either fiber regardless of fiber quality. HCF's advantage is consequently stated to be reach and baud at a given modulation order (e.g., 256-QAM reach ~170 km vs ~45 km at κ=-55 dB/km), with additional analysis of CO2 absorption, launch power, and IMI sensitivity. The model is a synthesis of published formulas and parametrized/calibrated components, with explicit caveats about unmeasured quantities.","tokens_in":25948,"tokens_out":20273,"duration_ms":172980,"significance":"If the transceiver B2B ceiling were anchored to direct measurement, this would be an important reframing of the HCF roadmap: it would show that HCF's near-vacuum nonlinearity does not automatically translate into ultra-high QAM at deployment baud rates, and that converter ENOB and analog bandwidth are the system bottleneck. The paper's transparency about its parameters and its external consistency checks (20 GBaud PS-1024-QAM and 60 GBaud PS-256-QAM demonstrations) are strengths, and the qualitative conclusion that the transceiver becomes the limiting factor at high baud is plausible. However, the absolute '1024-QAM infeasible at 64 GBaud' boundary is not yet anchored to a measured B2B curve; a ~1-bit ENOB shift at 64 GBaud would change the feasibility conclusion. The framework is valuable, but the headline quantitative claim needs additional support.","major_comments":[{"comment":"The central '1024-QAM infeasible at 64 GBaud' claim rests on the unmeasured, hand-calibrated transceiver B2B ceiling. The model uses ENOB0=5.5 at 32 GBaud, 0.6 bit/octave roll-off, S_afe=38 dB, and S_nl=42 dB. At 64 GBaud this gives ENOB≈4.9 and SNR_TRx≈25.3 dB, while the net 1024-QAM requirement from Eq. (3) is 29.57-1.53+1.0≈29.0 dB—a gap of about 3.7 dB. A ~1-bit increase in ENOB at 64 GBaud (ENOB≈6.0 instead of 4.9) raises the quantization floor by about 6 dB and, even with the parallel floors, brings the ceiling near or above 29 dB, making 1024-QAM feasible on HCF at short reach. The external anchors (Fan 20 GBaud, Wang 60 GBaud) exercise lower orders or lower baud rates, and the Varughese anchor near 100 GBaud is itself model-based; none pins the 64-GBaud high-order crossover to ±1 bit. Since the abstract presents the boundary as a physical result, the paper should either include a","section":"Section III-B, Eq. (8), Table II; Section IV-A; Abstract"},{"comment":"The closed form for the all-pairs splice-MPI sum appears incorrect as printed. For N_s=2, Eq. (18) gives S(2,ζ)=ζ, but the numerator shown in Eq. (19), if read as ζ^{N_s-1}-N_sζ+ζ^{N_s}, gives (ζ-2ζ+ζ^2)/(1-ζ)^2=(ζ^2-ζ)/(1-ζ)^2<0 for 0<ζ<1. The correct closed form should be [(N_s-1)ζ - N_sζ^2 + ζ^{N_s+1}]/(1-ζ)^2. If this is a typographical error, it should be corrected; if not, the numerical discrete-MPI results in Section IV-D and the conclusion that splice MPI is below the distributed-IMI floor are unreliable. This is a load-bearing point for the paper's 'distributed κ dominates; discrete splice MPI is small' message.","section":"Section III-F, Eq. (19)"}],"minor_comments":[{"comment":"The text says EEPN 'enters the transceiver ceiling of Eq. (8),' but Eq. (8) is the pure B2B ceiling and Section II-A explicitly excludes EEPN from that ceiling. EEPN should enter Eq. (2) as a separate SNR-equivalent penalty.","section":"Section III-D"},{"comment":"The analog-bandwidth floor S_afe(R_s) depends on an unspecified R_bw. 'Calibrated so the bandwidth penalty approaches ~10 dB near 100 GBaud' is not sufficient to reproduce Table II; please state R_bw or the explicit calibration formula.","section":"Section III-B"},{"comment":"The symbol κ is used both for the distributed IMI coefficient (dB/km) and for the linear splice coupling coefficients κ_in, κ_out. This is confusing; use different symbols (e.g., η for coupling).","section":"Section III-F"},{"comment":"The text says the ceiling 'crosses the 1024-QAM requirement near 24 GBaud.' Using the stated parameters and the net requirement of Eq. (3) (~29.0 dB), the crossover appears to be closer to 27–29 GBaud. Please clarify whether the comparison uses the raw threshold 29.57 dB or the net requirement, and derive the crossover from the formulas.","section":"Section IV-A"},{"comment":"The 'required-SNR lines' in Fig. 1 should state explicitly whether they include G_PCS and the 1.0 dB margin; the text uses both 'threshold' and 'net requirement' at different points.","section":"Figure 1 and Eq. (3)"},{"comment":"The phase-noise penalty δ_PN=(ΔνT_s/τ_1dB)^2 is not derived. The small-signal BPS variance analysis cited from Pfau scales linearly in ΔνT_s for the phase-error variance; a quadratic scaling should be justified or replaced with a linear form calibrated at the same 1 dB point.","section":"Eq. (9)"}],"recommendation":"major_revision","confidential_remarks":"The paper's synthesis is useful and the qualitative transceiver-bottleneck conclusion is likely robust, but the headline quantitative boundary at 64 GBaud depends on an unmeasured calibration. I would ask the authors for a measured B2B curve or a thorough sensitivity sweep before publication. The Eq. (19) issue also needs correction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper is worth reading as a system-design tool, but take the headline number with a grain of salt. The authors build the first per-channel SNR budget that puts HCF-specific fiber terms (IMI, pigtail NLI, CO2) in the same reciprocal frame as a symbol-rate-dependent transceiver B2B ceiling, and they use it to generate reach-versus-baud maps and a CO2 gas-line analysis. The qualitative conclusion—that at deployment baud rates the transceiver, not the fiber, caps the QAM order—is plausible and consistent with published demonstrations (PS-256 at 60 GBaud, PS-1024 at 20 GBaud). They are unusually honest about what is measured and what is calibrated: the B2B ceiling is a parameterization anchored to converter surveys and Varughese's ENOB analysis, and they flag phase-noise interpolation and MFA taper NLI as estimates.\n\nThe soft spot is the load-bearing calibration. The 64 GBaud transceiver ceiling is ~25.3 dB, which sits ~3.7 dB below the net 1024-QAM requirement. The stress-test note is correct: that gap corresponds to about 1 ENOB bit at 64 GBaud. A real best-in-class front-end with ENOB ~6 at 64 GBaud would make 1024-QAM feasible on HCF at short reach. So the abstract's \"infeasible on either fiber regardless of fiber quality\" is stated too strongly; the correct statement is \"infeasible under the modeled transceiver assumptions.\" The authors do include this caveat in Section V-F, but the abstract and conclusion don't carry it. That's an editorial fix, and it matters because the number will get quoted.\n\nAlso minor: reach numbers are single points, not bands (except the cabling IMI sensitivity), and the CO2 notch-ISI saturation constant is another calibration freedom. These are acknowledged. The all-pairs MPI closed form is a nice piece of work.\n\nMy recommendation: send it to review. The referee should press for either a measured back-to-back curve from a 64 GBaud-class transceiver or a sharpened abstract that scopes the claim. The paper's value as a design tool is real even if the exact crossover baud shifts. I would cite it for the reach maps and the gas-line analysis.","headline":"Useful integrated budget for HCF system design, but the '1024-QAM infeasible at 64 GBaud' headline rests on a hand-calibrated transceiver curve; the qualitative bottleneck shift is robust, the exact boundary is not.","tokens_in":26562,"tokens_out":2246,"would_cite":true,"duration_ms":23021,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"At 64 GBaud, a coherent transceiver's ~25 dB back-to-back ceiling makes 1024-QAM infeasible on standard and hollow-core fiber alike.","keywords":["hollow-core fiber","coherent optical transmission","higher-order QAM","transceiver SNR ceiling","effective SNR budget","inter-modal interference","CO2 absorption","quadrature amplitude modulation"],"falsifier":"Measure the back-to-back SNR of a state-of-the-art 64-GBaud coherent transceiver into an ideal link. If it exceeds about 29.6 dB (the 1024-QAM requirement with shaping credit and margin), the 64-GBaud infeasibility claim collapses on hollow-core fiber, since the corresponding link SNR there is about 35 dB. Alternatively, an end-to-end 64-GBaud 1024-QAM transmission over a few kilometers of hollow-core fiber with standard external-cavity lasers would directly refute the ceiling claim if successful.","tokens_in":1722,"feed_emoji":"📡","tokens_out":2600,"duration_ms":65569,"temperature":0.7,"pith_summary":"This paper tries to establish that the practical ceiling on modulation order in coherent optical systems is set by the transceiver's own back-to-back SNR, not by the fiber, once hollow-core fiber removes fiber nonlinearity. It builds a per-channel effective-SNR budget that combines optical-link noise, a symbol-rate-dependent transceiver ceiling, and secondary penalties in reciprocal form, then applies it at a deployment-realistic 64 GBaud, 75 GHz-spaced, 6 THz WDM system. The headline result: 1024-QAM needs about 29.6 dB Es/N0, but a best-in-class transceiver delivers only about 25.3 dB back-to-back at that baud, so 1024-QAM and above are infeasible on either fiber. Hollow-core fiber still wins on reach: 256-QAM reach improves from about 45 to about 170 km and 64-QAM from about 415 to about 2275 km at an inter-modal interference coefficient of -55 dB/km. CO2 gas lines matter mainly in the L-band, where a channel on a line loses one to two QAM orders, while line-avoiding C-band channels follow the gas-free baseline.","feed_headline":"Transceiver, not fiber, caps QAM at 256 for 64-GBaud links","feed_subtitle":"A 25 dB transceiver ceiling blocks 1024-QAM on both fibers; HCF still triples reach.","key_machinery":"The key object is the effective-SNR budget: 1/SNR_total = 1/SNR_link + 1/SNR_TRx(Rs) + sum_i 1/SNR_i, where SNR_link collects true optical noise powers (ASE, Kerr NLI, inter-modal interference, pigtail NLI), SNR_TRx is a parametrized transceiver back-to-back ceiling from ENOB-at-rate, analog bandwidth, and Tx/Rx nonlinearity, and the remaining impairments (phase noise, equalization-enhanced phase noise, timing jitter, PDL, line-system penalties, CO2 gas-notch ISI) enter as equivalent SNR floors. The reciprocal form is what makes the ceiling binding: the achievable SNR can never exceed the transceiver ceiling no matter how clean the link is.","core_discovery":"The central discovery is that a real coherent transceiver has a finite, symbol-rate-dependent back-to-back SNR ceiling—set by converter ENOB at the operating rate, analog bandwidth, and Tx/Rx nonlinearity—and that at 64 GBaud this ceiling (~25.3 dB for the high-performance class) sits below the 29.6 dB Es/N0 requirement for 1024-QAM. Because the total SNR combines the link SNR and the transceiver ceiling in parallel (reciprocally), no fiber improvement can lift the delivered SNR above the ceiling. The paper concludes that hollow-core fiber's near-vacuum Kerr nonlinearity does not enable ultra-high QAM at deployable baud rates; it enables longer reach and higher baud at a given modulation ord","pith_inferences":["If the parameterized transceiver ceiling is optimistic—for example, if a real 64-GBaud transceiver delivers more than about 30 dB back-to-back SNR—1024-QAM could become feasible on hollow-core fiber at 64 GBaud, so the exact crossover baud is uncertain even though the transceiver-limited picture likely survives.","The same reciprocal budget could be reused for future converter generations or other fiber types by swapping in a different ENOB-at-rate curve; the framework predicts a baud-versus-order design rule in which higher baud forces moderate order.","The CO2 notch-ISI result suggests a testable mitigation prediction: spectral pre-emphasis or digital-subcarrier avoidance should recover most of the C-band worst-channel reach penalty at modest spectral cost, while L-band avoidance remains spectrally expensive because the strong lines are denser than the channel bandwidth.","The IMI-reach prediction could be tested directly by measuring the cabled inter-modal interference coefficient on deployed hollow-core fiber; the paper itself identifies cabled-IMI characterization as the key open measurement for this class of result."],"forward_implications":["Ultra-high QAM (1024-QAM and above) is feasible only at low baud rates, roughly below 24 GBaud, even with a high-performance transceiver; at 64 GBaud the maximum feasible order is 256-QAM on either fiber.","Hollow-core fiber's practical value is reach and baud at a given order rather than raw modulation order: about 3.8x more reach for 256-QAM and 5.5x more for 64-QAM at the baseline IMI coefficient.","Improving the distributed inter-modal interference coefficient is the highest-leverage fiber-side lever: moving from -55 to -73 dB/km extends 256-QAM reach from about 170 km to about 975 km, though the paper flags this as a laboratory bound.","CO2 absorption forces spectral planning: line-avoiding C-band channels follow the gas-free baseline, while L-band channels on strong lines lose one to two QAM orders unless the core is sealed or mitigation is used.","Any future move beyond 256-QAM at deployable baud rates requires transceiver improvements—higher ENOB at rate, wider analog bandwidth, and lower Tx/Rx nonlinearity—rather than fiber improvements."],"fun_headline_variants":["HCF can't unlock 1024-QAM at 64 GBaud—transceiver ceiling","Transceiver ceiling blocks ultra-high QAM at 64 GBaud","Hollow-core fiber: reach wins, but transceiver caps QAM","At 64 GBaud, transceiver sets QAM ceiling, not HCF"],"cache_read_input_tokens":27520,"weakest_assumption_plain":"The load-bearing premise is that the parameterized back-to-back transceiver SNR ceiling—ENOB0 of 5.5 at 32 GBaud, a 0.6 bit/octave roll-off, a 38 dB analog floor, and a 42 dB nonlinearity floor—accurately represents real 64-GBaud coherent transceivers; it is calibrated to published trends rather than measured, and its absolute level shifts the exact baud at which 1024-QAM becomes infeasible.","fun_headline_variants_meta":{"raw":{"variants":["HCF can't unlock 1024-QAM at 64 GBaud—transceiver ceiling","Transceiver ceiling blocks ultra-high QAM at 64 GBaud","Hollow-core fiber: reach wins, but transceiver caps QAM","At 64 GBaud, transceiver sets QAM ceiling, not HCF"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000374,"raw_usage":{"total_tokens":1926,"prompt_tokens":933,"completion_tokens":993,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":677,"completion_tokens_details":{"reasoning_tokens":909}},"tokens_in":677,"tokens_out":993,"duration_ms":9040,"temperature":1.0,"reasoning_tokens":909,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T10:24:52.619056+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the back-to-back SNR of a state-of-the-art 64-GBaud coherent transceiver into an ideal link. If it exceeds about 29.6 dB (the 1024-QAM requirement with shaping credit and margin), the 64-GBaud infeasibility claim collapses on hollow-core fiber, since the corresponding link SNR there is about 35 dB. Alternatively, an end-to-end 64-GBaud 1024-QAM transmission over a few kilometers of hollow-core fiber with standard external-cavity lasers would directly refute the ceiling claim if successful.","supporting_citations":[],"review_version":3}