{"id":"7e182055-cbec-469f-bc0d-9b05e9f3650a","arxiv_id":"2608.03783","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In a U6GHz XL-MIMO prototype field trial, downlink throughput is strongly controlled by the target SNR, and model-based SNR analysis shows saturation as antenna count grows.","lead":"This paper reviews 6G spectrum plans and XL-MIMO channel modeling for the 6-24 GHz mid-band, then adds new measurements and a U6GHz field trial with a 1024-element prototype. It finds that the target signal-to-noise ratio is the dominant factor for downlink capacity, while uplink performance stays limited.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Field-trial conclusion overreaches: Fig. 20 is single-stream rate-vs-SNR, so it cannot support the claimed spatial-multiplexing benefit.","rationale":"The reader's verdict (CONDITIONAL) is broadly appropriate, but the specific weakest assumption they flag—stationarity of the virtual 128x6 array formed by sliding the 32x2 physical array—is not the most load-bearing risk for the paper's central claim. That virtual-array method supports the survey/measurement sections (III-IV) and the angular/capacity/near-field analyses, but the headline field-trial claim in Section VI uses a separate 1024-element physical AAU and a single-stream rate-vs-SNR plot. The load-bearing problem is that Fig. 20 cannot bear the 'spatial multiplexing gain' part of the conclusion: the caption and text describe single-stream rate, and the observed increase with SNR is a generic property of adaptive modulation and coding. Without rank/multi-stream data or a controlled comparison of array size and SNR, the central claim overstates what the experiment establishes. This is an internal-evidence problem, not a disagreement with community consensus. The model-based NUSW saturation result (V.D, Fig. 15) is a separate analytical point and does not validate the field-trial conclusion. I would keep the overall CONDITIONAL recommendation, but the revision requirement should focus on either adding multi-stream/rank evidence or narrowing the conclusion to single-stream link adaptation.","tokens_in":28196,"tokens_out":6397,"duration_ms":75579,"concrete_test":"Obtain the raw field-trial logs behind Fig. 20 and extract, for each reported measurement point, the rank indicator (RI) and the number of scheduled layers. If all points are single-stream (RI=1), the data cannot support the spatial-multiplexing claim; the paper should be revised to state only that higher SNR improves modulation order and single-stream rate. If RI>1 appears at high SNR, re-plot throughput versus SNR separated by rank and test whether the multiplexing gain increases with SNR at fixed channel conditions.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central field-trial claim (abstract; Section VI.C; Section VII.A) is that sufficiently high SNR substantially improves peak downlink capacity and enhances spatial multiplexing gain. The only reported evidence is Fig. 20, whose caption explicitly says 'single-stream rate and SNR'. The curve shows throughput rising through the QPSK, 16QAM, 64QAM, and 256QAM modulation regions; this is standard adaptive modulation and coding behavior for any single-user link and does not depend on XL-MIMO. No rank indicator, number of streams, or multi-stream throughput is reported, so the 'spatial multiplexing gain' part of the conclusion is not measured. Moreover, 'target SNR' appears to be the achieved/observed SNR, so the correlation with rate may be confounded by distance or channel realization; there is no controlled variation of target SNR at fixed array and channel conditions. The model-based NUSW analysis (Section V.D, Fig. 15) shows near-field SNR saturation, but that is a separate analytical/simulated comparison and does not provide field-trial evidence for multiplexing gain. Thus the paper's headline result is an overgeneralization of a single-stream link-adaptation curve.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper is a hybrid survey-and-experimental contribution on FR3 (6-24 GHz) XL-MIMO for 6G. It reviews spectrum allocation and standardization activities, describes a wideband TDM-MIMO channel sounder and a virtual 128x6 (1536-element) array formed by mechanically sliding a 32x2 array, summarizes measured channel characteristics (angular spreads, channel hardening, capacity, near-field phase, spatial non-stationarity), reviews channel estimation and beamforming algorithms, presents model-based SNR comparisons between far-field UPW and near-field NUSW models for 768 and 1536 antennas, and reports U6GHz field trials with a 1024-element prototype. The headline conclusion is that the target SNR is a critical factor for XL-MIMO performance: sufficiently high SNR substantially improves peak downlink capacity and spatial multiplexing gain, whereas uplink performance remains constrained.","tokens_in":28539,"tokens_out":6911,"duration_ms":77841,"significance":"If fully supported, the field-trial result would be a valuable datapoint for U6GHz XL-MIMO deployment, and the channel sounder covering 3-16 GHz with up to 1536 virtual elements is a useful experimental platform. The model-based UPW-versus-NUSW comparison also highlights an important qualitative point about near-field array gain saturation. However, the central field-trial claim currently overreaches: Fig. 20 is explicitly a single-stream rate-versus-SNR curve, and the demonstrated monotonic throughput growth across QPSK/16QAM/64QAM/256QAM is standard adaptive modulation behavior, not evidence of spatial multiplexing gain. The review portions are broad and cite a substantial body of work, though they lean heavily on the authors' own prior publications. Overall, the paper is a useful survey with an intriguing but not yet established experimental conclusion.","major_comments":[{"comment":"The claim that sufficiently high SNR 'substantially improves peak downlink capacity and enhances the spatial multiplexing gain' is not supported by the evidence presented. Fig. 20 is explicitly labeled 'single-stream rate and SNR'; the throughput increase with SNR through the QPSK, 16QAM, 64QAM, and 256QAM regions is standard adaptive modulation and coding for a single link and does not depend on XL-MIMO. No rank indicator, number of streams, or multi-stream throughput is reported, so the 'spatial multiplexing gain' part of the conclusion is not measured. In addition, 'target SNR' appears to be the achieved/observed SNR; without controlled variation at fixed array and channel conditions, the correlation may be confounded by distance, shadowing, or channel realization. The model-based NUSW analysis in Section V.D is a separate analytical/simulated comparison and does not provide field-tri","section":"Section VI.C, Fig. 20; abstract; Section VII.A"},{"comment":"The virtual 128x6 (1536-element) array is formed by mechanically sliding a 32x2 physical array through four horizontal and three vertical translations. The validity of the measurement-based results in Section IV (angular spreads, inverse condition number, channel capacity, near-field phase verification) hinges on the assumption that the propagation channel is stationary over the entire mechanical translation interval. The manuscript does not report any stationarity validation, such as repeated reference-path measurements during the sliding procedure, nor does it quantify the translation time or environment stability. If the environment changes during the multiple translations, the measured angular spreads, capacities, and near-field phase checks will be distorted. Please add a stationarity check or explicitly state and justify the stationarity assumption and its possible effect on the re","section":"Section III, Fig. 4"},{"comment":"The model-based SNR comparison is not reproducible as written. No closed-form expressions for the UPW and NUSW SNR are given, and the absolute path gain at a reference distance, transmit power, noise figure, bandwidth, and array normalization are not specified. Consequently, the key qualitative claim that the NUSW SNR converges to a constant bound while the UPW SNR grows unboundedly cannot be checked from the manuscript, and the reported 3.01 dB spacing between 1536 and 768 elements in Fig. 14 is asserted rather than derived from the stated model. In addition, the text after Fig. 13 says at 'the 100 MHz frequency point' the far-field UPW model breaks down because the array aperture expands to 'hundreds of meters'; with half-wavelength spacing and the described 16-column modular array, this aperture estimate is not consistent with the stated geometry, and the plotted frequency range in th","section":"Section V.D, Figs. 13-15"}],"minor_comments":[{"comment":"There is a sign inconsistency: Eq. (3) correctly writes S(k) = -1 when p_n_k - p_n_{k-1} <= -3 dB, but the surrounding text says 'when p_n_k - p_n_{k-1} <= 3 dB' without the minus sign. Please correct the text.","section":"Section IV.A.5, Eq. (3)"},{"comment":"The weights w_c, w_a, w_d and the threshold rho in Eq. (5) are not specified or referenced. Since the stationary-interval partitioning result depends on these choices, please provide default values or cite the estimation procedure.","section":"Section IV.A.5, Eq. (5)"},{"comment":"The figure would be substantially more informative with error bars or confidence intervals, the number of repeated trials, and a definition of how 'target SNR' is set or measured. The caption currently states only that modulation switching regions are indicated.","section":"Section VI.C, Fig. 20"},{"comment":"There are numerous typographical and style errors (e.g., 'Besides, The research' in Section I.3, 'U A V' in Section I.3, 'the 6G open innovation test device' repeated, and several missing articles). A full language edit is recommended before resubmission.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript bundles a broad survey with several experimental threads; the main novel field-trial claim is not yet supported by the single-stream evidence in Fig. 20. The review and model sections rely heavily on the authors' own prior publications (e.g., [12], [24], [25], [30]), which is acceptable if clearly delineated, but the paper should explicitly distinguish new contributions from review material. The claim in the title of reference [30] that a channel model was 'Adopted by 3GPP for 6G' may warrant editorial verification if it is used to bolster the model evaluation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is essentially a survey of FR3 XL-MIMO with two original pieces: a U6GHz field trial using a 1024-element prototype and a numerical SNR comparison for 768/1536-element arrays. The survey part is competent and reasonably current, and the description of the 3-16 GHz channel sounder plus the virtual-array measurement campaign is useful background for anyone entering this area. The field trial itself is a legitimate data point: 400 MHz, 22 m rooftop deployment, 1024 elements, and a rate-vs-SNR curve that shows ordinary link adaptation from QPSK to 256QAM.\n\nThe soft spot is the headline. The abstract and conclusion claim that high target SNR 'substantially improves peak downlink capacity and enhances the spatial multiplexing gain.' The only field-trial evidence is Fig. 20, whose caption says 'single-stream rate and SNR.' That curve shows standard adaptive modulation behavior for any single-user link and says nothing about the number of streams or rank. Spatial multiplexing gain is simply not measured. Nor is there error-bar, repetition, or statistical detail, and the 'target SNR' appears to be the achieved SNR, so the curve may be conflating distance or channel realization with link quality. The model-based SNR saturation in Section V.D is a well-known consequence of the near-field NUSW model and is not a new empirical result; the absolute path gain and reference power are not specified, so the numbers can't be checked. The virtual array used to emulate 1536 elements relies on channel stationarity over the mechanical translations, and that assumption is not validated.\n\nI don't think the paper is dishonest—it's a survey plus a thin experimental add-on, and the authors are open about relying on their own prior models. But the central claim is overgeneralized from a single-stream curve. A careful referee should ask for multi-stream throughput/rank reports, error bars, absolute SNR reference, and stationarity checks, and should push to reword the abstract. The survey content alone makes it worth a referee; the field trial does not resolve the open questions but is a real point of contact. I'd read it for the survey, not for the conclusion. I'd send it to review, but with a clear brief to check the evidence behind the spatial-multiplexing claim.","headline":"A broad, self-referential survey wrapped around a single-stream U6GHz field-trial curve; the spatial-multiplexing conclusion outruns the evidence.","tokens_in":29013,"tokens_out":3029,"would_cite":false,"duration_ms":33653,"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":"Field trials in the U6GHz band show that target signal-to-noise ratio, not raw antenna count, determines how much XL-MIMO downlink capacity is unlocked, while uplink remains constrained.","keywords":["FR3","new mid-band (6-24 GHz)","XL-MIMO","near-field propagation","spatial non-stationarity","channel modeling","U6GHz field trial","signal-to-noise ratio"],"falsifier":"Take a true 1536-element array and a sliding virtual array through the same UMa route and compare angular spread, capacity, and near-field phase correlation; if the two diverge beyond measurement uncertainty, the stationarity assumption behind the virtual array fails. Separately, plot measured or simulated SNR versus antenna count at a fixed FR3 frequency: continued 3 dB gain per doubling at large counts would falsify the NUSW saturation claim, while an observed plateau would confirm it.","tokens_in":1929,"feed_emoji":"📡","tokens_out":1972,"duration_ms":94568,"temperature":0.7,"pith_summary":"The paper is a system-level review and measurement study arguing that the new mid-band (FR3, 6–24 GHz) is a strong candidate spectrum for 6G, and that XL-MIMO with hundreds to thousands of antennas is the key technology to exploit it. Its central experimental claim comes from U6GHz (6425–7125 MHz) field trials: downlink throughput grows with target SNR, jumping from about 0.5 Gbps around the 64QAM threshold to nearly 1.5 Gbps in 256QAM, while uplink peaks near 0.44 Gbps. A companion model-based analysis shows why antenna-count scaling eventually saturates: under near-field non-uniform spherical wave propagation, SNR converges to a constant bound as the number of antennas grows, unlike the unbounded linear prediction of conventional far-field plane-wave models. The paper also reviews spectrum allocation, channel measurement, four XL-MIMO architectures (co-located, cell-free, sparse/movable, intelligent), near-field channel modeling, and estimation and beamforming algorithms. A careful reader would care because these results calibrate expectations: array size alone is not enough; the link budget and target SNR are what convert large arrays into throughput.","feed_headline":"XL-MIMO gains in the 6 GHz band hinge on target SNR, not antenna count","feed_subtitle":"Field trials show downlink throughput climbing toward 1.5 Gbps as SNR rises, while uplink plateaus near 0.44 Gbps.","key_machinery":"The load-bearing objects are (i) the near-field non-uniform spherical wave (NUSW) model, which treats each array element as seeing its own geometric distance and projected aperture to the source, and (ii) the virtual XL-MIMO array, built by sliding a 32×2 dual-polarized array through four horizontal and three vertical translations to emulate a 128×6 (1536-element) aperture for channel sounding. The NUSW model is what produces the SNR-saturation result; the virtual array is what turns a modest physical sounder into a large-aperture measurement without building a full 1536-element array. The field-trial counterpart is a 1024-element, 128-channel U6GHz prototype with 400 MHz bandwidth whose mea","core_discovery":"The paper's central discovery, stated on its own terms, is that the practical payoff of an extremely large array in the new mid-band is gated by the link SNR. In outdoor UMa field trials at 6425–6825 MHz with a 1024-element, 400 MHz prototype, downlink single-stream rate climbs monotonically with target SNR and enters higher-order modulation regions as SNR rises, while uplink throughput grows far more slowly. Complementing this, a model-based analysis of 768- and 1536-element modular arrays shows that under near-field non-uniform spherical-wave propagation the achievable SNR converges to a constant as antenna count grows, in contrast to the unbounded linear gain predicted by the far-field un","pith_inferences":["Editorial inference: the SNR-saturation curve implies an optimal array size below the physical maximum; beyond that point, marginal elements mainly add multi-user spatial separation rather than coherent single-user gain. The paper does not pursue this design trade-off.","Editorial inference: the virtual-array method is credible only if the propagation environment is frozen during the mechanical translations; if stationarity is violated, angular spreads could be inflated or subarray phase relationships distorted. A direct check would be comparing virtual-array results with a true 1536-element array along the same UMa route.","Editorial inference: the downlink/uplink asymmetry suggests the FR3 band may favor deployments with asymmetric link budgets, such as fixed wireless access or downlink-heavy traffic, unless uplink enhancement techniques mature.","Editorial inference: the same NUSW-based SNR saturation should appear at other frequencies within 6–24 GHz; whether the saturation point shifts with frequency is a clean next measurement to test the model's generality."],"forward_implications":["At a given site, engineering the link SNR (through coverage, power, beamforming gain, and modulation threshold) is the first-order lever; adding antennas beyond the point where SNR saturates yields little single-user rate under near-field propagation.","Near-field effects should be treated as a first-class constraint in XL-MIMO: algorithms and models that assume planar wavefronts will systematically overestimate SNR for very large arrays.","The U6GHz band can support multi-Gbps downlink in real outdoor deployments when high SNR is available, so system design should focus on extending high-SNR regions rather than only increasing array size.","Uplink requires a different solution set—UE transmit power, channel estimation accuracy, power control, or distributed/cell-free reception—since the trial shows it lags far behind downlink.","Measured angular spreads and channel-hardening trends provide calibration data for updating standardized channel models for the FR3 band."],"supporting_citations":[{"why":"Supplies the 6 GHz band measurement campaign and the TDM-MIMO sounder data behind the angular-spread and capacity results in Section IV.","marker":"[25]"},{"why":"The 3GPP TR 38.901 model is the far-field baseline that measured angular spreads and the proposed near-field and spatial non-stationarity extensions are compared against.","marker":"[13]"},{"why":"Provides earlier far-field-to-near-field 6 GHz experiments that validate spherical-wave phase behavior and motivate the near-field modeling.","marker":"[12]"},{"why":"Presents the unified near-field spatial non-stationarity channel model (spherical phases, element-wise SnS power, cluster variability) used in Section IV and noted as adopted by 3GPP for 6G.","marker":"[30]"},{"why":"Supplies the inverse-condition-number channel-hardening analysis used to show orthogonality grows with antenna count in the FR3 band.","marker":"[26]"},{"why":"Derives the closed-form maximum-SNR and MRC beamforming for modular XL arrays under the NUSW model, the basis for the SNR-convergence result in Fig. 15.","marker":"[48]"},{"why":"Extended version of the modular XL-array near-field modeling and performance analysis; supports the same SNR saturation conclusion.","marker":"[50]"},{"why":"Provides spectrum-allocation and propagation-perspective arguments for why the 6–24 GHz new mid-band is a 6G candidate, framing Section II.","marker":"[24]"},{"why":"Measurement-based comparison of cell-free versus conventional massive MIMO capacity in the FR3 band, supporting the cell-free XL-MIMO capacity analysis.","marker":"[36]"}],"fun_headline_variants":["XL-MIMO gains in mid-band tied to SNR, not antenna count","For 6G XL-MIMO, SNR is the real bottleneck, field trials show","Near-field limits: more antennas won't boost SNR in FR3","Field trials: SNR drives 6G XL-MIMO throughput, not array size","Mid-band XL-MIMO: SNR is king, antenna count saturates"],"cache_read_input_tokens":30720,"weakest_assumption_plain":"The load-bearing premise is that the virtual 1536-element array, assembled by sliding a 32×2 physical array across twelve translations, experiences an unchanged propagation channel during the whole measurement; any environmental change across translations distorts the measured angular spreads, capacities, and near-field phase checks.","fun_headline_variants_meta":{"raw":{"variants":["XL-MIMO gains in mid-band tied to SNR, not antenna count","For 6G XL-MIMO, SNR is the real bottleneck, field trials show","Near-field limits: more antennas won't boost SNR in FR3","Field trials: SNR drives 6G XL-MIMO throughput, not array size","Mid-band XL-MIMO: SNR is king, antenna count saturates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000535,"raw_usage":{"total_tokens":2457,"prompt_tokens":843,"completion_tokens":1614,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":587,"completion_tokens_details":{"reasoning_tokens":1513}},"tokens_in":587,"tokens_out":1614,"duration_ms":14832,"temperature":1.0,"reasoning_tokens":1513,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T12:34:50.239512+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a true 1536-element array and a sliding virtual array through the same UMa route and compare angular spread, capacity, and near-field phase correlation; if the two diverge beyond measurement uncertainty, the stationarity assumption behind the virtual array fails. Separately, plot measured or simulated SNR versus antenna count at a fixed FR3 frequency: continued 3 dB gain per doubling at large counts would falsify the NUSW saturation claim, while an observed plateau would confirm it.","supporting_citations":[{"cited_title":"Measurement-based massive MIMO channel characterization in 6 GHz band for 6G,","cited_arxiv_id":null,"evidence_quote":"Supplies the 6 GHz band measurement campaign and the TDM-MIMO sounder data behind the angular-spread and capacity results in Section IV."},{"cited_title":"Study on channel model for frequencies from 0.5 to 100 GHz(Release 15),","cited_arxiv_id":null,"evidence_quote":"The 3GPP TR 38.901 model is the far-field baseline that measured angular spreads and the proposed near-field and spatial non-stationarity extensions are compared against."},{"cited_title":"Far-field to near-field: Experimental studies of MIMO channel charac- terization and modeling in the 6 GHz band,","cited_arxiv_id":null,"evidence_quote":"Provides earlier far-field-to-near-field 6 GHz experiments that validate spherical-wave phase behavior and motivate the near-field modeling."},{"cited_title":"Near-field propagation and spatial non-stationarity channel model for 6–24 ghz (fr3) extremely large-scale mimo: Adopted by 3gpp for 6g,","cited_arxiv_id":null,"evidence_quote":"Presents the unified near-field spatial non-stationarity channel model (spherical phases, element-wise SnS power, cluster variability) used in Section IV and noted as adopted by 3GPP for 6G."},{"cited_title":"Channel hardening in 6G FR3 XL-MIMO: measurement-based analysis and modeling in a UMa scenario,","cited_arxiv_id":null,"evidence_quote":"Supplies the inverse-condition-number channel-hardening analysis used to show orthogonality grows with antenna count in the FR3 band."},{"cited_title":"Modular extremely large-scale array communication: Near-field modelling and performance analysis,","cited_arxiv_id":null,"evidence_quote":"Extended version of the modular XL-array near-field modeling and performance analysis; supports the same SNR saturation conclusion."},{"cited_title":"New mid-band for 6g: several considerations from the channel propagation characteristics perspective,","cited_arxiv_id":null,"evidence_quote":"Provides spectrum-allocation and propagation-perspective arguments for why the 6–24 GHz new mid-band is a 6G candidate, framing Section II."},{"cited_title":"Cell-free versus conventional massive mimo : An analysis of channel capacity based on channel measurement in the fr3 band,","cited_arxiv_id":null,"evidence_quote":"Measurement-based comparison of cell-free versus conventional massive MIMO capacity in the FR3 band, supporting the cell-free XL-MIMO capacity analysis."}],"review_version":1}