{"id":"d9a1abc5-5454-4953-946b-e60073f28d29","arxiv_id":"2504.15589","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A validation study claims the 3GPP TR 38.901 indoor hotspot path loss model remains valid for the 7-24 GHz band, based on fitted floating-intercept and alpha-beta-gamma parameters from existing NYU WIRELESS measurements at two mid-band frequencies.","lead":"This paper checks whether the 3GPP standard's indoor path loss formula still holds for the 7-24 GHz band that may be used by 6G, by fitting two standard path loss curves to existing NYU WIRELESS measurements at 6.75, 16.95, 28, and 73 GHz. The authors conclude the standard model is valid for 7-24 GHz, but the evidence rests on only two frequencies in one building.","discovery_kind":"replication","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper's 'close alignment' premise omits a consistent 5-6 dB FI LOS intercept offset and builds the NLOS band-wide gamma on only two frequencies; Sections V.A and VI concede the latter.","rationale":"The reader correctly identified the two-frequency gamma as a load-bearing weak assumption. I find a second, more direct issue: the comparison tables omit the intercept parameters that control absolute path-loss level. Table I shows FI LOS intercepts differing by 5.6-6.1 dB, and Table II shows ABG offset beta differing by 4.2 dB in LOS and up to 19.5 dB in NLOS, yet these are not among the reported |Delta| columns. Since the paper's only evidence for validity is parameter closeness, an unquantified path-loss bias of roughly 2 sigma in LOS, combined with an unstable two-point gamma in NLOS, does not support the unconditional 'remains valid' conclusion. The manuscript itself flags the frequency-point limitation in V.A ('The observed discrepancy for gamma may stem from the limited number of frequency points') and VI ('Using only two frequency points ... limits the robustness and generalizability'), so this is an internal limitation, not an external standard. The data and fit procedure in [21] are transparent enough to recompute a direct prediction-error metric; until that is done, the verdict should remain conditional rather than accept or reject.","tokens_in":10562,"tokens_out":11421,"duration_ms":106348,"concrete_test":"Recompute the mean absolute path-loss prediction error of 3GPP Eqs. (3)-(4) against the raw omnidirectional path-loss points at 6.75 and 16.95 GHz, separately for LOS and NLOS, over the measured TX-RX distances. If the LOS mean absolute error exceeds the model's 3 dB sigma_SF, or the NLOS error exceeds the option-specific 8.03/8.29 dB sigma_SF, the 'close alignment' premise fails and the Section VI conclusion should be softened; if the errors sit within those sigma_SF values, the intercept offset is absorbed by the model's declared stochastic range and the conditional claim can stand.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that 3GPP TR 38.901 InH path loss remains valid for 7-24 GHz, justified by 'close alignment' of FI/ABG parameters. The paper's own tables undercut this. In Table I, FI LOS intercepts are alpha = 43.4 vs 48.98 dB at 6.75 GHz and 50.9 vs 56.98 dB at 16.95 GHz, a consistent roughly 6 dB offset, yet no |Delta alpha| is reported; only distance-exponent and shadow-fading differences are shown. In Table II, ABG offset beta is 28.2 vs 32.4 dB in LOS and 12.9 vs 17.3/32.4 dB in NLOS, again omitted from the reported deltas. A 6 dB LOS offset is about twice the model's 3 dB shadow-fading sigma. In addition, the NLOS ABG gamma is fitted from only two frequencies in the same NYU building, giving gamma = 3.4 versus 2.49 (Option 1) and 2.0 (Option 2) in 3GPP, a difference of 0.91-1.4. Sections V.A and VI explicitly state that two frequency points make the ABG parameters sensitive and limit robustness. Since 'close alignment of parameters' is the only evidence offered, and both absolute-level and frequency-dependence parameters show unreported or substantial discrepancies, the conclusion that the model 'remains valid' for NLOS, and even the LOS offset, is not established by the paper as written.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper validates the 3GPP TR 38.901 indoor hotspot (InH) path loss model for the 7-24 GHz range, using NYU WIRELESS measurements at 6.75, 16.95, 28, and 73 GHz. The authors fit floating-intercept (FI) models at 6.75 and 16.95 GHz and alpha-beta-gamma (ABG) models over 7-24 GHz and 0.5-100 GHz, then compare the fitted parameters with the fixed 3GPP coefficients. They conclude that the existing 3GPP InH path loss model remains valid for 7-24 GHz in both LOS and NLOS conditions, and they frame the work as a contribution to the 3GPP Release 19 study on channel model validation.","tokens_in":10922,"tokens_out":2851,"duration_ms":24535,"significance":"If the conclusion were firmly established, the paper would provide a useful data point for the 3GPP Release 19 decision on whether to keep the TR 38.901 InH path loss model unchanged for 6G upper-mid-band simulations. The paper's strengths are that it uses real-world measured path loss data, applies closed-form fitting procedures from a well-known reference, and explicitly acknowledges some limitations of its two-frequency analysis in Section VI. However, the supporting evidence as presented is weaker than the conclusion because key discrepancies (intercept offsets and NLOS frequency exponents) are omitted or understated, and the validation switches between the two 3GPP NLOS options in a post-hoc manner.","major_comments":[{"comment":"The FI comparison in Table I and the text in Section IV entirely omit the intercept (alpha) difference. For LOS, the measured alpha is 43.4 dB versus 48.98 dB at 6.75 GHz and 50.9 dB versus 56.98 dB at 16.95 GHz, a consistent 5.6-6.0 dB offset that is about twice the 3 dB shadow-fading sigma. The paper reports only |Delta beta| and |Delta sigma_SF|, so the central claim of 'close alignment' is not supported for the absolute path loss level; the measured loss is systematically lower than the 3GPP model by roughly 6 dB.","section":"Section V.A, Table II"},{"comment":"The NLOS ABG frequency exponent gamma is fitted from only two frequencies (6.75 and 16.95 GHz), both measured in the same NYU building, and the result is gamma = 3.4, which differs from the 3GPP values 2.49 (Option 1) and 2.0 (Option 2) by 0.91 and 1.4, respectively. The paper itself states in Section V.A that the discrepancy may stem from the limited number of frequency points. A 0.91-1.4 difference in the frequency exponent is a substantial model mismatch, not a close alignment, and the conclusion that the NLOS model remains valid for 7-24 GHz is not established by the reported evidence.","section":"Section IV and Table I"},{"comment":"The NLOS FI validation switches between 3GPP Option 1 and Option 2 depending on the frequency: at 6.75 GHz the measured beta = 3.6 is compared favorably with Option 1 (3.83), while at 16.95 GHz the measured beta = 2.8 is compared favorably with Option 2 (3.19). No a priori criterion is given for selecting the option used in the comparison, so the procedure amounts to choosing the better-fitting option after seeing the data. This weakens the validity claim and should be addressed by a fixed decision rule or by reporting the comparison against both options consistently.","section":"Sections IV and V.A"},{"comment":"All comparisons are made at the level of point estimates with no measure of statistical uncertainty. The fitted parameters are based on 20 TX-RX locations at 6.75/16.95 GHz and 48 locations at 28/73 GHz, and the paper reports differences in exponents as small as 0.03 as if they were meaningful. The authors should provide standard errors, confidence intervals, or a sensitivity analysis (e.g., bootstrap over locations) so that the reader can judge whether the observed differences are within sampling variability.","section":"Sections IV-V"}],"minor_comments":[{"comment":"There are typos in the header line: 'V erification', 'V alidation', 'Y oshimura', 'I shii', and '6G a nd Beyond' contain stray spaces or uppercase letters from the converted text.","section":"Equation (1)"},{"comment":"Equation (1) repeats the word 'where': 'where, where alpha and gamma...' should be a single 'where'.","section":"Footnotes"},{"comment":"The footnote on page 1 reads '16.75 GHz and 16.95 GHz was selected'; this should be '6.75 GHz and 16.95 GHz were selected'.","section":"Section III"},{"comment":"The statement that the FI model is obtained 'when setting gamma = 0 or 2 in the ABG path loss model' is unclear; gamma is normally set to zero for a single frequency, while setting gamma = 2 corresponds to a free-space frequency dependence. The sentence should be rephrased for accuracy.","section":"Table II"},{"comment":"The caption and text describe |Delta sigma_SF| as 'the absolute difference' but the table also reports |Delta alpha| and |Delta gamma|; consider adding units (dB for alpha and sigma, dimensionless for gamma) directly in the column headers to avoid ambiguity.","section":"Table II"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a short conference paper (VTC2025-Spring) submitted to a journal. The central idea of validating TR 38.901 with FR3 band measurements is timely, and the underlying data are from the authors' own prior measurement campaign. The main technical concerns are the omitted FI intercept deltas, the post-hoc selection between NLOS Options 1 and 2, and the absence of uncertainty quantification; these are fixable within the manuscript's scope, but they are load-bearing for the 'remains valid' conclusion. I would also suggest the editor check whether the underlying point data are accessible via the cited arXiv data set, since the paper does not include the raw measurement points."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a legitimate 3GPP Release 19 contribution, but the headline conclusion is stronger than the evidence. The paper fits FI and ABG models to existing NYU measurements at 6.75 and 16.95 GHz and compares the fitted parameters to the 3GPP TR 38.901 coefficients. That comparison for the 7-24 GHz band is new, and the paper is clearly written and honest about its limitations in Section VI. For those reasons, it deserves a serious referee.\n\nThe soft spots are real and visible in the paper's own tables. The FI LOS intercepts are 43.4 vs 48.98 dB at 6.75 GHz and 50.9 vs 56.98 dB at 16.95 GHz — a consistent ~6 dB offset, roughly twice the 3 dB shadow-fading sigma. Table I reports |Δβ| and |Δσ| but not |Δα|. The ABG offset β is likewise 28.2 vs 32.4 dB in LOS and 12.9 vs 17.3 or 32.4 dB in NLOS, again unreported. Omitting these numbers gives the 'close alignment' claim more cover than it deserves. The NLOS ABG frequency exponent γ is 3.4 against 2.49/2.0, a difference of 0.91-1.4, fitted from only two frequencies in one building; the paper concedes this in Section V.A. The conclusion then says the model 'remains valid' — that does not follow from the evidence as presented. On top of that, the NLOS comparison switches between 3GPP Option1 and Option2 depending on frequency, which is hard to defend without an a priori rule, and there are no confidence intervals on any fitted parameter.\n\nTo be fair: the LOS distance exponents and shadow-fading values are close, the data are externally published, and the fitting method is standard. The paper is not circular — it compares external measurements against fixed standard coefficients. The self-citations are to the data papers, which is appropriate.\n\nThis paper is for 3GPP RAN1 delegates, not for a general propagation audience. The underlying data are fine, but the validation should be reframed: the LOS intercept offset and the NLOS γ discrepancy mean the current model is approximately right in slope but not demonstrably valid across the band. A revision that reports all parameter deltas, adds uncertainty quantification (or at least acknowledges its absence), and commits to one NLOS option would make this a solid contribution.\n\nRecommendation: send it to peer review. It is useful and honest, but it needs a careful revision before acceptance.","headline":"A useful 3GPP contribution that overclaims: the parameter deltas in its own tables, especially the ~6 dB LOS intercept offset and the two-frequency NLOS gamma, do not support the 'valid for 7-24 GHz' conclusion as stated.","tokens_in":11488,"tokens_out":2857,"would_cite":false,"duration_ms":25387,"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":"Measurements at 6.75 and 16.95 GHz confirm the existing 3GPP TR 38.901 indoor hotspot path loss model remains valid for the 7-24 GHz band.","keywords":["3GPP","TR 38.901","6G","7-24 GHz","ABG path loss model","FI path loss model","FR3","upper mid-band"],"falsifier":"Measure the path loss in an indoor hotspot at one or two additional frequencies inside 7-24 GHz, for example 10 GHz and 20 GHz, in a different building, and fit the ABG model to all points. If the NLOS frequency exponent gamma moves away from 2.49 or 2 by more than the current 0.91-1.4 discrepancy, or if the model's predicted path loss at those frequencies falls outside the measured shadow-fading sigma, then the paper's conclusion that the model remains valid for NLOS would fail.","tokens_in":10344,"feed_emoji":"📶","tokens_out":5559,"duration_ms":40990,"temperature":0.7,"pith_summary":"This paper tries to establish that the existing 3GPP TR 38.901 indoor hotspot (InH) path loss model, which was largely interpolated between sub-6 GHz and millimeter-wave data, remains accurate in the 7-24 GHz upper-mid band. Using wideband measurements at 6.75 GHz and 16.95 GHz, plus 28 GHz and 73 GHz, the authors fit floating-intercept (FI) and alpha-beta-gamma (ABG) path loss models and compare the fitted coefficients with the TR 38.901 equations. The close agreement in distance exponents, frequency exponents, and shadow fading, especially in line-of-sight conditions, supports keeping the model unchanged for 6G simulations in this band. The result matters because standardization bodies are deciding whether to revise the channel model for 7-24 GHz, and confirming the existing model would save effort and preserve continuity.","feed_headline":"Indoor path loss model for 7-24 GHz survives new tests","feed_subtitle":"Measurements at 6.75 and 16.95 GHz match the TR 38.901 hotspot model in LOS and NLOS.","key_machinery":"The validation is carried by two standard path loss models: the floating-intercept (FI) model, which describes how signal attenuation grows with distance at a single frequency, and the alpha-beta-gamma (ABG) model, which adds a frequency term. The 3GPP TR 38.901 InH equations have the same functional form as ABG, so the authors convert the standard equations into FI and ABG form, fit those forms to measured omnidirectional path loss data, and compare coefficients. The load-bearing step is the frequency exponent gamma in the ABG fit: when gamma from the measured 7-24 GHz data is close to the model's gamma, the model's frequency scaling is validated; when gamma deviates, as it does in NLOS, the paper argues the deviation is an artifact of using only two frequency points.","core_discovery":"On its own terms, the paper claims that the 3GPP TR 38.901 InH path loss model is valid for the 7-24 GHz frequency range in both LOS and NLOS conditions. For LOS, the measured distance exponent (FI) is 1.7 at both 6.75 and 16.95 GHz, matching the model's 1.73 to within 0.03, and the frequency exponent gamma from the ABG fit is 1.9 versus the model's 2. For NLOS, the measured values depend on which of the model's two options is chosen: at 6.75 GHz the measured distance exponent 3.6 is within 0.23 of Option1, while at 16.95 GHz the measured 2.8 is within 0.39 of Option2. The paper acknowledges that the NLOS gamma discrepancy is larger (0.91 to 1.4), which it attributes to having only two frequency points, but concludes that the overall alignment of alpha, beta, and shadow-fading parameters demonstrates validity for 7-24 GHz.","pith_inferences":["If two-frequency gamma is as sensitive as the paper concedes, then a band-wide NLOS validity claim should be treated as provisional until at least a third frequency in the middle of the band is measured; the paper's own limitations text supports this caution.","The discrepancy pattern, with the model's gamma lower than the measured gamma, suggests that the TR 38.901 NLOS model may systematically underestimate path loss at the upper end of 7-24 GHz in enclosed indoor offices, which could matter for interference and coverage planning in 6G.","A testable extension: applying the same FI/ABG comparison to outdoor urban or rural scenarios would tell whether the interpolation concern that motivated the 7-24 GHz study is an indoor-only issue.","If the existing model is retained, the practical consequence is that some NLOS indoor deployments in the upper mid-band may need a small additional attenuation margin; this is an inference, not a claim of the paper."],"forward_implications":["If the conclusion holds, 3GPP can keep the InH path loss model unchanged for 6G studies in the 7-24 GHz band, avoiding re-fitting or retuning.","LOS conditions in indoor hotspots show negligible model error (distance exponent within 0.03), so link budget and coverage estimates in open indoor spaces can rely on the existing model.","The NLOS case is less certain: the measured gamma 3.4 versus the model's 2.49 or 2 implies the model may underpredict path loss at higher frequencies in NLOS, so system-level simulations of indoor NLOS links should carry this uncertainty.","The 0.5-100 GHz ABG fit also aligns, suggesting the model's frequency scaling is consistent across the whole range when more frequency points are included.","Future measurements at intermediate frequencies, such as 10 GHz and 20 GHz, can reduce the gamma uncertainty and either confirm or refine the model."],"supporting_citations":[{"why":"Supplies the 6.75 GHz and 16.95 GHz indoor hotspot measurement data and the companion channel model parameters used for validation.","marker":"[9]"},{"why":"Provides the detailed propagation measurement campaign and channel models at 6.75 GHz and 16.95 GHz in the indoor environment.","marker":"[10]"},{"why":"Defines the 3GPP Release 19 study on channel model validation for 7-24 GHz that this paper directly feeds into.","marker":"[16]"},{"why":"The TR 38.901 specification whose InH path loss equations are the target of validation.","marker":"[17]"},{"why":"Provides the point-by-point omnidirectional path loss data for the mid-band frequencies used to derive the FI and ABG fits.","marker":"[20]"},{"why":"Supplies the 28 GHz and 73 GHz measurement data and the closed-form FI/ABG fitting procedure used throughout the analysis.","marker":"[21]"}],"fun_headline_variants":["Indoor path loss model validated for 7-24 GHz in new tests","3GPP hotspot model holds for 7-24 GHz in LOS and NLOS","New measurements confirm path loss model across 7-24 GHz","Path loss model stands up to tests at 6.75 and 16.95 GHz"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The validity claim for the 7-24 GHz band assumes that a frequency-dependence exponent gamma fitted to just two measured frequencies, 6.75 GHz and 16.95 GHz, both in the same indoor building, is representative of the whole band; if the true band-wide gamma differs, the NLOS conclusion is not established.","fun_headline_variants_meta":{"raw":{"variants":["Indoor path loss model validated for 7-24 GHz in new tests","3GPP hotspot model holds for 7-24 GHz in LOS and NLOS","New measurements confirm path loss model across 7-24 GHz","Path loss model stands up to tests at 6.75 and 16.95 GHz"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000626,"raw_usage":{"total_tokens":2959,"prompt_tokens":1072,"completion_tokens":1887,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":688,"completion_tokens_details":{"reasoning_tokens":1802}},"tokens_in":688,"tokens_out":1887,"duration_ms":13052,"temperature":1.0,"reasoning_tokens":1802,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:22:13.116609+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the path loss in an indoor hotspot at one or two additional frequencies inside 7-24 GHz, for example 10 GHz and 20 GHz, in a different building, and fit the ABG model to all points. If the NLOS frequency exponent gamma moves away from 2.49 or 2 by more than the current 0.91-1.4 discrepancy, or if the model's predicted path loss at those frequencies falls outside the measured shadow-fading sigma, then the paper's conclusion that the model remains valid for NLOS would fail.","supporting_citations":[{"cited_title":"Comprehensive FR1(C) and FR3 Lower and Up per Mid-Band Propagation and Material Penetration Loss Measur ements and Channel Models in Indoor Environment for 5G and 6G,","cited_arxiv_id":null,"evidence_quote":"Supplies the 6.75 GHz and 16.95 GHz indoor hotspot measurement data and the companion channel model parameters used for validation."},{"cited_title":"RP-234018 New SID: Study on channel modelling enhance ments for 7-24 GHz for NR,","cited_arxiv_id":null,"evidence_quote":"Defines the 3GPP Release 19 study on channel model validation for 7-24 GHz that this paper directly feeds into."},{"cited_title":"Study on channel model for frequencies from 0.5 t o 100 GHz,","cited_arxiv_id":null,"evidence_quote":"The TR 38.901 specification whose InH path loss equations are the target of validation."},{"cited_title":"In- door Office Wideband Millimeter-Wave Propagation Measurem ents and Channel Models at 28 and 73 GHz for Ultra-Dense 5G Wireless Networks,","cited_arxiv_id":null,"evidence_quote":"Supplies the 28 GHz and 73 GHz measurement data and the closed-form FI/ABG fitting procedure used throughout the analysis."}],"review_version":1}