{"id":"f0cecf64-1e18-4c0a-80b6-2a14e71c31f3","arxiv_id":"2608.09179","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A dynamic port-reconfigurable near-field channel model and a greedy subarray partition scheme enable fluid-antenna UAV-to-ground links to track high-gain ports with lower complexity.","lead":"This paper introduces a dynamic near-field channel model for drones with fluid antenna systems, where antenna ports can be switched on and off, and pairs it with a fast greedy algorithm for grouping active ports. It matters because drone-to-ground links often operate in the near-field regime, where standard far-field channel models are inaccurate and can hurt reliability.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Equation (3) collapses the planar FAS port grid onto a single ray, so the model cannot support the claimed 2D near-field spatial distribution.","rationale":"I read the paper in good faith. The greedy partition algorithm, dynamic update procedure, and complexity accounting are clearly described, and the idea of grouping FAS ports under a Rayleigh-distance constraint is reasonable. However, the single most load-bearing element of the central claim is the port position model in Equation (3), and that equation is internally inconsistent with the stated planar FAS geometry. The reader's weakest_assumption focuses on whether the subarray planar-wave approximation is validated against an exact spherical-wave model. That is a legitimate concern, but it is secondary: even a perfect error bound for the subarray approximation would not help if the ports themselves are not located on a planar aperture. Since k_(p_h,p_v) in Equation (3) is a nonnegative norm, all port offsets point along one ray from the array center, so the 2D near-field spatial non-uniformity that the paper claims to characterize cannot exist in the model. Figures 4-13 and the associated accuracy and capacity conclusions are therefore not evidence for a planar FAS. If the authors' code uses the correct two-dimensional position formula, the text should be corrected and the simulations re-verified; but as submitted, the manuscript's model statement does not support the abstract's claim. I would therefore move the verdict from CONDITIONAL to REJECT for the current version, while acknowledging the framework might be salvageable with a corrected position model.","tokens_in":19551,"tokens_out":15501,"duration_ms":144138,"concrete_test":"Compute the nine port offset vectors from Equation (3) for P_h = P_v = 3, Delta d_h = Delta d_v = lambda/2, psi_ver^T = psi_azi^T = 0, and form the 3 by 9 matrix of coordinates. A correct planar FAS model would give a rank-2 Gram matrix with signed coordinates spanning both sides of the center; Equation (3) gives rank 1, with all non-center ports on the positive x-ray and the four corners coincident. If the rank is 1, Equation (3) must be replaced by a proper two-basis-vector formula such as d = k_p_h Delta d_h e_h + k_p_v Delta d_v e_v, and all numerical results in Section IV must be re-run before the central claim can be assessed.","verdict_should_be":"REJECT","load_bearing_attack":"Section II, Equation (3) defines the position of port (p_h,p_v) relative to the FAS center as d_(p_h,p_v),T = k_(p_h,p_v) * [cos(psi_ver^T) cos(psi_azi^T), cos(psi_ver^T) sin(psi_azi^T), sin(psi_ver^T)]^T, where k_(p_h,p_v) = sqrt(k_p_h^2 Delta d_h^2 + k_p_v^2 Delta d_v^2). Because k_(p_h,p_v) is a nonnegative scalar norm, every port offset is a positive multiple of the same unit vector. The intended L_h by L_v planar port grid therefore collapses onto a single ray from the array center: ports with equal radial distance coincide, and no port lies on the opposite side of the center or in a perpendicular direction. For example, with P_h = P_v = 3 and Delta d_h = Delta d_v = lambda/2, psi_ver^T = psi_azi^T = 0, the four corner ports all map to the same point on the positive x-axis, and the nine ports occupy at most three distinct positions, all on one ray. This is an internal inconsistency in the model itself, not a subarray approximation issue. The non-uniform 2D spatial distribution claimed in the abstract and used in Figures 4-13 cannot be produced by Equation (3). As written, the central claim is unsupported unless the implementation uses a different, uncited position formula.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a dynamic port-reconfigurable near-field channel model for FAS-assisted UAV-to-mobile-user links, combining LoS/NLoS decomposition with UAV motion and a binary port activation matrix. It also introduces a greedy subarray partition scheme with a dynamic update algorithm to reduce computational complexity, a maximum-gain port selection strategy, a modeling-accuracy metric against a UPA baseline, and a MIMO-style capacity analysis. Numerical results are presented for modeling error and channel capacity under varying FAS dimensions, port spacing, active port ratios, and UAV dynamics.","tokens_in":19852,"tokens_out":3496,"duration_ms":32713,"significance":"If the model were geometrically sound, the paper would offer a useful simulation and design tool for FAS-enabled UAV near-field links, and the greedy subarray scheme with explicit complexity analysis (O(PhPv + Pact^2) partition, O(epsilon) updates) is a practical contribution. The derivation of capacity formulas and the port selection strategy are standard but clearly presented. However, the central geometric definition in Eq. (3) collapses the intended planar port grid onto a single ray, which invalidates the claimed 2D non-uniform spatial distribution and calls into question all numerical results built on it. The modeling-accuracy metric in Eq. (33) is also internal to the modeling assumptions rather than a comparison to an exact spherical-wave model or measurements, so the central 'accurate characterization' claim is not established by the evidence provided.","major_comments":[{"comment":"Equation (3) defines the port position as d_(ph,pv),T = k_ph,pv times a unit vector, where k_ph,pv = sqrt(k_ph^2 Delta d_h^2 + k_pv^2 Delta d_v^2) is a nonnegative scalar. Since every port offset is therefore a nonnegative multiple of the same unit vector, all Ph by Pv ports collapse onto a single ray from the FAS center; ports with equal k_ph,pv coincide, and no port lies on the opposite side of the center or in a perpendicular direction. For example, with P_h = P_v = 3 and zero orientation angles, the four corner ports all map to the same point on the positive x-axis, and the nine ports reduce to at most three distinct points. This is an internal inconsistency in the model, not a subarray approximation issue. The subsequent channel expressions, including the LoS CIR in Eq. (20), the NLoS CIR in Eq. (23), and the vectorized response in Eqs. (37)-(39), use separate horizontal and vertical offsets and therefore assume a genuine 2D planar array. As written, the claimed non-uniform 2D spatial distribution in the abstract and in Figures 4-13 cannot be produced by the model. This needs to be fixed by replacing Eq. (3) with a proper planar position vector, e.g., d_(ph,pv),T = [k_ph Delta d_h, k_pv Delta d_v, 0]^T in the FAS local frame before applying orientation rotations, and all subsequent results must be regenerated.","section":"Sec. II, Eq. (3)"},{"comment":"The modeling accuracy metric Delta in Eq. (33) compares the proposed FAS channel to a UPA channel model generated under the same assumptions and with the same subarray partition procedure. This does not measure accuracy against an exact spherical-wave model or against measured channels; it measures the difference between two model variants that share the same approximations. Consequently, the claim in the abstract that 'the proposed model accurately characterizes the non-uniform spatial distribution of near-field channels' is not supported by the reported modeling-error results. The authors should either compare against an exact per-port spherical-wave model (as a ground truth) or against measured data, and should provide an error bound, perhaps in terms of the maximum phase error across the subarray aperture, to justify the accuracy-complexity tradeoff.","section":"Sec. III-B, Eq. (33)"},{"comment":"The load-bearing premise of the subarray partition is that all ports within a subarray can share one distance and one angle parameter set because the subarray aperture satisfies the Rayleigh-distance bound D_sub < sqrt(lambda * xi / 2). However, this is asserted without an error bound on the intra-subarray phase variation. The Rayleigh distance is a heuristic far-field boundary, not a guarantee that the planar-wave approximation is accurate for the CIR phase terms in Eq. (20). For the claimed 'limited accuracy loss' relative to per-port spherical-wave processing, a quantitative bound is needed, e.g., bounding the maximum phase error across the subarray aperture as a function of D_sub, wavelength, and distance, and showing that this bound is small for the scenarios simulated. Without such a bound, the complexity-accuracy tradeoff is not demonstrated.","section":"Sec. II-A, Eq. (10) and Sec. II-B, Eq. (20)"},{"comment":"Because all numerical results are generated from the collapsed geometry in Eq. (3), the channel-gain distributions in Figures 4, the active-port patterns in Figure 5, and the modeling-error and capacity curves in Figures 6-13 do not correspond to any physical planar FAS. Even if the geometry is corrected, the simulations must be rerun, and the qualitative conclusions (e.g., dependence on port spacing and active-port ratio) may change. The current set of results therefore cannot be used to validate the model's claims.","section":"Sec. IV, Figs. 4-13"}],"minor_comments":[{"comment":"The denominator of Eq. (22) uses (d^G_psub_sub,x(t) - d_p,x(t))^2, but d_p,x(t) is not defined in the context of a subarray; it should presumably be the subarray center coordinate. Please clarify the notation.","section":"Sec. II-B, Eq. (22)"},{"comment":"Algorithm 2 updates the subarray center when a new port is added, but the description does not specify how the center is recomputed (e.g., average of member ports or first port). This should be stated for reproducibility.","section":"Sec. II-A, Algorithm 2"},{"comment":"Reference [23] lists 'IEEE Wireless Commun. Lett.' twice in the same entry, and the venue information for [11] and [46] should be checked for duplicated journal names.","section":"References"},{"comment":"Please correct typographical issues such as 'the the' in the text after Eq. (15), 'UA V' spacing artifacts, and 'the the components' in the lines following Eq. (22).","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper's scope is appropriate for the journal, and the algorithmic complexity analysis is a genuine contribution. However, the geometric collapse in Eq. (3) is a fundamental modeling error that invalidates the numerical studies as presented. I recommended major revision rather than reject because the error is fixable by correcting the port position formula and rerunning the simulations, but the authors should be asked to either provide a corrected and fully regenerated paper or clearly state that the current results are based on a non-planar one-dimensional port layout. I also note that the modeling-accuracy metric is inherently circular and should be supplemented with an external benchmark before the central claims can be accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe short version: the central channel model is broken as written. Equation (3) defines each port's position as a nonnegative scalar k_{ph,pv} times a single unit vector, which forces the entire Ph×Pv planar port grid onto one ray from the array center. For a 3×3 FAS, the four corners map to the same point; no port can sit on the opposite side or in a perpendicular direction. Equation (20) and the appendix CIR formulas later use separate horizontal and vertical offsets, so Eq. (3) contradicts the paper's own channel equations. This is not a subarray approximation issue; it is an internal inconsistency in the geometry, and it invalidates the abstract's claim that the model captures the non-uniform 2D near-field spatial distribution.\n\nThe genuinely new piece is the greedy, dynamic subarray partition for sparsely activated FAS ports. An O(ε) per-update cost after O(PhPv + Pact²) construction is a practical idea, and the complexity decomposition is clearly presented. The gain-based multi-port selection is standard but sensibly extended to the multi-active-port case. The authors are also honest that the greedy partition is not globally optimal and that mutual coupling is deliberately excluded.\n\nSoft spots beyond Eq. (3): the accuracy metric in Eq. (33) compares the proposed FAS model to a UPA model built under the same assumptions and partitioned the same way. That cannot establish accuracy against an exact spherical-wave model or against measurements; it only shows internal consistency. The capacity analysis reuses the same generated channels, so there are no independently falsifiable predictions. Several smaller notation slips exist (e.g., a ψazi/ψver mix-up in Eq. (23)) but those are fixable.\n\nThe Eq. (3) problem is load-bearing. If the intended formula was the standard planar offset vector, the text needs a corrected equation and the simulations should be rerun with it; if the collapsed geometry is actually what was simulated, then the 2D results are invalid. Either way, the paper is not publishable as-is.\n\nThis paper is for researchers working on FAS-aided UAV links and near-field subarray methods; the algorithmic idea is worth knowing, but only after the geometry is fixed and validation is strengthened. I would send it to peer review with a clear demand to fix Eq. (3) and add an exact spherical-wave or measurement comparison. As written, I would not cite it.","headline":"Equation (3) collapses the planar FAS onto a single ray, making the central model internally inconsistent; the subarray-partition idea is worth pursuing but the paper needs major revision.","tokens_in":20403,"tokens_out":3921,"would_cite":false,"duration_ms":34346,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94A05","94A40"],"pacs":[],"model":"deepseek-v4-flash","headline":"A greedy subarray partition makes near-field FAS-UAV channel modeling lightweight without losing non-uniform spatial accuracy.","keywords":["fluid antenna system","near-field communication","channel modeling","subarray partition","UAV communications","channel capacity","port selection","modeling accuracy"],"falsifier":"At the paper's simulation settings ($f_c = 5$ GHz, $H_0 = 20$ m, $D_0 = 60$ m, 50$\\times$50 FAS with $\\Delta d = 2/5\\lambda$ and 3/5 active ports), compute the normalized modeling error $\\Delta$ of (33) against an exact per-port spherical-wave channel for a sweep of $D_{\\mathrm{sub}}$ values. If the error does not shrink toward zero as $D_{\\mathrm{sub}}$ approaches one-port size and does not stay within a few decibels at the working $D_{\\mathrm{sub}}$, the shared-parameter assumption collapses.","tokens_in":19326,"feed_emoji":"🛸","tokens_out":8977,"duration_ms":149547,"temperature":0.7,"pith_summary":"Fluid antenna systems (FASs) let a UAV's antenna reconfigure which of many dense ports is active, but air-to-ground links between a UAV and a ground user often sit in the near field, where distance and angle differ from port to port and the spatial signal distribution is strongly non-uniform. This paper claims that a greedy subarray partition can tame that complexity: active ports are grouped by an aperture threshold derived from the Rayleigh distance, so all ports in a subarray share one distance and one angle parameter set, and the complex impulse response is computed once per group. The accompanying dynamic update absorbs newly activated ports with $O(\\varepsilon)$ distance checks after an initial $O(P_h P_v + P_{\\mathrm{act}}^2)$ partition, replacing per-port spherical-wave recomputation. The paper also adds a channel-gain-based active port selection that avoids deep-fading ports, and it derives modeling accuracy against a uniform planar array baseline plus MIMO channel capacity. If the claim holds, the model provides a lightweight, real-time-capable tool for designing FAS-enabled UAV air-to-ground links in 6G scenarios.","feed_headline":"Fluid antenna UAV links get a fast near-field model","feed_subtitle":"Greedy subarray grouping shares one distance and angle across ports, cutting per-update cost to O(ε).","key_machinery":"The load-bearing object is the greedy subarray partition with aperture threshold $D_{\\mathrm{sub}}$. Active ports are grouped so that any two ports in a subarray are separated by no more than $D_{\\mathrm{sub}}$, with $D_{\\mathrm{sub}}$ chosen below the maximum satisfying $D^{\\mathrm{sub}}_{\\max} < \\sqrt{\\lambda \\xi_{\\mathrm{sub},R}/2}$, the Rayleigh-distance condition that keeps the receiver in the far field of every subarray. This turns per-port spherical-wave phase computations into one shared distance and angle parameter set per subarray, so the CIR is evaluated once per group, the partition costs $O(P_h P_v + P_{\\mathrm{act}}^2)$, and a port switch only triggers $O(\\varepsilon)$ distance checks against existing subarray centers.","core_discovery":"The paper's central claim is that a planar FAS on a moving UAV can be modeled with near-field accuracy without computing a spherical-wave response for every active port. The channel is built as a sum of LoS and NLoS components, with a binary port-state matrix tracking which ports are active and UAV motion kinematics updating all geometry with time. The key move is to partition active ports into subarrays whose aperture satisfies the Rayleigh-distance bound $D_{\\mathrm{sub}} < \\sqrt{\\lambda \\xi_{\\mathrm{sub},R}/2}$, so that within each subarray the planar-wave approximation holds and all ports share one representative distance and one set of departure/arrival angles. A greedy algorithm forms these subarrays, and a dynamic update rule lets a newly activated port join or create a subarray in $O(\\varepsilon)$ time. Ports are chosen by maximum channel gain to skip deep fades, and the paper derives the normalized modeling error against a UPA baseline and the MIMO channel capacity. The reported simulations show the partitioned model tracks the spatially non-uniform near-field channel and that the partition induces only limited capacity loss.","pith_inferences":["Testable extension: comparing the proposed model against an exact per-port spherical-wave simulation, rather than the UPA baseline used in the paper, would quantify the true error of the shared-parameter assumption and validate where the accuracy boundary at the chosen $D_{\\mathrm{sub}}$ holds.","The greedy partition is scanning-order dependent; choosing the center port more cleverly could reduce the number of subarrays $\\varepsilon$ further, cutting the dynamic update cost without changing $D_{\\mathrm{sub}}$.","Connecting neighbouring problems: the same subarray philosophy could be applied to FAS receiver arrays or to reconfigurable intelligent surfaces, where sparse activation makes fixed mechanical partitions inapplicable.","A natural extension is to let $D_{\\mathrm{sub}}$ adapt to the instantaneous UAV-to-user distance, giving a seamless handover between near-field and far-field regimes as the UAV flies."],"forward_implications":["Port reconfiguration on a UAV FAS becomes fast enough for real-time tracking: each port switch costs $O(\\varepsilon)$ rather than a full per-port spherical-wave recomputation.","The threshold $D_{\\mathrm{sub}}$ gives designers an explicit accuracy–complexity dial, with smaller subarrays approaching the spherical-wave model.","The model makes the non-uniform near-field gain landscape of FAS visible, so capacity and diversity can be optimized over FAS physical parameters such as port spacing and activation ratio.","The channel-gain selective activation predicts that a moderate active-port ratio (e.g., 3/5) beats all-port activation, because all-port activation includes deep-fading positions.","Because the subarrays simply carry planar-wave phase while the overall geometry stays time-varying, the framework extends naturally to other mobile air-to-ground platforms beyond UAVs."],"supporting_citations":[{"why":"Defines fluid antenna systems and the port-reconfiguration premise this paper builds on.","marker":"[10]"},{"why":"Prior dynamic channel modeling for FAS in UAV communications; the baseline this work extends to near-field.","marker":"[11]"},{"why":"Supplies the near-field/far-field distinction and the Rayleigh-distance boundary for UAV links.","marker":"[31]"},{"why":"Tutorial on near-field communications giving the non-uniform spatial channel properties the model targets.","marker":"[33]"},{"why":"Introduces subarray partition for near-field air-ground RIS modeling, which the paper adapts to dynamic FAS ports.","marker":"[37]"},{"why":"Mechanical subarray partition for large-scale MIMO, the fixed-partition baseline the proposed greedy scheme improves on.","marker":"[38]"},{"why":"Reports microsecond-level FAS reconfiguration used to argue that the dynamic update overhead is not dominant.","marker":"[44]"}],"fun_headline_variants":["Near-field fluid antenna UAV links get a fast model","Greedy subarray trick speeds up fluid antenna UAV links","UAV fluid antenna near-field model cuts per-update cost","Greedy partition makes fluid-antenna UAV near-field practical"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the planar-wave approximation inside each subarray is accurate enough for one distance and one angle to represent all ports in the group; the paper gives no bound on the phase error this introduces across the subarray aperture.","fun_headline_variants_meta":{"raw":{"variants":["Near-field fluid antenna UAV links get a fast model","Greedy subarray trick speeds up fluid antenna UAV links","UAV fluid antenna near-field model cuts per-update cost","Greedy partition makes fluid-antenna UAV near-field practical"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00091,"raw_usage":{"total_tokens":3953,"prompt_tokens":1028,"completion_tokens":2925,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":644,"completion_tokens_details":{"reasoning_tokens":2855}},"tokens_in":644,"tokens_out":2925,"duration_ms":34331,"temperature":1.0,"reasoning_tokens":2855,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:56:23.697155+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"At the paper's simulation settings ($f_c = 5$ GHz, $H_0 = 20$ m, $D_0 = 60$ m, 50$\\times$50 FAS with $\\Delta d = 2/5\\lambda$ and 3/5 active ports), compute the normalized modeling error $\\Delta$ of (33) against an exact per-port spherical-wave channel for a sweep of $D_{\\mathrm{sub}}$ values. If the error does not shrink toward zero as $D_{\\mathrm{sub}}$ approaches one-port size and does not stay within a few decibels at the working $D_{\\mathrm{sub}}$, the shared-parameter assumption collapses.","supporting_citations":[{"cited_title":"Fluid antenna systems,","cited_arxiv_id":null,"evidence_quote":"Defines fluid antenna systems and the port-reconfiguration premise this paper builds on."},{"cited_title":"Dynamic channel modeling of fluid antenna systems in UA V communications,","cited_arxiv_id":null,"evidence_quote":"Prior dynamic channel modeling for FAS in UAV communications; the baseline this work extends to near-field."},{"cited_title":"Cross far- and near-field wireless communications in terahertz ultra-large antenna array systems,","cited_arxiv_id":null,"evidence_quote":"Supplies the near-field/far-field distinction and the Rayleigh-distance boundary for UAV links."},{"cited_title":"Near-field communications: A tutorial review,","cited_arxiv_id":null,"evidence_quote":"Tutorial on near-field communications giving the non-uniform spatial channel properties the model targets."},{"cited_title":"Large-scale RIS enabled air-ground channels: Near- field modeling and analysis,","cited_arxiv_id":null,"evidence_quote":"Introduces subarray partition for near-field air-ground RIS modeling, which the paper adapts to dynamic FAS ports."},{"cited_title":"High-efficient near-field channel characteristics analy sis for large-scale MIMO communication systems,","cited_arxiv_id":null,"evidence_quote":"Mechanical subarray partition for large-scale MIMO, the fixed-partition baseline the proposed greedy scheme improves on."},{"cited_title":"Meta fluid antenna: Architecture design, performance analysis, experimental examination","cited_arxiv_id":null,"evidence_quote":"Reports microsecond-level FAS reconfiguration used to argue that the dynamic update overhead is not dominant."}],"review_version":1}