{"id":"45f5c6bb-c38f-41ae-b1b9-ffb7a0cfcadc","arxiv_id":"2508.15673","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A coded random access scheme using near-field spatial beams as replica resources and cluster-based combining plus SIC is shown by simulation to reduce packet loss in IIoT settings.","lead":"This paper presents a grant-free random access protocol in which factory sensors send the same packet along several beam directions at once to a large rooftop antenna array, using near-field effects to create parallel channels. Simulations show far fewer lost packets than a single-antenna-element approach, with the interference cancellation step playing a key role.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Near-field mode realization via DFT beams in Eq. (6) is the pivotal unverified premise; if the effective channel vectors H(k)b(k,j) are not near-orthogonal, the R spatial replicas do not act as independent CRA resources, undermining Sec. IV's PLR conclusions.","rationale":"The reader's weakest assumption correctly identifies the realization of near-field communication modes via simple DFT beamforming as the load-bearing premise. The paper cites [18] for the mode count and the orthogonality claim but does not verify these for its specific ELAA geometry. If the effective channel vectors H(k)b(k,j) for different j are not nearly orthogonal, then the spatial replicas sent by a single NU are not independent CRA resources; they would interfere at the AP, clustering would merge, and the SIC could not cleanly separate them. This would directly invalidate the central mechanism that distinguishes CSRA from a single-antenna CRA scheme. The proposed concrete test—computing the Gram matrix of the effective channel vectors and the singular-value spectrum of H—would settle whether this premise holds. The additional power-normalization inconsistency (Eq. (1) vs. Eqs. (4)-(5)) is a secondary concern that could affect the absolute PLR values, but it does not change the primary condition. Since the reader's verdict is CONDITIONAL and our identified concern is the same, the verdict should remain unchanged, pending the outcome of the test. This is an honest assessment: the scheme is plausible and the simulations may well be correct, but the core physical assumption is currently unverified.","tokens_in":9399,"tokens_out":18783,"duration_ms":226164,"concrete_test":"For the Sec. IV.A geometry, place a user at y=0, z=0, x=0 (d=8 m). Generate the exact 8000×20 channel matrix H from Eq. (2). (1) Compute singular values of H and count those within 10 dB of the largest; compare to Eq. (3). (2) For each DFT beam vector b_j from Eq. (6) whose angle θ_n from Eq. (7) lies within the AP angular window (arcsin(±10/8)), compute the normalized inner products G_ij = |(H b_i)^H (H b_j)| / (||H b_i|| ||H b_j||). If any off-diagonal |G_ij| > 0.1, the replicas are not independent. (3) Re-run the PLR simulation with the power normalization of Eq. (1) (i.e., divide the sum in Eq. (4) by R) and compare Fig. 2; if PLR degrades substantially, the absolute performance claim is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. III.A-B asserts that the R(k)_max communication modes of Eq. (3), taken from [18], are realized by the DFT steering vectors in Eq. (6), and that these modes are orthogonal in the angular domain, so each NU can transmit R independent replicas per slot. This is the pivotal premise of CSRA: without it, the 'spatial replicas' are just correlated copies that interfere at the ELAA, and the clustering/SIC in Sec. III.C cannot separate them. The paper does not derive or verify this for its geometry (LR=20 m, d=8 m, λ=5 mm, NT=20). In the near-field of the AP, a DFT beam does not produce a single plane wave across the aperture; its footprint is a function of the angular-to-position mapping, and the orthogonality of the codebook in Eq. (6) is a far-field property. Whether the effective channel vectors H(k)b(k,j), j=1..R, are near-orthogonal and each has significant norm (strongly coupled) is never reported. An additional inconsistency: Eq. (1) uses b(k)=1/R sum b(k,j) (implying total power PT/R), while Eqs. (4)-(5) sum b(k,j) without this factor (implying total power R·PT), which could inflate the absolute PLR gains. The load-bearing check is to compute the Gram matrix of the effective channel vectors and the singular-value spectrum of H for the Sec. IV.A scenario.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a grant-free coded random access scheme for near-field communications, termed coded spatial random access (CSRA). Multi-antenna network users transmit multiple spatial replicas of the same packet toward an extremely large aperture array (ELAA) access point, using DFT-based beam steering directions. The AP clusters antenna elements by energy detection per pilot, performs per-cluster MRC decoding, and then runs successive interference cancellation (SIC). The authors argue that near-field communication modes provide multiple orthogonal spatial channels, enabling CRA diversity without channel estimation. Numerical Monte Carlo simulations report packet loss rate (PLR) as a function of the number of active users K and the number of replicas R, comparing CSRA with and without SIC and against a single-element processing variant (CSRA-SE). The claimed contribution is improved reliability and reduced latency for dense uncoordinated IIoT networks.","tokens_in":9797,"tokens_out":11162,"duration_ms":127332,"significance":"The topic is timely, and the protocol architecture is clearly described at an algorithmic level. If the central premise holds—namely, that the R spatial replicas are near-orthogonal, strongly coupled effective channels that act as independent CRA resources—the scheme could be a useful grant-free option for near-field XL-MIMO IIoT. The manuscript gives credit for not fitting any parameter to the PLR curves; the mode-count formula is taken from prior work [18] and is not extracted from the current simulations, so the qualitative trends are not circular. The simulation results are internally plausible (e.g., the existence of an optimal R depending on load). However, the current evidence does not yet establish the main claim: the power normalization is inconsistent, the mode-realization premise is not verified for the simulated geometry, and the chosen baseline does not isolate the benefit of spatial CRA relative to conventional time-domain CRA. The significance of the contribution therefore remains conditional on these points being addressed.","major_comments":[{"comment":"There is a power-normalization inconsistency. Eq. (1) defines b^(k) = (1/R^(k)) Σ_j b^(k,j), so for unit-norm mutually orthogonal DFT beams the total radiated power per NU is PT/R^(k). Eqs. (4) and (5), however, use Σ_j b^(k,j) without the 1/R factor, corresponding to total power R^(k)·PT. The two conventions differ by a factor R^2 in power. Since the PLR curves in Sec. IV depend on SNR, the absolute PLR values may be optimistic. Please state the intended transmit-power constraint and re-run the simulations under a single consistent normalization.","section":"§III.A, Eqs. (1), (4)–(5)"},{"comment":"The load-bearing premise is that the R^(k)_max modes counted by Eq. (3), taken from [18], are realized by the DFT beamformers (6), yielding near-orthogonal and strongly coupled effective channel vectors H^(k)b^(k,j). This is asserted based on [18] but not verified for the geometry used in Sec. IV (LR=20 m, d≈8 m, λ=5 mm, NT=20). The DFT codebook orthogonality in (6) is a far-field property of the NU array; it does not automatically imply orthogonality of the cascade H^(k)b^(k,j) across the finite ELAA. Please report the Gram matrix or singular-value spectrum of the effective channel for representative NU positions, and confirm that the values of R used in the simulations (2–7) correspond to near-orthogonal, strongly coupled replicas. Without this, the interpretation of the R-replica PLR gains as spatial CRA diversity is not supported.","section":"§III.B, Eq. (6)"},{"comment":"The only performance baseline is CSRA-SE, which removes clustering and combining. This comparison conflates the benefit of the proposed spatial-replica CRA with the receive-array gain obtained from cluster MRC. To support the claim that integrating near-field spatial multiplexing with CRA improves reliability, the paper should compare with a time-domain CRA scheme using the same number of replicas and the same total energy per message (e.g., R replicas in R slots with a single beam), or at least with a single-beam, non-spatial CSRA variant. The current results show that array combining helps, but they do not demonstrate that the spatial-mode diversity mechanism itself is beneficial relative to conventional CRA.","section":"§IV.C"},{"comment":"Pilot collisions are not addressed. The protocol lets each active NU draw pilots from a set of size P with P ≪ K, yet the energy detection and clustering in Eqs. (8)–(9) operate per pilot. If two active NUs choose the same pilot and their footprints overlap on the ELAA, the channel estimate in (8) is a superposition, and the cluster is ambiguous. The manuscript does not state the value of P used in the simulations, how same-pilot collisions are resolved before SIC, or whether the reported PLR already includes such collisions. This is a load-bearing detail for the dense, uncoordinated scenario claimed.","section":"§III.C, §IV.A"},{"comment":"For the stated geometry (λ=5 mm, NT=20, LT≈0.05 m, LR=20 m, d between 8 and 8.54 m), Eq. (3) gives Rmax ≈ 15–16, not 20. The text states 'each NU has up to Rmax = 20 orthogonal transmission directions ... leaving each NU with Rmax = 13 usable orthogonal directions.' Please reconcile the mode-count formula with the simulated geometry and report the actual distribution of usable R per NU. Since only R ≤ 7 is simulated, this does not invalidate the PLR curves, but the description of the near-field diversity is internally inconsistent.","section":"§IV.A, Eq. (3)"}],"minor_comments":[{"comment":"The threshold η = 2σ/NP appears dimensionally inconsistent: |\\hat h_r(pj)|² is a squared amplitude, while σ/NP has amplitude units. Please correct the expression (probably η = 2σ²/NP or similar) and include a sensitivity analysis with respect to η, since the clustering result depends on this heuristic threshold.","section":"§III.C.1, Eq. (9)"},{"comment":"The SIC procedure relies on the AP knowing the NU position to identify the set of antenna elements UW. This is an important assumption; it should be stated explicitly in the system model and discussed with respect to positioning accuracy requirements.","section":"§III.C.3"},{"comment":"The assumption of perfect slot synchronization and ideal pilot orthogonality is stated, but its practical impact on the PLR results is not discussed. A sentence on synchronization error tolerance would improve the paper's applicability.","section":"§II, §III.A"},{"comment":"The Monte Carlo simulation parameters do not include the pilot set size P, the number of Monte Carlo runs, or confidence intervals. These should be reported to make the numerical results reproducible.","section":"§IV.A"}],"recommendation":"major_revision","confidential_remarks":"The paper is a system-level protocol paper; the near-field physics is cited rather than derived. The main technical risk is the unverified orthogonality of the effective spatial replicas under the simulated geometry, together with the power-normalization inconsistency. If the authors can supply the Gram-matrix/SVD verification, fix the normalization, and add a time-domain CRA baseline, the contribution could become acceptable. The self-citation [18] is legitimate and not a circularity issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a legitimate new combination—using near-field spatial directions as replicas in coded random access, with clustering and SIC—and the Monte Carlo results look credible. The load-bearing premise, that DFT beams from the NU yield orthogonal effective channel vectors across the ELAA, is not actually verified in the paper, but for their specific parameters it appears to hold: NT=20, LR=4000λ means adjacent DFT beams differ by 0.1 in sine space, which is exactly an integer number of cycles over the 8000-element array. So the stress-test candidate concern about non-orthogonality does not land as fatal, though the authors should state it.\n\nWhat is actually new: the idea of treating near-field spatial beams as CRA replica resources, with cluster-based MRC and SIC at the AP, sidestepping channel estimation. The protocol is fully described, and the simulations show a large improvement over single-element processing and over the no-SIC variant. That is worth having.\n\nSoft spots, in rough order of importance. First, a power inconsistency: Eq (1) defines b(k)=1/R sum b(k,j), implying total transmit power PT, but Eqs (4)-(5) drop that normalisation, implying each replica is sent at full power PT. The paper never says which one the simulations used. If it is the latter, the absolute PLR values are optimistic; the relative comparisons may survive, but the model needs to be made consistent. Second, position knowledge is quietly assumed: the SIC step relies on knowing where each NU is to know which antenna elements its replicas hit. That should be in the system model, not implied later. Third, the latency claim in the abstract is not supported by any metric. Fourth, there is no comparison with a conventional time-domain CRA baseline; the paper only compares against its own variants. That is acceptable for a short paper, but it limits what you can say about the scheme's absolute merit.\n\nOverall, I trust the qualitative result. The paper is not a breakthrough, but it is a plausible incremental contribution with a clear story. I would send it to peer review: a competent reviewer could get a clean revision. Worth citing if you work on CRA or near-field massive access.","headline":"A solid incremental CRA scheme for near-field massive access; the core premise holds for their parameters, but fix power normalization and position assumption.","tokens_in":10209,"tokens_out":8583,"would_cite":true,"duration_ms":106227,"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":"Near-field spatial modes can carry grant-free access without channel estimation.","keywords":["grant-free random access","coded random access","near-field communications","extremely large aperture array","successive interference cancellation","massive machine-type communications","spatial multiplexing","industrial IoT"],"falsifier":"Take a 20-element transmitting ULA and a 20 m, 60 GHz ELAA separated by a distance inside the Fresnel region, and measure the singular-value spectrum of the LOS channel matrix. If fewer than floor(1 + 2 L_T L_R / (lambda sqrt(4 d^2 + L_R^2))) singular values are above the coupling threshold, or if the DFT beams of Eq. (6) have pairwise correlation well above zero, then the independent-replica assumption underlying the simulated packet loss rates is violated.","tokens_in":9380,"feed_emoji":"📡","tokens_out":8472,"duration_ms":84144,"temperature":0.7,"pith_summary":"The paper sets out to show that the parallel spatial channels created when an extremely large aperture array operates in the near field can be used as extra replica resources in a coded random access protocol. It proposes a grant-free scheme, CSRA, in which each active machine transmits copies of its packet along several beam directions, and the access point clusters antenna elements that see the same pilot, combines them, and runs successive interference cancellation. The simulations indicate that this design reaches packet loss rates around 10^-4 with light load and a few times 10^-3 with 45 active users, while processing each antenna element independently cannot reach 10^-2 even with five users. The reason to care is that dense, sporadic industrial traffic could then be handled without scheduling or channel estimation, using only the geometry of the array.","feed_headline":"Near-field beams turn spatial directions into extra packet replicas","feed_subtitle":"Clustering antennas and cancelling interference lets dense, uncoordinated devices transmit at low packet loss.","key_machinery":"The central object is the set of near-field communication modes: parallel LOS channels created by spherical wavefronts between a small user array and a much larger aperture array. Eq. (3) counts them as floor(1 + 2 L_T L_R / (lambda sqrt(4 d^2 + L_R^2))). The paper's mechanism is to realize these modes with DFT beam-steering vectors, so that a user can transmit independent replicas in distinct directions; the AP then clusters contiguous antenna elements per pilot, applies MRC inside each cluster to get array gain, and runs SIC to resolve collisions.","core_discovery":"The paper proposes coded spatial random access (CSRA), a grant-free protocol in which each active network user sends R copies of its packet along R of the near-field communication modes available between its small array and the extremely large aperture array at the access point. Because the wavefront is spherical, the channel matrix has rank larger than one even in pure line-of-sight, and the paper relies on the known result that the number of strongly coupled modes is given by Eq. (3). The AP does not estimate the full channel; it detects pilots per antenna element, clusters contiguous elements that receive the same pilot, combines each cluster with MRC, decodes, and then cancels decoded re","pith_inferences":["If the mode-count premise holds, the same spatial-replica trick should work wherever the Fresnel condition holds, so the scheme's core idea is portable to other frequency bands and array sizes by rescaling Eq. (3).","Because the paper simulates only a fixed geometry, a natural extension is to let users move or tilt, which would change which DFT beams intercept the ELAA and test whether the cluster-to-beam mapping survives.","The slot-synchronous beacon model is a simplifying assumption; an asynchronous or misaligned variant would show how much of the gain depends on the common time reference.","Clusters themselves encode the user's angular position, so a future scheme could exploit cluster location for user identification or localization without extra overhead."],"forward_implications":["Dense, uncoordinated IIoT access can operate grant-free in the near field without per-user channel estimation, as long as the ELAA clusters and combines received signals.","SIC is essential: without it the simulated scheme cannot reach a packet loss rate of 10^-2, so the reliability gain is tied to iterative interference cancellation.","The optimal number of replicas R depends on load: R=4 gives near 10^-4 PLR at K=25, while R=4 or R=5 is better at K=45, suggesting an adaptive replica-count rule.","Clustering and combining at the ELAA is what enables low-PLR operation; single-element processing stays above 10^-2 even for K=5 active users.","The protocol requires fixed user positions to select the beam directions that intercept the ELAA, which fits stationary factory machinery."],"supporting_citations":[{"why":"Supplies Eq. (3), the count of strongly coupled near-field communication modes, and the assertion that simple beam steering realizes them.","marker":"[18]"},{"why":"Provides the DFT beamforming codebook of Eq. (6) used to generate orthogonal transmission directions.","marker":"[19]"},{"why":"Establishes that near-field intelligent surfaces support multiple parallel communication modes in LOS conditions.","marker":"[6]"},{"why":"Defines coded random access with SIC, the protocol family this scheme extends.","marker":"[5]"},{"why":"Shows coded random access with SIC in a cell-free massive MIMO setting, the multi-antenna context this scheme adapts to the near field.","marker":"[4]"},{"why":"Models the ELAA as a radio stripe linear array, the implementation assumed in the simulations.","marker":"[16]"},{"why":"Justifies the near-field/Fresnel propagation regime and its consequences for XL-MIMO.","marker":"[7]"},{"why":"Supports the SIC procedure by showing how the AP derives replica and antenna-element information from a decoded message.","marker":"[20]"}],"fun_headline_variants":["Near-field modes replace grants with multiple packet copies","Coded random access goes near-field for uncoordinated 6G devices","Near-field spatial modes serve as packet replicas in grant-free access","Multiplying packet copies via near-field waves without channel state info","Near-field arrays encode spatial diversity as random-access gain"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that the near-field communication modes counted by Eq. (3) are actually realized by the simple DFT beamforming vectors of Eq. (6), and that those modes are orthogonal and strongly coupled enough to behave as independent replica channels.","fun_headline_variants_meta":{"raw":{"variants":["Near-field modes replace grants with multiple packet copies","Coded random access goes near-field for uncoordinated 6G devices","Near-field spatial modes serve as packet replicas in grant-free access","Multiplying packet copies via near-field waves without channel state info","Near-field arrays encode spatial diversity as random-access gain"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000472,"raw_usage":{"total_tokens":2140,"prompt_tokens":660,"completion_tokens":1480,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":404,"completion_tokens_details":{"reasoning_tokens":1408}},"tokens_in":404,"tokens_out":1480,"duration_ms":12516,"temperature":1.0,"reasoning_tokens":1408,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T17:45:00.827722+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a 20-element transmitting ULA and a 20 m, 60 GHz ELAA separated by a distance inside the Fresnel region, and measure the singular-value spectrum of the LOS channel matrix. If fewer than floor(1 + 2 L_T L_R / (lambda sqrt(4 d^2 + L_R^2))) singular values are above the coupling threshold, or if the DFT beams of Eq. (6) have pairwise correlation well above zero, then the independent-replica assumption underlying the simulated packet loss rates is violated.","supporting_citations":[{"cited_title":"Communication modes with large intelligent surfaces in the near field,","cited_arxiv_id":null,"evidence_quote":"Supplies Eq. (3), the count of strongly coupled near-field communication modes, and the assertion that simple beam steering realizes them."},{"cited_title":"DFT-based beamforming weight- vector codebook design for spatially correlated channels in the unitary precoding aided multiuser downlink,","cited_arxiv_id":null,"evidence_quote":"Provides the DFT beamforming codebook of Eq. (6) used to generate orthogonal transmission directions."},{"cited_title":"Holographic communication using intelligent surfaces,","cited_arxiv_id":null,"evidence_quote":"Establishes that near-field intelligent surfaces support multiple parallel communication modes in LOS conditions."},{"cited_title":"Coded random access: Applying codes on graphs to design random access protocols,","cited_arxiv_id":null,"evidence_quote":"Defines coded random access with SIC, the protocol family this scheme extends."},{"cited_title":"Access point cooperation strategies for coded random access in cell-free massive MIMO,","cited_arxiv_id":null,"evidence_quote":"Shows coded random access with SIC in a cell-free massive MIMO setting, the multi-antenna context this scheme adapts to the near field."},{"cited_title":"Cell-free massive MIMO with radio stripes and sequential uplink processing,","cited_arxiv_id":null,"evidence_quote":"Models the ELAA as a radio stripe linear array, the implementation assumed in the simulations."},{"cited_title":"A tutorial on near-field XL-MIMO communications towards 6G,","cited_arxiv_id":null,"evidence_quote":"Justifies the near-field/Fresnel propagation regime and its consequences for XL-MIMO."},{"cited_title":"Irregular repetition slotted ALOHA in an information-theoretic setting,","cited_arxiv_id":null,"evidence_quote":"Supports the SIC procedure by showing how the AP derives replica and antenna-element information from a decoded message."}],"review_version":1}