{"id":"a81e6e13-7f86-48d3-8a9a-e62b65f2a38a","arxiv_id":"2505.19902","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A fixed multi-pinching-antenna downlink is modeled as a finite-impulse-response channel, and an OFDMA max-min subcarrier and power allocation is shown by simulation to raise the worst-user rate over single-carrier baselines.","lead":"This paper shows that when multiple pinching antennas along a leaky waveguide serve several users at once, the channel becomes frequency-selective and interferes with itself, and it proposes an OFDMA scheme with a low-complexity allocation algorithm to restore high minimum user rates. A smart generalist might read it because pinching antennas are a low-cost candidate for indoor mmWave coverage, and this work tackles the main practical obstacle to serving many users at once.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The numerical OFDMA gains rest on an unmeasured hardware idealization: unit-gain phase-only waveguide taps (Eq. 3) and a 1/N uniform power split (Eq. 10); real leaky-waveguide attenuation or unequal pinch coupling would change the FIR taps and could shrink the reported gains.","rationale":"The reader's conditional verdict is appropriate. I looked for a more internal break in the OFDMA argument and did not find one: from the FIR model in Eq. (4), the frequency response in Eq. (7), and the stated OFDMA orthogonality constraints, the subcarrier-level rate expression is self-consistent, and the greedy allocation plus water-filling is a plausible heuristic even though \"near-optimal\" is not demonstrated against an optimum. The place where the central numerical claim is least secure is the step between the physical leaky waveguide and the channel used in simulation. A phase-only waveguide tap (Eq. 3) combined with a forced 1/N power split (Eq. 10) is an energy-conserving idealization only if each PA radiates exactly 1/N of the input power with no propagation loss; real passive pinch apertures on a dielectric waveguide have a coupling profile and attenuation. Because every reported OFDMA-vs-TDMA margin in Figs. 2 and 3 is generated under this idealization, the hardware assumption is load-bearing. This is the same weakest assumption the reader identified, so I agree. The proposed sensitivity test would settle whether realistic loss profiles change the headline conclusion; until then CONDITIONAL remains the right verdict.","tokens_in":7916,"tokens_out":7223,"duration_ms":83363,"concrete_test":"Re-run the simulations of Figs. 2 and 3 with physically motivated tap amplitudes: replace h_{0,n} in Eq. (4) by sqrt(L_n) exp(-j 2π x_n/λ_g) with x_n = nD_x/(N+1), where L_n = exp(-α x_n) models waveguide/leakage loss, and normalize the total extracted power to P_t (so the 1/N factor in Eq. (10) becomes 1/Σ_n L_n). Sweep α over 0.1, 0.5, and 1 dB/m, and additionally test an empirically measured coupling profile if available. If at N=30 the OFDMA minimum-rate advantage over the TDMA/SC-FDE benchmark shrinks by more than roughly 25%, or if the CP/T_FFT design must be revised, the equal-amplitude assumption is load-bearing for the headline claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. (4) sets the waveguide contribution to h_{0,n} = exp(-j 2π ||ψ0-ψP_n||/λ_g), a unit-magnitude phase, and Eq. (10) inserts an extra 1/N in the SNR denominator because \"the BS spreads the power P_{m,k} uniformly across the N PAs.\" These two choices jointly fix the equal-amplitude, N-tap frequency-selective channel that produces Figs. 2–3. A passive leaky waveguide fed at one end does not deliver equal power to every pinch: each aperture couples out a fraction of the traveling wave, and the remaining guided power decays along the guide, so the n-th tap in Eq. (4) should carry a real amplitude sqrt(L_n) with L_n decreasing in distance from the feed (waveguide attenuation and coupling profile). Once L_n is included, the effective number of significant taps drops, the RMS delay spread and coherence bandwidth used to set T_CP and Δf change, and the SNR in Eq. (10) is no longer |H|²/(N N0 Δf). The paper provides no measurement, no leakage model, and no citation for this hardware assumption. Since the headline rate comparison is computed entirely under the idealized equal-tap model, the central gain claim is not yet tied to a physical PA implementation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers a downlink from a base station feeding a leaky waveguide with N fixed pinching antennas to M single-antenna users. The authors model the resulting channel as an N-tap finite impulse response filter (Eq. 4), characterize its frequency selectivity, and propose an OFDMA framework with cyclic-prefix and FFT sizing. They formulate a max-min rate-fairness problem (P1) and solve it with a two-stage heuristic: greedy subcarrier assignment followed by per-user water-filling (Algorithm 1). Simulations compare the proposed scheme against TDMA with single-carrier frequency-domain equalization and against a single-PA baseline, reporting higher minimum user rates across various N, M, blockage, and power settings.","tokens_in":8142,"tokens_out":9319,"duration_ms":99706,"significance":"If the channel model is accepted, the paper gives a clear, low-complexity solution to the ISI problem that arises when fixed, spatially separated PAs serve multiple users, and it demonstrates the value of frequency diversity in PA systems. The derivation from geometry to Eq. (7) is explicit, the problem formulation is standard, and Algorithm 1 is easy to implement. However, the numerical gains rest on two unverified hardware idealizations: unit-magnitude waveguide taps and a uniform 1/N power split. The 'near-optimal' fairness claim is also not benchmarked against an optimal solver or an upper bound. These issues make the qualitative idea credible but the quantitative claims conditional.","major_comments":[{"comment":"The channel model assumes that the waveguide contribution h_{0,n} is a unit-magnitude phase and that the BS spreads each subcarrier's power uniformly across the N PAs, producing the factor 1/N in the SNR denominator of Eq. (10). A physical leaky waveguide radiates a fraction of the guided power at each aperture, so the tap amplitudes in Eq. (4) should generally decrease with distance from the feed; no measurement, coupling model, or citation is given for equal-amplitude taps. Since the CP length, coherence bandwidth, and the rate comparisons in Figs. 2-3 all depend on these tap amplitudes, the headline gain over TDMA is not yet tied to a physical implementation. Please add a waveguide attenuation/coupling model and re-evaluate the results, or justify equal amplitudes with a measurement.","section":"II, Eq. (3); III, Eq. (10)"},{"comment":"The claim that Algorithm 1 achieves 'near-optimal' max-min fairness is not supported. P1 is an NP-hard mixed-integer nonlinear program, and the paper provides no comparison with an optimal solver for small problem sizes, no upper bound from a relaxation, and no approximation guarantee. I recommend adding an exhaustive-search comparison for small values of K and M, or a relaxation-based upper bound, to quantify the optimality gap before using the phrase 'near-optimal'.","section":"Abstract; III, P1 and Algorithm 1"},{"comment":"The chosen CP/FFT parameters are not mutually consistent with OFDM orthogonality. The paper sets T_FFT = T_CP + 1/B_c and then chooses K = 2^{floor(log2(B T_FFT))}, so that Delta_f = B/K. For standard OFDM, Delta_f must equal 1/T_FFT, which holds only if K = B T_FFT exactly; with the floor operation this is generally false. The inconsistency affects the efficiency factor eta = T_FFT/(T_FFT+T_CP) and the numerical rate results. Please redefine the design so that T_FFT = K/B (or Delta_f = 1/T_FFT) and update the simulations accordingly.","section":"III, CP/FFT sizing paragraph"},{"comment":"The per-user water-filling with a fixed budget P_t/M does not necessarily solve the max-min problem P1. The greedy stage assigns subcarriers based on provisional rates computed with equal power per subcarrier, but after stage 2 the actual user rates depend on the final power allocation, which can change the ordering of users; no guarantee or final max-min check is provided. Either the algorithm should be modified to iterate between assignment and power allocation, or the claim should be weakened to a heuristic that improves fairness rather than achieving near-optimal max-min fairness.","section":"III, Algorithm 1, Stage 2"}],"minor_comments":[{"comment":"The composite delay tau_{m,n} is used implicitly in Eq. (4) but only defined later in the CP/FFT paragraph; please define it before first use.","section":"II, Eq. (4) and following text"},{"comment":"The phrase 'the m-th use' should be corrected to 'the m-th user'.","section":"III, text after Eq. (8)"},{"comment":"The formula K = 2^{floor(log2(B T_FFT))} selects the largest power of two not exceeding B T_FFT, not the 'next power of two'; please reword to avoid ambiguity.","section":"III, CP/FFT sizing paragraph"},{"comment":"The caption 'Minimum data rate versus number of PA' should read 'number of PAs'.","section":"IV, Fig. 2 caption"},{"comment":"The notation for the power variable is introduced as p_{m,k} in Eq. (9) but the sentence before Eq. (10) refers to P_{m,k}; please unify the notation.","section":"III, Eq. (10) and surrounding text"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the scope of eess.SP and the topic is timely. I see no novelty concern, but the quantitative contributions are conditional on the idealized waveguide model and the unverified near-optimality claim; both are fixable within the manuscript's scope through additional modeling and small-scale optimality benchmarks. The paper relies on recent same-group works for the channel and blockage model, which is acceptable but should be acknowledged more explicitly if the waveguide model is not changed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is the first paper I've seen that treats a fixed multi-PA downlink as a frequency-selective FIR channel and builds OFDMA on it. That is a legitimate step for the PA literature, and the geometry-to-FIR derivation (Eq. 4 to Eq. 7) is clean. The CP/FFT sizing follows standard coherence-bandwidth reasoning, and the two-stage greedy assignment plus per-user water-filling is a sensible way to attack the max-min MINLP. The simulations directionally support the claim that OFDMA beats single-PA and TDMA/SC-FDE baselines, increasingly so with more PAs. Credit where due: no parameter is fitted to produce the gains; the model is constructed from geometry and standard OFDMA machinery. Soft spots, in order of importance. First, the hardware assumption in Eqs. (3) and (10) is the load-bearing one. The waveguide contribution is modeled as a unit-magnitude phase, and the SNR denominator carries a 1/N from 'the BS spreads the power uniformly across the N PAs.' A real leaky waveguide loses power along its length, and pinch coupling is not uniform, so the FIR taps should have real, decreasing amplitudes. That changes the delay spread, coherence bandwidth, CP length, and the absolute rates. I don't think this sinks the central idea - unequal taps still give you a frequency-selective channel - but it could shrink the simulated advantage, and the paper offers no measurement, leakage model, or citation to justify it. The authors should either add a loss model or show sensitivity. Second, the 'near-optimal' claim is never checked. No comparison to an optimal or even a good upper bound for the max-min problem, so that word is doing work without evidence. Third, the benchmark formula in Eq. (12) is garbled (the parentheses don't parse), and no code or data is released. Both are fixable but matter for reproducibility. Fourth, the paper leans on same-group references for the channel and blockage model. That's not disqualifying, but it makes the unverified part of the model more visible. Net: the central argument holds up as a modeling-and-simulation contribution, and the limitations are addressable. I'd send this to peer review and ask for a leakage-aware channel model or sensitivity analysis, an optimality check, and cleaned-up benchmarks. A serious referee can turn this into a solid paper.","headline":"First clean OFDMA treatment of fixed multi-pinching-antenna downlinks, with real simulation gains, but the headline numbers rest on an unverified equal-amplitude waveguide model.","tokens_in":743,"tokens_out":832,"would_cite":true,"duration_ms":29755,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A fixed multi-pinching-antenna downlink is a frequency-selective FIR channel, and OFDMA with greedy subcarrier assignment and water-filling raises the worst user's data rate well above single-carrier scheduling.","keywords":["pinching antennas","OFDMA","frequency-selective channel","finite impulse response","max-min fairness","resource allocation","millimeter wave","leaky waveguide"],"falsifier":"Measure the magnitude of the frequency response $|H_m(f)|$ on a real 30 m leaky waveguide with $N$ evenly spaced pinch points at 28 GHz and compare the measured tap amplitudes and coherence bandwidth with Eq. (7); if the taps are unequal because of waveguide loss or coupling variation, the predicted OFDMA-over-TDMA minimum-rate gap should narrow with $N$, and the simulation curves would not be reproduced.","tokens_in":7625,"feed_emoji":"📡","tokens_out":8362,"duration_ms":74983,"temperature":0.7,"pith_summary":"Pinching-antenna systems route millimeter-wave signals through a leaky waveguide and radiate them at many small apertures, and this paper studies the practical case where those apertures are fixed at evenly spaced positions while several indoor users are served at once. In that setting every user receives multiple delayed copies of its symbol, one per aperture, so the downlink behaves like an $N$-tap finite-impulse-response channel whose delay spread at 28 GHz can exceed a 500 MHz symbol duration by an order of magnitude. The paper's claim is that this strong frequency selectivity is not a nuisance to be equalized away but a resource: an orthogonal frequency-division multiple access (OFDMA) frame with a cyclic prefix, a greedy max-min subcarrier assignment, and per-user water-filling raises the worst user's rate substantially above time-division single-carrier baselines. If correct, dense multi-user pinching-antenna coverage can work with fixed aperture positions and no mechanical steering.","feed_headline":"OFDMA lifts worst-user rate in multi-pinching-antenna downlinks","feed_subtitle":"Fixed pinch points create severe inter-symbol interference; splitting subcarriers across users turns the distortion into a fairness gain.","key_machinery":"The load-bearing object is the FIR channel model in Eq. (4) and its Fourier transform in Eq. (7), in which the $n$-th tap has amplitude $h_{0,n}h_{m,n}$, where $h_{0,n}$ is a unit-magnitude phase from the waveguide feed to the $n$-th aperture and $h_{m,n}$ is the free-space path from that aperture to user $m$, located at the composite delay $\\tau_{m,n} = \\|\\psi_m-\\psi_P^n\\|/(\\lambda f_c) + \\|\\psi_0-\\psi_P^n\\|/(\\lambda_g f_c)$. These taps are what make the channel frequency-selective and set the cyclic-prefix length (from the worst excess delay) and the FFT size (from the RMS delay spread via the coherence-bandwidth rule of thumb). The framework then rides on the OFDMA formulation in problem (P1), a max-min mixed-integer program whose binary subcarrier variables make it NP-hard, and on the two-stage heuristic--greedy subcarrier assignment maximizing $\\Gamma_{m,k}=|H_{m,k}|^2/\\max_{q\\neq m}|H_{q,k}|^2$ for the least-served user, then classical water-filling per user--that keeps polynomial complexity while exploiting the ripples.","core_discovery":"The central discovery is that a fixed, uniformly placed multi-pinching-antenna downlink has an unavoidable, strongly frequency-selective channel: the impulse response $h_m(\\tau)$ in Eq. (4) is a sum of $N$ delayed replicas, each combining a guided delay through the waveguide and a free-space delay, and the transfer function $H_m(f)$ in Eq. (7) shows corresponding spectral ripples. With apertures meters apart, the composite delay spread is tens of nanoseconds, versus a 2 ns symbol at 500 MHz, so single-carrier reception without long guard intervals collapses under inter-symbol interference. The paper shows that OFDMA converts the ripples into usable frequency diversity: after choosing the cyclic prefix from the worst-case excess delay and the FFT window from the RMS delay spread, a two-stage allocation--greedy assignment of subcarriers to the currently lowest-rate user using the ratio $\\Gamma_{m,k} = |H_{m,k}|^2 / \\max_{q\\neq m}|H_{q,k}|^2$, followed by per-user water-filling--approximately maximizes the minimum user rate. Simulations for a 30 m by 10 m room at 28 GHz with two or four users and low or moderate blockage show the OFDMA minimum rate consistently above the TDMA with single-carrier frequency-domain equalization and above a single centered PA.","pith_inferences":["Editorial inference: the equal-tap waveguide model is the part a hardware team would test first; if real leaky waveguides attenuate the traveling wave, the FIR taps become unequal and the reported OFDMA gains are an upper bound, though the general FIR-plus-OFDMA method would still apply.","Editorial inference: the same reasoning transfers to the uplink and to sub-6 GHz or lower-cost emulations; at lower carrier frequencies the physical aperture delays shrink relative to the symbol, so the frequency-selective effect and the OFDMA advantage should be smaller but measurable.","Editorial inference: the max-min greedy rule could be extended to jointly optimize PA positions, which the paper lists as future work; a testable prediction is that such joint placement gives diminishing returns once the cyclic prefix covers the worst-case delay spread.","Editorial inference: a direct rate-versus-$N$ experiment in a real 28 GHz waveguide would isolate the model's signature--OFDMA minimum rate rising with $N$--from other effects such as blockage, because the predicted slope is specific to the equal-amplitude FIR assumption."],"forward_implications":["Fixed PA placements can serve multiple users without repositioning or tracking, because the OFDMA frame absorbs the multipath instead of avoiding it.","Minimum user rate improves with the number of PAs under OFDMA, whereas the TDMA equalizer's effective SNR degrades as more taps add inter-symbol interference.","The scheme remains useful under moderate LoS blockage, where a single centered PA nearly flattens in performance.","System design parameters--cyclic prefix, subcarrier count, and FFT length--follow directly from measurable delays, giving a concrete OFDM frame-construction recipe for PA deployments.","OFDMA's joint subcarrier and power control is the mechanism for fairness; the greedy allocation approaches the max-min optimum at polynomial cost."],"supporting_citations":[{"why":"Supplies the simulation parameter set (28 GHz carrier, $n_e=1.4$, $-90$ dBm noise, $30 \\times 10$ m room) and the multi-PA scheduling context.","marker":"[9]"},{"why":"Supplies the LoS blockage model and the $\\beta$ values (0.05 to 0.15) used in the simulations.","marker":"[10]"},{"why":"Supplies the exponential LoS-probability formula $P(\\alpha_{m,n}=1)=e^{-\\beta\\|\\psi_m-\\psi_n^P\\|}$ used for blockage.","marker":"[11]"},{"why":"Supplies the capacity/rate expressions for the single-carrier and single-PA benchmarks against which OFDMA is compared.","marker":"[12]"}],"fun_headline_variants":["OFDMA turns pinch-antenna ISI into fairness gains","Pinching antennas: OFDMA lifts worst-user rate","OFDMA overcomes ISI in pinching-antenna downlinks","Multi-pinch antennas get fairer with OFDMA"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results assume the waveguide delivers an equal-amplitude, phase-shifted copy of the signal to every pinch point and the base station spreads each subcarrier's power uniformly across all $N$ apertures; a real waveguide's attenuation or position-dependent coupling would break the equal-tap channel and could shrink the OFDMA advantage.","fun_headline_variants_meta":{"raw":{"variants":["OFDMA turns pinch-antenna ISI into fairness gains","Pinching antennas: OFDMA lifts worst-user rate","OFDMA overcomes ISI in pinching-antenna downlinks","Multi-pinch antennas get fairer with OFDMA"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000486,"raw_usage":{"total_tokens":2431,"prompt_tokens":1017,"completion_tokens":1414,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":633,"completion_tokens_details":{"reasoning_tokens":1343}},"tokens_in":633,"tokens_out":1414,"duration_ms":9917,"temperature":1.0,"reasoning_tokens":1343,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:05:34.377673+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the magnitude of the frequency response $|H_m(f)|$ on a real 30 m leaky waveguide with $N$ evenly spaced pinch points at 28 GHz and compare the measured tap amplitudes and coherence bandwidth with Eq. (7); if the taps are unequal because of waveguide loss or coupling variation, the predicted OFDMA-over-TDMA minimum-rate gap should narrow with $N$, and the simulation curves would not be reproduced.","supporting_citations":[{"cited_title":"A primer on spatial modeling and analysis in wireless networks,","cited_arxiv_id":null,"evidence_quote":"Supplies the exponential LoS-probability formula $P(\\alpha_{m,n}=1)=e^{-\\beta\\|\\psi_m-\\psi_n^P\\|}$ used for blockage."},{"cited_title":"Capacity analysis of frequency-selective mimo channels with sub- optimal detectors,","cited_arxiv_id":null,"evidence_quote":"Supplies the capacity/rate expressions for the single-carrier and single-PA benchmarks against which OFDMA is compared."}],"review_version":1}