{"id":"36f69cd0-5898-4a4f-874c-b073af9d630a","arxiv_id":"2412.03916","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Outdoor measurements at 220-230 GHz show rain causes strong attenuation but keeps Rician K-factors above 40 dB, and simulations indicate QAM and lower frequencies are more robust for terahertz links.","lead":"This paper reports outdoor measurements of 220 to 230 GHz terahertz links during rain, finding significant power loss but little multipath scattering, with strong line-of-sight components preserved. It uses those measurements plus simulations to compare modulation schemes and recommends adaptive modulation and lower-frequency planning for future 6G links.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 40-50 dB Rician K-factors may be an artifact of power-sensor noise rather than evidence of minimal rain multipath; without a noise-floor calibration, the central 'no ISI' claim is not yet established.","rationale":"The reader's conditional verdict is based largely on the BER model being ad hoc; I agree that Eq. 5 is under-specified and that the BER analysis is not a direct measurement. But the more load-bearing issue is upstream: the experimental K-factor values that motivate the whole 'rain does not cause multipath' narrative. The reported values are so extreme that they are near or below the resolution of many power sensors, and no instrument characterization is given. In addition, the paper's own Table 2 shows mixed goodness of fit, with Weibull sometimes beating Rician, which weakens the claim that the 40-50 dB K values are physically resolved. This is not an accusation of error; it is an unstated measurement-system assumption that must be checked before the central claim can be accepted. A fixed-attenuator calibration test is cheap and would settle it. If a static channel also yields K>40 dB, then the rain K conclusion is likely an artifact; if it yields much lower K, then the rain K values are meaningful. Because the concern is addressable with additional calibration data, the verdict should remain conditional rather than move to accept or reject.","tokens_in":12792,"tokens_out":8898,"duration_ms":86613,"concrete_test":"Perform a zero-fading calibration: acquire 1-minute, 7 Hz power records in the same 54 m link under stable clear weather with a calibrated fixed attenuator inserted at the receiver, and fit Rician and Weibull to those records exactly as in Section 3 and Table 2. If the fitted K is again above 40 dB and the Rician/Weibull R2 separation is comparable to the rain rows, then the reported K-factors are dominated by the measurement system's noise floor and cannot support the rain-multipath conclusion. A supporting check is to report the raw power trace and the sensor's noise-equivalent power or resolution, with 95% confidence intervals on K from bootstrap resampling.","verdict_should_be":"UNCHANGED","load_bearing_attack":"To support the central claim that a 220-230 GHz link in rain retains K>40 dB and hence negligible multipath/ISI, the measured CDF must actually resolve the tiny diffuse component implied by such K values. At K=50 dB, the relative power standard deviation is roughly 0.02 dB (sqrt(2/K)); the paper does not report the power sensor's resolution, noise floor, or any calibration run through a static channel. Table 2 is also ambiguous: at 220 GHz, 8 mm/hr, the Weibull fit has R2=0.9961 versus Rician R2=0.9953, and at 229 GHz, 23 mm/hr, Weibull has R2=0.9976 versus Rician R2=0.9960, so the claim that Rician 'performed relatively better' is not consistently supported. If the fitted K is determined by sensor noise, slow drift, or scintillation rather than by rain-induced multipath, the headline conclusion about minimal ISI is unsupported, and using K=46 dB in the subsequent BER analysis becomes circular. The rain-rate representativeness issue identified by the reader is real but secondary: it affects the attenuation-versus-rate comparison, whereas the K-factor issue is load-bearing for the 'strong LoS' part of the central claim.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an outdoor terahertz channel measurement campaign at 220, 225, and 229 GHz over a 54 m line-of-sight path during rain and clear weather. The authors analyze the cumulative distribution functions of the received power, fit Rician and Weibull distributions, and infer Rician K-factors of 40–50 dB, concluding that rain-induced multipath and inter-symbol interference are negligible. They compare measured rain attenuation with the ITU-R P.838-3 model and Mie scattering predictions for several raindrop size distributions, finding the measurements bounded above by ITU-R and below by the Joss-Drizzle model. A bit error rate (BER) analysis is then carried out using a Rician fading model with an assumed K-factor formula, comparing modulation schemes and evaluating the effects of rain rate, temperature, humidity, and frequency. The central conclusions are that rain mainly attenuates the LoS component without strong multipath, QAM performs well in rain, and lower THz frequencies are more resilient.","tokens_in":13037,"tokens_out":5382,"duration_ms":46886,"significance":"If the central claims hold, the paper provides valuable experimental evidence on THz propagation in rain at 220–230 GHz, a frequency range of direct relevance to 6G backhaul and access links. The finding that Rician K-factors remain above 40 dB would support the use of simple LoS channel models and adaptive modulation in rainy conditions, and the attenuation bounds with respect to ITU-R and Mie models are practically useful for link budgeting. The paper also benefits from a real outdoor measurement campaign with co-located meteorological data. However, the significance is conditional on the K-factor extraction being physically meaningful, and on the BER model parameters being justified; these are the points of concern in the major comments.","major_comments":[{"comment":"The claim that the Rician distribution 'performed relatively better' is not consistently supported by the reported R² values. For example, at 220 GHz with 8 mm/hr rain, the Weibull fit has R²=0.9961 versus Rician R²=0.9953, and at 229 GHz with 23 mm/hr rain, Weibull R²=0.9976 versus Rician R²=0.9960. The paper should justify the Rician choice, perhaps with a formal goodness-of-fit test that accounts for the differing number of parameters, or explicitly acknowledge that the two distributions are statistically indistinguishable for several rows of Table 2. This is load-bearing because the K-factors and the 'minimal multipath' conclusion derive from the Rician fit.","section":"Section 3.1, Table 2"},{"comment":"The measurement system's noise floor, power sensor resolution, and static-channel calibration are not reported. At K-factors of 40–50 dB, the relative power standard deviation from multipath is on the order of sqrt(2/K) ≈ 0.02 dB, which is extremely small. If the measured power CDF spread is dominated by sensor noise, slow drift, or atmospheric scintillation rather than by rain-induced multipath, the fitted K-factor would not be a valid indicator of multipath severity. The authors should provide a noise-floor measurement, a static (no-rain) calibration run, and an estimate of the achievable dynamic range to demonstrate that the K-factors are resolved by the measurement system.","section":"Section 2 and Section 3.1"},{"comment":"The proposed K-factor formula is under-specified. The exponent denoted α in Eq. (5) is not clearly defined; the text first introduces α as a path loss exponent in the formula for path loss, but it is not stated whether the same α is used in the K-factor formula. The parameters K0 and Rc (set to 50 mm/hr) are selected ad hoc rather than derived from the measurement data. Since this formula is used to generate the BER predictions in Figs. 6 and 7, the quantitative BER results are not reproducible and their sensitivity to the free parameters is unknown. The authors should derive the distance and rain-rate dependence of K from the measurements, or clearly present the BER analysis as an illustrative scenario with a sensitivity study over the free parameters.","section":"Section 4, Eq. (5)"},{"comment":"The BER-versus-frequency predictions rely on an atmospheric absorption model that is not described or referenced. The statement that clear-weather operation is limited to 350 GHz and rainy operation to 300 GHz depends on the specific gaseous absorption model used (e.g., ITU-R P.676 or a line-by-line model), which is not identified. Without specifying the absorption model and its parameters, the frequency-dependent BER thresholds and the recommendation for lower-frequency planning are not verifiable. The model should be named, and its sensitivity to humidity and temperature should be discussed.","section":"Section 4, Fig. 7(a)"},{"comment":"The rain rate used in the attenuation comparison is taken from a China Meteorological Administration station described only as 'proximate' to the campus. The distance between the station and the 54 m channel, and the spatial uniformity of rain along the 41.5 m exposed path, are not quantified. If the station rain rate differs from the path-average rain rate, the attenuation-versus-rain-rate curves in Fig. 4 and the conclusion that measurements fall between the ITU-R and J-D bounds could be biased. The authors should report the station distance and, if possible, provide an uncertainty estimate for the rain rate along the channel.","section":"Section 2 and Section 3.2, Fig. 4"}],"minor_comments":[{"comment":"The Mie scattering formula in Eq. (2) is garbled; the constant 3.3429 and the integration variable are not clearly defined. Please rewrite the equation in standard notation and specify the units of the extinction efficiency and raindrop radius.","section":"Section 3.2, Eq. (2)"},{"comment":"The text states that QAM has 'superior BER performance,' but the simulation results show that BPSK, 2-PAM, and 4-QAM have identical BER under the Rician channel. The claim should be softened to 'comparable BER with higher spectral efficiency' to avoid overstatement.","section":"Section 4, Fig. 5"},{"comment":"The caption says 'linear K-factor is calculated by for Eq. (5),' which is a typo; it should read 'calculated by Eq. (5).'","section":"Section 4, Fig. 7 caption"},{"comment":"The table header contains the typo 'Ricain'; it should be 'Rician.'","section":"Table 2"},{"comment":"There are numerous typographical and grammatical errors (e.g., 'the influence of rain-induced fading on power profile and bit error performances remains poorly understood,' 'this impact it not so obvious'). A thorough language edit is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely and important topic, and the experiment is nontrivial. However, the central claim of minimal multipath rests on K-factors that may be below the resolution of the measurement system; this needs to be addressed with calibration data. The BER model in Eq. (5) is also under-specified and currently functions as an arbitrary parameterization. I would be willing to re-review a revised version that adds a noise-floor calibration, a clearer justification of the Rician versus Weibull fit, a fully specified Eq. (5), and a reference to the atmospheric absorption model used in Fig. 7(a). The single-day measurement campaign is a limitation that the authors do acknowledge, but the generalizability claims should be correspondingly moderated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this is a straightforward measurement paper with a useful new dataset at 220-230 GHz, but the central claim that rain leaves the Rician K-factor above 40 dB needs a noise-floor calibration before it can carry the 'no multipath, no ISI' conclusion. The BER analysis is a simulation, not a measurement.\n\nWhat is new: a 54 m outdoor channel at 220, 225, and 229 GHz, with rain rates up to 25 mm/hr, and CDF-based fits to Rician and Weibull distributions. That frequency/path combination is genuinely new. The attenuation data falls between the ITU-R upper bound and the Joss-Drizzle lower bound, and the paper sensibly presents this as a prediction range rather than forcing a single model. The observation that power loss varies at constant rain rate, attributed to raindrop size distribution, is consistent with earlier work and worth having documented at these frequencies.\n\nWhat is soft: First, the K-factors above 40 dB imply received power fluctuations of order 0.02 dB; the paper does not report a static-channel calibration, noise floor, or sensor resolution, so the fitted K may reflect sensor noise rather than rain multipath. The stress-test note is right that this is load-bearing. Second, Eq. (5) defines a K-factor scaling with distance and rain rate but never defines alpha_c, and Rc=50 mm/hr is manually selected. The BER curves in Figs. 6 and 7 are consequences of that assumed form, so calling them predictions overstates what is known. Third, the claim of QAM superiority is overstated: BPSK, 2-PAM, and 4-QAM give identical BER, and the paper simply moves on to 16-QAM. Also, in two of the rain cases the Weibull fit has slightly higher R2 than Rician, so the statement that Rician performed relatively better is not uniformly supported. Minor, but the paper should acknowledge it.\n\nThe measurements appear carefully done within their scope, and the paper is candid about limitations. The issues are addressable with a calibration run and a clearer description of the K-factor formula. I would send this to peer review — the dataset deserves scrutiny, and the K-factor question can be settled with proper calibration.\n\nFor your reading group, worth a look if anyone works on THz link budgets; otherwise it's a pass.","headline":"New 220-230 GHz rain dataset is worth a look, but the high-K-factor claim needs a noise-floor calibration before it can carry the 'no multipath' conclusion.","tokens_in":13661,"tokens_out":3104,"would_cite":false,"duration_ms":26895,"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":"This paper claims that a 220-230 GHz terahertz link in rain keeps a strong line-of-sight path—Rician K-factors stay above 40 dB—so rain mostly costs power, not coherence.","keywords":["terahertz communications","rain attenuation","Rician K-factor","bit error rate","Mie scattering","raindrop size distribution","QAM modulation"],"falsifier":"Install a disdrometer along the exposed part of the path and compare its one-minute raindrop size distributions with the station rain rate during the same storms; if measured attenuation systematically falls outside the ITU-R upper and Joss-Drizzle lower bounds when the local distribution is used, the prediction-range claim fails.","tokens_in":1517,"feed_emoji":"🌧️","tokens_out":2044,"duration_ms":126106,"temperature":0.7,"pith_summary":"The paper is trying to establish that rain does not destroy the basic usability of terahertz links in the 220-230 GHz band: it attenuates the signal but leaves a strong line-of-sight path, so the channel stays Rician with K-factors above 40 dB and inter-symbol interference stays small. If true, this matters because it separates rain's effect into a power-budget problem—solvable by transmit power, adaptive modulation, and frequency choice—rather than a waveform-coherence problem. The authors support this with outdoor measurements over 54 meters under natural rain and with a paired theoretical analysis using ITU-R and Mie-scattering attenuation models, plus BER calculations for QAM and other modulations.","feed_headline":"Rain keeps THz links LOS-strong: K-factor above 40 dB","feed_subtitle":"220-230 GHz measurements over 54 m show multipath stays weak; rain mainly cuts power, so adaptive modulation can hold the link.","key_machinery":"The Rician K-factor—the ratio of line-of-sight power to scattered multipath power—fitted to one-minute cumulative distribution functions of received SNR is the quantity that carries the claim that rain does not add significant multipath. Supporting machinery includes the ITU-R power-law attenuation model and Mie-scattering attenuation computed with exponential raindrop size distributions, with the ITU and Joss-Drizzle models bracketing the measured loss. A K-factor formula extended with rain rate and path-loss exponent then feeds a 16-QAM bit-error-rate expression under Rician fading.","core_discovery":"The paper reports that over a 54 m outdoor path at 220, 225, and 229 GHz, rain lowers received SNR but leaves the channel Rician with K-factors above 40 dB, meaning the line-of-sight component dominates and multipath-induced inter-symbol interference is negligible. Measured rain attenuation falls between the ITU-R prediction (upper bound) and the Joss-Drizzle Mie-scattering prediction (lower bound), while power loss at equal rain rates varies because raindrop size distribution changes. In BER terms, 16-QAM matches BPSK and 2-PAM under strong LOS fading, 2-FSK is worst, a 41.5 m link at 20 dBm stays near BER $2\\times10^{-7}$ up to 50 mm/hr, and a 1 km link needs 40 dBm to stay below the forward-error-correction threshold for rain rates below about 30 mm/hr. Higher frequencies show more humidity sensitivity than lower ones, so the paper concludes that adaptive modulation and lower-THz frequency planning are the practical levers.","pith_inferences":["If K-factors stay above 40 dB across rain rates, receiver equalizers for terahertz links could be far simpler than those for multipath-dominated bands; the paper does not design equalizers, but its Rician fits point in that direction.","The observed constant-rate power spread implies link margins should be set by raindrop-size-distribution variability, not just the rain-rate reading, since a margin built only on a single attenuation curve would miss the spread at fixed rain rates.","A natural extension is to measure 140 GHz and 220-230 GHz in the same storms to test the paper's frequency-sensitivity conclusion directly, rather than comparing to a separate earlier 140 GHz study.","The BER model sets the critical rain rate at 50 mm/hr; in regions with more extreme rain, the predicted operating window for 1 km links would close faster, so that threshold should be validated locally before deployment."],"forward_implications":["At 220-230 GHz over tens of meters, rain-induced multipath is weak enough that inter-symbol interference is not the limiting factor in rain.","Measured attenuation falls between the ITU-R upper bound and the Joss-Drizzle lower bound, giving a prediction range rather than a single value for link budgets.","16-QAM achieves the same BER as BPSK and 2-PAM under the strong line-of-sight channel, so higher spectral efficiency is available in rain without a BER penalty.","At 1 km, 20 dBm transmit power fails the FEC threshold in rain, while 40 dBm sustains reliable BER for rain rates below about 30 mm/hr.","Higher THz frequencies are more sensitive to humidity, so lower-frequency channels offer greater environmental resilience in rainy conditions."],"supporting_citations":[{"why":"Supplies the ITU-R power-law attenuation coefficients used as the upper-bound prediction for rain loss.","marker":"[43]"},{"why":"Provides the Joss-Drizzle raindrop size distribution used for the lower-bound Mie-scattering attenuation prediction.","marker":"[46]"},{"why":"Gives the Marshall-Palmer raindrop size distribution used as the moderate-to-heavy rain reference model.","marker":"[44]"},{"why":"Provides the double-Debye dielectric model for water that sets the frequency-dependent rain attenuation in the Mie calculations.","marker":"[49]"},{"why":"Earlier outdoor 140 GHz rain and snow measurements whose multipath behavior motivates the comparison of CDF shapes.","marker":"[9]"},{"why":"Shows that conventional prediction models underestimate BER degradation in heavy rain, motivating the BER analysis here.","marker":"[21]"},{"why":"Laboratory emulated-rain 16-QAM study used as a modulation baseline for the BER calculations.","marker":"[13]"},{"why":"Documents spatial variation in raindrop size distributions, used to explain the measured power variability at constant rain rate.","marker":"[27]"},{"why":"Prior terahertz rain measurement and modeling at 140 GHz that this campaign extends to 220-230 GHz.","marker":"[25]"}],"fun_headline_variants":["Rain vs THz: LOS holds, power falls, BER survives","220-230 GHz rain test: multipath negligible, power varies","THz rain: raindrop size, not rate, drives power loss","Rain on THz: 16-QAM shines, FSK suffers, LOS strong","54m THz in rain: K-factor >40 dB, adaptive modulation wins"],"cache_read_input_tokens":15744,"weakest_assumption_plain":"The results assume the rain rate reported by the nearby weather station matches what actually falls on the 54-meter path during each one-minute measurement; if local rain differs, the attenuation-versus-rate curves and the constant-rate variability conclusion shift.","fun_headline_variants_meta":{"raw":{"variants":["Rain vs THz: LOS holds, power falls, BER survives","220-230 GHz rain test: multipath negligible, power varies","THz rain: raindrop size, not rate, drives power loss","Rain on THz: 16-QAM shines, FSK suffers, LOS strong","54m THz in rain: K-factor >40 dB, adaptive modulation wins"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000458,"raw_usage":{"total_tokens":2318,"prompt_tokens":985,"completion_tokens":1333,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":601,"completion_tokens_details":{"reasoning_tokens":1233}},"tokens_in":601,"tokens_out":1333,"duration_ms":12477,"temperature":1.0,"reasoning_tokens":1233,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:56:16.529592+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Install a disdrometer along the exposed part of the path and compare its one-minute raindrop size distributions with the station rain rate during the same storms; if measured attenuation systematically falls outside the ITU-R upper and Joss-Drizzle lower bounds when the local distribution is used, the prediction-range claim fails.","supporting_citations":[],"review_version":1}