{"id":"1b9778bd-ba7d-4e7d-944f-d47ea3a8586a","arxiv_id":"2504.12885","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Optimized movable antenna positions raise uplink sum rates by up to 57% over compact uniform arrays in LOS-dominant wideband MIMO, but only a few percent in rich scattering or at high SNR.","lead":"This paper examines whether physically movable base-station antennas improve data rates in wideband multi-user MIMO systems. It finds that the gains are large only in line-of-sight, interference-limited settings and largely disappear under rich scattering, many subcarriers, or strong hardware distortion.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Missing fixed irregular-array baseline: the paper's own conclusion concedes a pre-optimized fixed array could match movable antennas, so reported gains over uniform arrays may not show a benefit of actual movement.","rationale":"The reader's weakest assumption is the missing optimized fixed-array baseline, and the paper's own conclusion explicitly concedes that a pre-optimized fixed irregular array could achieve similar rates. This is a load-bearing concern because the abstract frames movable antennas as a way to achieve higher rates with few antennas; if the same rate is achievable with a fixed irregular array, the contribution reduces to geometry optimization rather than mechanical movement. The paper's Lemma 1 and Corollary 1 are correctly derived, and the qualitative dependence on EVM, subcarrier count, and scattering richness is supported by the simulations. However, the quantitative headline numbers (57%, 28%, 2.4%) compare against non-optimized uniform baselines, and no wideband comparison with a fixed optimized irregular array is provided. The proposed test would settle whether movability itself adds value. I therefore keep the reader's CONDITIONAL verdict: the model is sound, but the central claim needs the missing baseline comparison and ideally code or error bars to substantiate the exact gains. This is a condition that can be met with an additional numerical experiment.","tokens_in":7493,"tokens_out":5989,"duration_ms":62897,"concrete_test":"Re-run the numerical study of Figs. 3 and 4 (or a statistically matched subset) with an additional baseline: a fixed irregular array whose M positions are optimized once by the same PSO with the same constraints, using the average channel over all considered user realizations (or a training set) as the objective, and then evaluated on the same realizations. Also include the pre-optimized array from reference [15] if available. If the fixed-optimized-array rate is within, say, 1% of the movable-antenna rate in LOS-dominant and rich-scattering conditions, the benefit of movement is negligible and the central claim should be reframed. If the gap exceeds the reported differences, dynamic movement is essential.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claim is that optimized movable antennas achieve up to 57% higher sum rate than a compact UPA and 28% higher than a sparse UPA in LOS-dominant wideband channels. These gains are computed against fixed uniform arrays that are not optimized for the channel. The load-bearing assumption is that the appropriate baseline reflects the best achievable performance of a fixed-geometry array; otherwise the improvement attributed to 'movable antennas' is actually an improvement from geometry optimization that could be realized without any mechanical movement. Section V explicitly concedes: 'one could potentially use a fixed irregular array optimized for the propagation scenario (as proposed in [15]) to achieve similar rates without antenna movements.' If that is true, the headline percentages overstate the advantage of dynamic antenna relocation. The paper does not simulate such a baseline in the wideband setting, so the incremental value of movability remains unquantified. This concern does not invalidate the channel model or the qualitative dependence on impairments/multipath, but it directly affects the paper's motivating claim that movable antennas are a distinct enabler of higher rates.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies a wideband multi-user MIMO uplink with movable antennas at the base station. The authors derive an OFDM system model with a geometric multipath channel that includes pulse-shaping, and model user hardware impairments via the error vector magnitude (EVM). Lemma 1 gives an achievable average sum rate expression, and Corollary 1 shows a high-SNR ceiling K log2(1/EVM^2) that is independent of the antenna positions. The antenna positions are optimized by particle swarm optimization under per-antenna movement regions and a minimum-separation constraint. Numerical results in LOS-dominant and rich-scattering scenarios compare the optimized movable array against sparse and compact uniform arrays, leading to the claimed conditional gains. The paper concludes that movable antennas are most useful in LOS-dominant, interference-limited wideband systems, and it explicitly concedes that a pre-optimized fixed irregular array may achieve similar rates without antenna movement.","tokens_in":7633,"tokens_out":3916,"duration_ms":39064,"significance":"The derivation of the wideband channel model with pulse shaping and the achievable sum-rate expression is a solid contribution, and the high-SNR ceiling in Corollary 1 is a neat result that is independent of the array geometry. If the numerical claims are robust, the paper would provide useful guidance on when movable antennas help in wideband systems. Strengths include the explicit worst-case noise argument in the proof of Lemma 1, the algebraic proof of the EVM ceiling, and the identification of interference-limited regimes as the relevant operating point for movable antenna gains. The main weakness is that the numerical baselines do not include a pre-optimized fixed irregular array, so the headline gains may overstate the value of physical movement. The PSO solver also lacks a global optimality guarantee, so the reported rates may be local optima that could affect the quantitative conclusions.","major_comments":[{"comment":"The conclusion explicitly states: 'In scenarios where movable antennas outperform a sparse ULA/UPA, one could potentially use a fixed irregular array optimized for the propagation scenario (as proposed in [15]) to achieve similar rates without antenna movements.' The three baselines considered in Section IV (sparse UPA, sparse ULA, and compact UPA) are all uniform arrays and are not optimized for the channel. Consequently, the reported gains of up to 57% versus a compact UPA and 28% versus a sparse UPA (Section IV-A) conflate geometry optimization with physical antenna movement. To support the central claim that movable antennas provide a distinct benefit, the paper should compare against a fixed irregular array optimized for the same propagation scenario in the wideband setting, or at least bound the best achievable fixed-geometry performance. Without such a baseline, the paper's central quantitative claims are not substantiated.","section":"V (Conclusion) and IV (Baselines)"},{"comment":"The PSO algorithm is terminated after 100 iterations and no optimality-gap analysis is provided. The statement in Section IV-A that 'The proposed PSO algorithm finds excellent antenna positions, which are only 3 % from the interference-free upper bound' is not supported by any convergence test or comparison with a global optimum. Since the objective in (13) is non-concave and the search space is effectively 32-dimensional (16 antennas, each movable in two dimensions), the solutions may be local optima, and the reported differential gains over fixed arrays could change with a better optimizer. Please provide evidence of near-global optimality, for example by running multiple restarts, comparing against exhaustive search on a low-dimensional instance, or refining the PSO output with a local gradient-based method.","section":"III-A and IV-A"}],"minor_comments":[{"comment":"The exponent in (4) is written as e^{-j2πλ(τ_{i,n}-η)/c}; using λ/c = 1/f_c makes this algebraically correct, but the notation is nonstandard and may confuse readers. Rewriting it as e^{-j2π f_c (τ_{i,n}-η)} would make the units clear.","section":"II-A, Eq. (4)"},{"comment":"The y-axis label 'Average sum rate' is used while the text refers to 'rate per subcarrier' in the discussion of increasing S. Since R_Σ in (8) is already an average over subcarriers, please clarify whether the plotted quantity is the per-subcarrier average or the total sum rate, and adjust the label and captions accordingly.","section":"IV-A, Figs. 3 and 4"},{"comment":"The sentence 'The loss is largest with movable antennas' is ambiguous because the total sum rate increases with S; the underlying observation is that the average rate per subcarrier decreases. Consider stating this explicitly in relative terms.","section":"IV-A"},{"comment":"The parenthetical 'which was not observed in [10]' is unclear: it is not obvious whether [10] claimed the opposite effect or simply did not study the high-SNR asymptote. Clarify the comparison to [10] or remove the parenthetical.","section":"IV-B"},{"comment":"The numerical percentages in the conclusion (22%, 5.3%, 14%, 4.8%) are not all directly traceable to the figures because the reference baseline changes between the sparse ULA and sparse UPA. Please specify which baseline each percentage refers to so that the summary is reproducible.","section":"V"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a communications journal and the modeling part is competent. The main issue is that the authors' own conclusion concedes that a fixed pre-optimized irregular array could achieve similar rates without antenna movement; this directly affects the interpretive claim of the numerical section. I would not reject the paper, because the wideband channel model, the sum-rate derivation, and the EVM ceiling are valuable contributions, but a revision should integrate a fixed irregular-array baseline and strengthen the optimality evidence for the PSO solutions before the quantitative claims can be accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid, honest extension of movable-antenna optimization from narrowband to wideband multi-user MIMO. The model is new: it includes pulse-shaping effects and EVM impairments at the user side, and the derivation of the sum rate in Lemma 1 is clean. The high-SNR ceiling in Corollary 1 follows algebraically and is neat because it is independent of antenna positions. The numerical study is suggestive and the qualitative conclusion—movable antennas help mainly in LOS-dominant, interference-limited scenarios with few subcarriers—is supported by the simulations.\n\nThe soft spot is the one the authors themselves concede: the headline gains (up to 57% over compact UPA, 28% over sparse UPA) are measured against fixed uniform arrays that are not optimized for the propagation scenario. Section V admits that a pre-optimized fixed irregular array, as in their own related work [15], could achieve similar rates without any movement. That baseline is never simulated in the wideband setting, so the incremental value of actually moving antennas remains unquantified. The comparison against a sparse ULA is fairer, but still not the best possible fixed geometry. This does not invalidate the model or the qualitative findings, but it does mean the motivating claim—that movability itself is a distinct enabler—is not proven.\n\nAlso, the PSO solver has no optimality guarantee, the results are shown without error bars, and the number of scenarios is limited. These are minor-to-moderate issues, not fatal. The analytical parts hold up, and the paper is transparent about its limitations, which counts for something.\n\nWho is this for? Researchers working on movable antennas or fluid-antenna systems who need a wideband MIMO reference model. The contribution is incremental but useful, and the paper is worth engaging with seriously.\n\nRecommendation: send it to peer review. The authors should be asked to run the optimized fixed-irregular-array baseline and, ideally, report variance across channel realizations. With that added, the quantitative claims would be on much firmer ground.","headline":"Honest, well-built wideband movable-antenna model, but the paper never quantifies the benefit of movement itself against the best fixed irregular array.","tokens_in":8226,"tokens_out":1433,"would_cite":true,"duration_ms":15733,"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":"Movable antennas in wideband MIMO mostly pay off in line-of-sight, interference-limited channels, not in rich scattering at practical subcarrier counts.","keywords":["movable antennas","wideband MIMO","OFDM","hardware impairments","error vector magnitude","particle swarm optimization","sum rate maximization","multipath channels"],"falsifier":"Simulate the same LOS-dominant scenario with a fixed irregular array whose positions are optimized once for the user distribution, using the same PSO objective and the same $5\\lambda \\times 5\\lambda$ aperture per antenna. If that static array's average sum rate matches the movable-antenna result within a few percent, then the claim that dynamic movement drives the reported 28–57% gains is falsified; the gains would come from geometry selection rather than movement.","tokens_in":1983,"feed_emoji":"📡","tokens_out":4409,"duration_ms":117105,"temperature":0.7,"pith_summary":"This paper asks whether movable antennas, whose positions at a base station can be physically adjusted, meaningfully improve data rates in wideband multi-user MIMO systems, where prior work had mostly considered narrowband operation. It derives an OFDM uplink model that includes the pulse-shaping filter and transceiver hardware impairments, then optimizes antenna positions with particle swarm optimization and compares against fixed uniform arrays. The central finding is that the benefit of moving antennas is conditional: in line-of-sight-dominant channels, optimized movable antennas achieve up to 57% higher average sum rate than a compact uniform planar array and up to 28% higher than a sparse one, but in rich scattering with practical subcarrier counts the gain over a sparse uniform linear array falls to about 2.4%. Hardware impairments cap the achievable rate at a level independent of antenna positions, so movable antennas help mainly in interference-limited rather than hardware-limited regimes. These results matter because they identify when the complexity of physically moving antennas is actually worth deploying.","feed_headline":"Movable antennas gain up to 57% — but only in LOS channels","feed_subtitle":"The advantage shrinks to about 2.4 percent in rich scattering with many subcarriers, so fixed arrays often suffice.","key_machinery":"The mechanism is a geometric wideband channel model in which each antenna's position enters through the array response vector $a_P(\\varphi,\\theta) = [e^{j p_1^T k(\\varphi,\\theta)},\\ldots,e^{j p_M^T k(\\varphi,\\theta)}]^T$, with wave vector $k(\\varphi,\\theta)$ determined by azimuth and elevation angles. The multipath channel becomes a finite-impulse-response filter whose taps combine path amplitudes, delays, and the position-dependent array response, and each subcarrier's channel is the DFT of those taps. The sum rate is the average over subcarriers of a log-det expression with two terms: the ideal-hardware rate minus a hardware-impairment penalty controlled by the error vector magnitude; the high-SNR limit $K \\log_2(1/\\mathrm{EVM}^2)$ is independent of positions. Particle swarm optimization then searches the antenna positions, constrained to non-overlapping squares with a minimum $\\lambda/2$ spacing, to maximize this sum rate.","core_discovery":"The paper establishes that in a wideband multi-user MIMO uplink with a base station carrying few antennas ($M \\approx K$), optimizing antenna positions with particle swarm optimization yields the largest rate gains when the channel is line-of-sight dominant and the system is interference-limited. For a 16-antenna array serving 10 users, the optimized movable configuration reaches within 3% of an interference-free upper bound and outperforms a compact uniform planar array by up to 57% and a sparse uniform planar array by up to 28%. In rich non-line-of-sight scattering, the gains shrink rapidly as subcarriers increase; at practical subcarrier counts the improvement over a sparse uniform linear array is only about 2.4%. The paper also derives an achievable sum-rate expression under error-vector-magnitude hardware impairments and shows that as transmit power grows the rate converges to $K \\log_2(1/\\mathrm{EVM}^2)$, a ceiling independent of antenna positions. Hence movable antennas are beneficial precisely when inter-user interference, not hardware distortion, is the bottleneck.","pith_inferences":["An implication the paper leaves implicit is a sharper test: in a static deployment where the user distribution is known in advance, a one-time optimized irregular array should capture most of the movable-antenna gain, so the real advantage of movable antennas may be adaptability to changing user locations rather than peak rate.","Because the gains are tied to channel similarity across subcarriers, a natural extension is to predict that in wideband systems with high mobility or fast-changing multipath, the optimal positions would need frequent re-optimization, and the PSO-based approach as presented would need an online, low-complexity update rule to remain practical.","The same model could be applied to the downlink with hardware impairments at the base station; since the asymptotic ceiling is independent of positions there as well, one would expect the interference-limited LOS regime to again be the only favorable case.","A testable design rule follows from the high-SNR limit: when the operating SNR already puts the system near $K \\log_2(1/\\mathrm{EVM}^2)$, investing in antenna movement is pointless; the distance between the current rate and that ceiling tells whether position optimization is worth the mechanical cost."],"forward_implications":["In LOS-dominant wideband scenarios with moderate subcarrier counts, movable antennas can nearly match an interference-free upper bound (within about 3%), yielding gains up to 57% over a compact UPA and 28% over a sparse UPA.","In rich scattering with many subcarriers, different subcarriers prefer different antenna positions, so the gain over a sparse ULA falls to about 2.4%; fixed arrays are nearly as good.","Hardware impairments set an EVM-dependent rate ceiling $K \\log_2(1/\\mathrm{EVM}^2)$ that no antenna movement can exceed; at high SNR all array geometries converge to the same limit, and movable-antenna advantages become negligible.","The relative benefit of movable antennas degrades with bandwidth: from roughly 14% at 20 or fewer subcarriers down to 4.8% at 100 or more subcarriers in non-LOS conditions.","Because a fixed irregular array pre-optimized for the propagation scenario could plausibly achieve similar rates without any movement, the value of dynamic movement itself is narrower than the headline gains over uniform arrays suggest."],"supporting_citations":[{"why":"Surveys prior narrowband movable-antenna systems, defining the baseline that wideband operation must extend.","marker":"[3]"},{"why":"Provides the prior single-user wideband movable-antenna model that this paper generalizes to multi-user MIMO with pulse shaping.","marker":"[8]"},{"why":"Supplies the error-vector-magnitude hardware-impairment model and the rate analysis that the sum-rate and high-SNR limit build on.","marker":"[9]"},{"why":"Gives the finite-impulse-response tap description used to convert geometric multipath into the OFDM channel.","marker":"[11]"},{"why":"Its worst-case uncorrelated additive noise theorem is used to prove the achievable sum rate under hardware distortion.","marker":"[12]"},{"why":"Defines the particle swarm optimization method used to search antenna positions.","marker":"[13]"},{"why":"The authors' related work on pre-optimized irregular arrays, which the conclusion cites as a possible fixed alternative to movement.","marker":"[15]"},{"why":"Provides the urban-microcell spatial channel parameters used in the numerical comparisons.","marker":"[16]"}],"fun_headline_variants":["Movable antennas boost wideband MIMO only in LOS channels","Movable antenna gains in wideband MIMO depend on scattering","LOS channels key for movable antennas in wideband MIMO","Movable antennas help in wideband MIMO only with little scattering"],"cache_read_input_tokens":10368,"weakest_assumption_plain":"The reported gains are measured against fixed uniform arrays that were not optimized for the propagation scenario, and if a fixed irregular array pre-optimized for the scenario, as the paper itself notes is possible, is the benchmark, the advantage of physically moving antennas would shrink or disappear.","fun_headline_variants_meta":{"raw":{"variants":["Movable antennas boost wideband MIMO only in LOS channels","Movable antenna gains in wideband MIMO depend on scattering","LOS channels key for movable antennas in wideband MIMO","Movable antennas help in wideband MIMO only with little scattering"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000522,"raw_usage":{"total_tokens":2533,"prompt_tokens":959,"completion_tokens":1574,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":575,"completion_tokens_details":{"reasoning_tokens":1503}},"tokens_in":575,"tokens_out":1574,"duration_ms":11005,"temperature":1.0,"reasoning_tokens":1503,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:21:06.151515+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate the same LOS-dominant scenario with a fixed irregular array whose positions are optimized once for the user distribution, using the same PSO objective and the same $5\\lambda \\times 5\\lambda$ aperture per antenna. If that static array's average sum rate matches the movable-antenna result within a few percent, then the claim that dynamic movement drives the reported 28–57% gains is falsified; the gains would come from geometry selection rather than movement.","supporting_citations":[{"cited_title":"A tutorial on movable antennas for wireless networks,","cited_arxiv_id":null,"evidence_quote":"Surveys prior narrowband movable-antenna systems, defining the baseline that wideband operation must extend."},{"cited_title":"Performance analysis and optimization for movable antenna aided wideband communications,","cited_arxiv_id":null,"evidence_quote":"Provides the prior single-user wideband movable-antenna model that this paper generalizes to multi-user MIMO with pulse shaping."},{"cited_title":"Massive MIMO networks: Spectral, energy, and hardware efficiency,","cited_arxiv_id":null,"evidence_quote":"Supplies the error-vector-magnitude hardware-impairment model and the rate analysis that the sum-rate and high-SNR limit build on."},{"cited_title":"Björnson and Ö","cited_arxiv_id":null,"evidence_quote":"Gives the finite-impulse-response tap description used to convert geometric multipath into the OFDM channel."},{"cited_title":"How much training is needed in multiple-antenna wireless links?","cited_arxiv_id":null,"evidence_quote":"Its worst-case uncorrelated additive noise theorem is used to prove the achievable sum rate under hardware distortion."},{"cited_title":"Pre-Optimized Irregular Arrays versus Moveable Antennas in Multi-User MIMO Systems","cited_arxiv_id":"2502.03994","evidence_quote":"The authors' related work on pre-optimized irregular arrays, which the conclusion cites as a possible fixed alternative to movement."},{"cited_title":"3GPP TS 25.996, Jul","cited_arxiv_id":null,"evidence_quote":"Provides the urban-microcell spatial channel parameters used in the numerical comparisons."}],"review_version":1}