{"id":"4e7d5edd-02a8-424d-8cb7-3b8fec489040","arxiv_id":"2412.19974","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Adjusting the positions of STARS elements can improve the weighted sum rate in multiuser downlink by exploiting spatial diversity.","lead":"This paper proposes a movable-element simultaneously transmitting and reflecting surface (ME-STARS) for wireless networks, where the positions of the surface elements can be adjusted to boost signal quality. The authors show by simulation that this added mobility can raise the weighted sum rate of multiuser downlink by around 14 to 20 percent over fixed-element designs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"MS/TS algorithm extensions are asserted rather than derived; reported gains for these protocols are not independently verifiable.","rationale":"The reader's weakest_assumption was the far-field plane-wave channel model. That is a reasonable concern, but it is standard in the movable-antenna literature and applies equally to the ME-RIS baseline and FPE-STARS, so it does not single out the paper's contribution. A more concrete and explicitly acknowledged gap is the under-specification of the MS and TS algorithm extensions. The manuscript admits that the Θ_κ and τ_κ updates are omitted, referring to [6] for details. As a result, the numerical results for MS and TS, which contribute to the abstract's claim of gains 'for each ME-STARS operating protocol,' are not independently verifiable. This is a reproducibility weakness rather than a demonstrated error, and it does not undermine the fully derived ES algorithm, which is the paper's main contribution and shows the largest gain. I also note a secondary concern: the monotonic-convergence argument for Algorithm 3 relies on each subproblem improving the true WSR, but the position subproblem maximizes a penalized objective and may not guarantee a WSR increase across outer iterations; this is not explicitly proven. The concrete test I propose would settle the primary concern by checking whether the omitted MS/TS details can be reconstructed to match the figures. The verdict remains CONDITIONAL, matching the reader's assessment, because the ES results are plausible and the channel model is conventional, but the MS/TS claims require either added derivations or a scoped claim.","tokens_in":12,"tokens_out":13700,"duration_ms":198900,"concrete_test":"Implement the MS and TS variants strictly from the paper and the referenced STARS work [6], filling the omitted Θ_MS and Θ_TS/τ updates as described there. Run the Table I setup and compare the resulting MS/TS WSR curves to Figs. 4–5. If the reproduced MS/TS values differ by more than a few percent (e.g., >5%) or require unstated assumptions to match, the reported MS/TS gains are not reproducible from the manuscript, and the central claim should be narrowed to the ES protocol.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim covers all three STARS operating protocols, but the algorithms for MS and TS are not actually presented. Section III-B states that MS and TS are extensions of the ES algorithm, yet the binary-amplitude penalty for MS and the Θ_κ and τ_κ updates for TS are omitted: 'The details are omitted here for brevity' (MS) and 'Due to the page limit, the solving approaches for Θ_κ and τ_κ are omitted here' (TS). Consequently, the MS and TS curves in Figs. 4–8 cannot be reproduced from the paper alone, and the reported gains (14.33% and 14.21% in Fig. 5) are not independently checkable. Because the abstract and conclusion claim improvement 'for all operating protocols,' this missing support is load-bearing for the full claim. The ES algorithm is fully derived and delivers the largest gain, so the ES portion of the claim is much better supported; the concern is specifically that the MS/TS results rest on unstated algorithmic choices that may or may not match the text.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes a movable-element enabled simultaneously transmitting and reflecting surface (ME-STARS) for a multiuser downlink MISO system. The element positions on the STARS are optimized together with the base-station active beamforming and the STARS passive transmission/reflection coefficients, for the energy-splitting (ES), mode-switching (MS), and time-switching (TS) operating protocols. A weighted sum rate (WSR) maximization problem is formulated for each protocol, and an alternating optimization (AO) algorithm is developed for ES, combining gradient descent for the element-position subproblem, WMMSE for the active beamforming, and SCA for the passive beamforming. The paper states that the ES algorithm extends to MS and TS, and numerical results report WSR gains over fixed-position-element STARS (FPE-STARS) and over movable-element RIS (ME-RIS), with the largest ES gains around 20% in Fig. 5.","tokens_in":22572,"tokens_out":11918,"duration_ms":121554,"significance":"If the results hold, the paper identifies a useful new degree of freedom for STARS-based communications: the ability to move elements within a confined region can reduce multiuser interference and improve the weighted sum rate. The ES branch is presented in substantial detail, including closed-form gradient expressions in Appendix A, a step-size rule, complexity estimates, and a convergence argument. The numerical study is reasonably broad, covering number of users, transmit power, region size, number of multipath components, and number of elements. However, the MS and TS extensions are only sketched, with explicit statements that key solving details are omitted; since the abstract and conclusion claim improvements for all three protocols, the incomplete derivations for MS and TS are a load-bearing weakness. The assumed far-field geometric channel model is standard in the movable-antenna literature and is stated clearly, so it is a limitation to be acknowledged rather than an internal error.","major_comments":[{"comment":"The MS and TS algorithms are asserted rather than derived, and the text explicitly says so: for MS, 'The details are omitted here for brevity', and for TS, 'Due to the page limit, the solving approaches for Θκ and τκ are omitted here'. For TS, Algorithm 4 instructs the reader to update U via Algorithm 1 by substituting R_ES with R_TS, but Appendix A derives gradients specifically for R_ES, which has no τκ prefactor and includes interference from all users; the TS rate in Eq. (13) has a time-sharing factor and interference only within Jκ, so the gradient formulas do not transfer without additional derivation. The Θκ and τκ updates are not specified at all. Consequently, the MS and TS curves in Figs. 4-8, including the 14.33% and 14.21% gains reported in Fig. 5, cannot be reproduced from the manuscript. Because the abstract, contributions, and conclusion claim WSR improvement for all three operating protocols, this missing support is load-bearing. Please provide complete derivations for MS and TS in the main text or an appendix, or alternatively restrict the central claim to the ES protocol and present MS/TS as preliminary extensions with clearly labeled heuristic implementations.","section":"Section III-B and Algorithm 4"},{"comment":"The termination criterion of Algorithm 1 is inconsistent with the penalty constraint used in the optimization. Eq. (23b) enforces g(ũn, ũn') = 2D0/A - ||tanh(ũn) - tanh(ũn')||^2 ≤ 0, i.e., the squared tanh distance must be at least 2D0/A. Algorithm 1 line 15 instead stops when ||tanh(ũn) - tanh(ũn')||^2 ≥ D0. For normalized region sizes with A < 2, the algorithm may terminate with element positions that violate the minimum-distance constraint (14c); for the simulated A = 2.5λ the condition is stricter than the derived one but still does not match the stated constraint. Please correct the termination condition to match Eq. (23b) and confirm that the simulation code uses the same criterion.","section":"Algorithm 1 line 15 vs Eq. (23b)"},{"comment":"The claim that Algorithm 3 converges to a solution that is 'at least locally optimal' is not fully supported. The argument given is that the WSR is non-decreasing under alternating optimization and is bounded, which establishes convergence of the objective value but not stationarity or local optimality of the final iterate. To justify the local-optimality claim, one would need to show that each block update (including the penalized gradient ascent in Algorithm 1 and the SCA rank-one relaxation in Algorithm 2) converges to a stationary point of its subproblem and that the alternating sequence has the corresponding fixed-point property. Please either provide such an argument or weaken the statement to convergence of the objective to a limit point.","section":"Section III-A4"}],"minor_comments":[{"comment":"There are several typographical errors: 'degress-of-freedom' in the Abstract, 'scehmes' in Section IV-A, and 'activiated' in Section III-B; these should be corrected.","section":"Abstract, Section IV-A"},{"comment":"The text states that ME-STARS outperforms FPE-STARS 'by around 20.15%, 14.33%, and 14.21% ... in average', but it does not define what the average is taken over, nor does it report confidence intervals despite using 10^3 random realizations for each point. Please clarify the computation of these percentages and consider adding statistical dispersion information.","section":"Fig. 5 and Section IV"},{"comment":"The notation for the minimum-distance constraint is inconsistent: Eq. (14c) uses the Euclidean norm ‖un - un'‖2 ≥ D0, while Eq. (23b) uses the squared norm ‖tanh(ũn) - tanh(ũn')‖^2. Please make the notation uniform and state explicitly that the distance constraint is on the physical positions un.","section":"Eqs. (14c) and (23b)"},{"comment":"For TS, the time-allocation subproblem is described only as a 'typical resource allocation' problem with a reference to [6]. After fixing U, Wt, Wr, and Θt/Θr, the objective is linear in τt and τr, so a closed-form solution (allocate all time to the mode with larger weighted rate) exists and should be provided instead of being omitted.","section":"Section III-B, TS time allocation"},{"comment":"The paper relies on the far-field plane-wave geometric channel model with a small number of discrete paths. This is a standard assumption, but given that element-position optimization is the central mechanism, the authors should explicitly note that the predicted gains are contingent on accurate angle information and may be reduced under near-field propagation, diffuse scattering, or angle estimation errors.","section":"Section II-B"}],"recommendation":"major_revision","confidential_remarks":"The ES part of the paper is a solid and well-documented contribution for the movable-element STARS concept. The main issue is scope: the paper promises results for all three STARS protocols but omits the derivations for MS and TS, making the corresponding numerical claims non-reproducible. I would ask the authors to either complete those derivations in an appendix or a supplementary file, or to transparently reframe the paper's contribution around the ES protocol. The inconsistency in the Algorithm 1 termination condition should also be fixed and checked against the simulation code. The paper is within the scope of the journal and, with these fixes, could become publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe ES half of this paper is worth your time; the MS/TS half is not fully there yet. The new idea is to put movable elements on a STAR surface, so element positions become an extra degree of freedom alongside transmit/reflect beamforming. That combination is genuinely new relative to the cited work, which treats movable elements on reflection-only RIS or fixed positions on STAR. The system model and the ES optimization are carefully done: field-response channel model, position-dependent gradients derived in Appendix A, penalty method with tanh mapping for the feasible region, and WMMSE plus SCA for beamforming. The convergence argument is standard but at least stated. Simulation results show ME-STARS beating FPE-STARS by about 20% for ES and also improving MS/TS, with gains growing in user count and scatterer richness. That is plausible and matches the physics of having more spatial degrees of freedom.\n\nThe soft spot is exactly what the stress-test flags. The MS and TS algorithms are not derived. The text says 'the details are omitted here for brevity' for MS and 'due to the page limit, the solving approaches are omitted' for TS. The binary-amplitude penalty for MS and the Theta/tau updates for TS are not given. So the MS/TS curves in Figures 4–8 are not reproducible from the paper, and the reported 14% gains for those protocols rest on unstated algorithmic choices. Because the abstract and conclusion claim gains for all three operating protocols, this missing support is load-bearing for the full claim. The ES result, which carries the largest gain, is much better supported. My read is that the core contribution should be stated as ME-STARS with fully solved ES, with MS/TS as sketched extensions needing either derivation or a clear disclaimer.\n\nMinor items: no code/data, no error bars despite averaging over 1000 realizations, and the far-field plane-wave model is standard but may not hold under near-field or diffuse-scattering conditions, which would affect the optimized positions.\n\nThe paper deserves a serious referee. It makes a real, incremental but useful step in an active area, and the ES algorithm is derived well enough to check. I would accept it for review, with the condition that the authors either fill in the MS/TS details or narrow the claims.","headline":"Solid and new ES algorithm for movable-element STARs, but the MS/TS claims are placeholders that need to be either derived or retracted.","tokens_in":23105,"tokens_out":2382,"would_cite":true,"duration_ms":25328,"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":"This paper proposes a movable-element simultaneously transmitting and reflecting surface (ME-STARS) and claims that optimizing element positions improves weighted sum rate over fixed-position STARS by roughly 14–20% across operating…","keywords":["simultaneously transmitting and reflecting surface","movable element","position optimization","beamforming","weighted sum rate","reconfigurable intelligent surface","spatial degrees of freedom","multiuser MISO"],"falsifier":"Simulate or measure the same system with a near-field spherical-wave channel model, or with ray tracing that includes diffuse scattering, and run the plane-wave-based optimization algorithm; if the weighted sum rate gain of ME-STARS over fixed-position STARS shrinks from the reported roughly 20% toward zero, or if moving elements to the optimized positions yields no better sum rate than half-wavelength-spaced fixed elements, the central claim is falsified.","tokens_in":22224,"feed_emoji":"📡","tokens_out":7452,"duration_ms":66263,"temperature":0.7,"pith_summary":"This paper proposes a new type of intelligent surface: a simultaneously transmitting and reflecting surface (STARS) whose individual elements can be moved within a small confined region. It claims that letting element positions be optimized, on top of the usual phase and amplitude tuning, adds a valuable spatial degree of freedom that improves the weighted sum rate in a multiuser downlink. The paper develops an alternating-optimization algorithm that jointly optimizes base station beamforming, element positions, and passive transmission/reflection coefficients for three operating protocols (energy splitting, mode switching, time switching). Numerically, the movable-element design is reported to outperform a fixed-position STARS by about 20% in weighted sum rate for the energy-splitting protocol, with smaller but consistent gains for the other protocols. The practical promise is better full-space coverage with less inter-user interference, at the cost of mechanically adjustable elements.","feed_headline":"Moving STARS elements yields ~20% higher weighted sum rate","feed_subtitle":"Tuning element positions adds a spatial degree of freedom that cuts inter-user interference in full-space coverage.","key_machinery":"The load-bearing object is the field-response channel model, in which each multipath component is described by fixed angles so that the channel response between the base station and a user via a STARS element at position $u_n$ becomes a sum of plane-wave exponentials of the form $e^{j 2\\pi(\\rho_{S,j}^p(u_n)-\\rho_{S,in}^o(u_n))/\\lambda}$, where the $\\rho$ terms are linear functions of the element coordinates $x_n$ and $y_n$. The paper rewrites this cascaded channel as the matrix $F_j(u_n)=f_j(u_n)(f_{in}(u_n))^H$ and derives closed-form gradients of the weighted sum rate with respect to the element coordinates. Those gradients, combined with a penalty method for the region and minimum-distance constraints and log-sum-exp smoothing, drive the gradient-ascent position update in Algorithm 1. The remaining blocks are the standard WMMSE update for the base station beamforming and an SCA rank-one relaxation for the passive transmission and reflection coefficients.","core_discovery":"The central claim is that moving the elements of a STARS within a confined two-dimensional region, rather than fixing them at half-wavelength spacing, creates an extra spatial degree of freedom that can be optimized to reduce channel correlation among users and improve the weighted sum rate. Treating each element's position as a continuous variable, the paper formulates a WSR maximization problem for the energy-splitting (ES), mode-switching (MS), and time-switching (TS) protocols, and solves it by alternating between gradient-descent position updates, WMMSE active beamforming, and SCA-based passive beamforming. The simulations show ME-STARS outperforming fixed-position STARS by 20.15% (ES), 14.33% (MS), and 14.21% (TS) in weighted sum rate at the reference transmit power, and outperforming a movable-element RIS used for full-space coverage in the ES and MS modes. The gains grow with the number of users and with the number of multipath components, because more scatterers give the position optimization more local maxima to exploit.","pith_inferences":["The paper leaves implicit that the same gradient machinery could be applied to other element-controlled surfaces, such as holographic or active STARS, provided the phase response model is extended accordingly.","A testable implication outside the paper is that position optimization acts as a kind of spatial interference whitening; one could try a simpler heuristic that moves elements to maximize the angular separation between users' cascaded channel vectors and check how much of the reported gain it recovers.","Angle estimation error is an untested vulnerability: the gradients exploit fine phase differences, so even small angle-of-arrival and angle-of-departure estimation errors could erase the gain in a real deployment, where fixed-position beamforming would be more robust.","Because the paper assumes positions are constant over the fading block, the benefit in mobile channels depends on how fast elements can be physically moved; predictive position tracking would be a natural follow-up."],"forward_implications":["If the central claim is right, adding movable elements to a STARS gives a new spatial degree of freedom that is most valuable when the number of users approaches the number of surface elements, because position optimization decorrelates otherwise similar user channels.","Energy-splitting mode reaps the largest benefit from element movement (about 20% WSR gain over fixed positions at 30 dBm), since it already has the most passive-beamforming flexibility.","The gain from movable elements grows with the number of multipath components, so the approach is better suited to rich-scattering environments than to sparse line-of-sight channels.","A finite movable region is sufficient: WSR saturates as the region size grows, which suggests that practical mechanical constraints need not be the bottleneck.","The alternating-optimization algorithm converges in a small number of outer iterations (around 5 for ES and 10 for TS), making the joint design computationally feasible for moderate element counts."],"supporting_citations":[{"why":"Defines the three STARS operating protocols (ES, MS, TS) and their feasible transmission/reflection coefficient sets that the paper adopts.","marker":"[6]"},{"why":"Gives the field-response channel model and the position-dependent field-response vector formulation used to write cascaded channels as functions of element positions.","marker":"[14]"},{"why":"Establishes the multiuser movable-antenna position-optimization problem and the alternating-optimization structure that the proposed algorithm follows.","marker":"[15]"},{"why":"Presents the closest prior system, an intelligent reflecting surface with two-dimensional movable elements, used here as the movable-element RIS baseline and the motivation for 2D surface movement.","marker":"[27]"},{"why":"Supplies the weighted minimum mean square error method used to solve the active beamforming subproblem at the base station.","marker":"[32]"},{"why":"Supplies the successive convex approximation method used to solve the passive beamforming subproblem at the STARS.","marker":"[35]"},{"why":"Introduces the fluid-antenna concept of position-adjustable antennas, the broader family this work extends to full-space surfaces.","marker":"[12]"}],"fun_headline_variants":["Move STARS elements for a 20% sum-rate boost","STARS with movable elements lift sum rate ~20%","Shifting STARS element positions boosts WSR by 20%","Movable STARS elements add spatial diversity, lifting sum rate"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole design rests on the far-field plane-wave channel model: each link is a small number of discrete paths with known fixed angles, and moving an element only changes the phase of each path linearly; if real channels have near-field curvature, diffuse scattering, or angle uncertainty, the optimized positions and the predicted gains need not materialize.","fun_headline_variants_meta":{"raw":{"variants":["Move STARS elements for a 20% sum-rate boost","STARS with movable elements lift sum rate ~20%","Shifting STARS element positions boosts WSR by 20%","Movable STARS elements add spatial diversity, lifting sum rate"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001522,"raw_usage":{"total_tokens":6134,"prompt_tokens":1019,"completion_tokens":5115,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":5044}},"tokens_in":635,"tokens_out":5115,"duration_ms":35260,"temperature":1.0,"reasoning_tokens":5044,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T23:42:28.716924+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or measure the same system with a near-field spherical-wave channel model, or with ray tracing that includes diffuse scattering, and run the plane-wave-based optimization algorithm; if the weighted sum rate gain of ME-STARS over fixed-position STARS shrinks from the reported roughly 20% toward zero, or if moving elements to the optimized positions yields no better sum rate than half-wavelength-spaced fixed elements, the central claim is falsified.","supporting_citations":[{"cited_title":"Simultaneou sly transmitting and reflecting (STAR) RIS aided wireless commu nications,","cited_arxiv_id":null,"evidence_quote":"Defines the three STARS operating protocols (ES, MS, TS) and their feasible transmission/reflection coefficient sets that the paper adopts."},{"cited_title":"Modeling and performance an alysis for movable antenna enabled wireless communications,","cited_arxiv_id":null,"evidence_quote":"Gives the field-response channel model and the position-dependent field-response vector formulation used to write cascaded channels as functions of element positions."}],"review_version":1}