{"id":"c05bba14-9cf6-4867-af4e-3711c263db23","arxiv_id":"2507.18793","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A flexible-intelligent-metasurface-parameterized doubly dispersive MIMO channel model is proposed, and optimizing the surface shape at both link ends is shown by simulation to improve achievable rate and angle-of-arrival estimation for OFDM, OTFS, and AFDM in integrated sensing and communications.","lead":"This paper adds flexible intelligent metasurfaces, antenna arrays that can physically bend, to the math used for high-mobility wireless channels that change quickly in both time and frequency. It claims that optimizing the surface shape, not just tuning phases, can noticeably improve both data rates and radar-like sensing in future 6G networks.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed FIM gains reduce to per-element phase control; without a same-DoF phase-array baseline or physical shape constraints, the FIM-specific claim is not established.","rationale":"The reader's weakest assumption was the phase-only point-radiator idealization in Equation (3), and I agree that this is the soft spot. My stress-test sharpens the concern: because y_b spans a full wavelength, the optimization variable is exactly a per-element phase shifter, so the paper's numerical machinery never exercises anything beyond phase control. This makes the absence of a same-DoF phase-array baseline particularly damaging to the claim that FIM morphing itself is the enabler of the gains. The paper is otherwise internally consistent: the input-output relations in Equations (18), (24), and (30) are structurally correct, the gradient ascent formulas follow from standard matrix calculus, and the waveform-rate equality follows from unitary equivalence of the transforms. The concern is not a mathematical contradiction but an unvalidated physical interpretation plus an unfair baseline. A conditional verdict is therefore appropriate: the results can be accepted as an optimization study of phase-controlled arrays, but the FIM-specific conclusions require either a phase-array baseline or a physical validation of the morphing model. Since the reader already assigned CONDITIONAL, no verdict change is needed.","tokens_in":18876,"tokens_out":12756,"duration_ms":164108,"concrete_test":"Re-run the simulations with each y_b replaced by an independent phase variable φ_b ∈ [0,2π) in Equation (4), optimizing the same objective and applying the same MUSIC routine; if the rate-versus-SNR curves and spectra coincide with Figures 2-6, the FIM-specific claim collapses to conventional phase control. As a physical check, augment the optimization with a mechanical smoothness constraint such as |y_{b+1} - 2y_b + y_{b-1}| ≤ ε or use full-wave simulated array responses for several morphed FIM shapes; if the optimized gains shrink substantially, the unconstrained phase-only model is optimistic.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equations (3) and (4) place each y_b only in the phase e^{j(2π/λ)y_b sinθ sinφ}; with y_b ranging over [-λ,λ], each per-element phase is independently reachable. Thus the optimization problem (35) and the gradients in Equations (36)-(45) are mathematically identical to choosing arbitrary phase shifters at the TX and RX elements of a conventional array. The numerical comparisons are only 'no FIM' (y = 0) versus random or optimized y; there is no baseline with the same number of phase-only degrees of freedom, such as a conventional phase-controlled array or an RIS with independent phase shifts. Consequently, the reported ~2.5 dB gain for random FIMs, the additional ~2 dB gain for optimized FIMs, and the sharper MUSIC peaks in Figures 2-6 may demonstrate phase optimization rather than any benefit specific to the physical morphing of a flexible metasurface. Moreover, the model imposes only the box constraint (1c) on y_b and ignores mechanical coupling of a continuous elastic surface, so the independently optimized y values may not be physically realizable. Footnote 3 addresses only delay and Doppler variations, not these issues.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a FIM-parameterized doubly-dispersive (FPDD) MIMO channel model in which the y-coordinates of the meta-atoms enter the array response through phase-only factors, and applies the model to OFDM, OTFS, and AFDM in a high-mobility ISAC setting. It derives effective input-output relations for the three waveforms, formulates a joint achievable-rate maximization problem with a sensing constraint, solves it by gradient ascent with closed-form gradients, and reports simulations showing rate gains and sharper MUSIC-based angle-of-arrival estimates when FIMs are optimized. The paper is clearly written and the signal-processing derivations are mostly standard, but the numerical evidence for the FIM-specific claim is incomplete.","tokens_in":19129,"tokens_out":8427,"duration_ms":90101,"significance":"If the results held, the paper would provide a unified optimization framework for metasurface shape design across three important DD-channel waveforms. The derivation of the effective channel matrices for OFDM, OTFS, and AFDM is careful, and the closed-form gradient expressions appear correct, which are useful assets. However, the paper does not compare against a conventional phase-controlled array or RIS with the same number of phase-only degrees of freedom, and the optimization and evaluation use the same idealized model. The paper is therefore more convincing as a rate-optimization framework for phase-configurable arrays than as evidence that physical FIM morphing provides intrinsic gains beyond what ordinary per-element phase control can achieve.","major_comments":[{"comment":"The response vector depends on the FIM coordinates only through the phase factor e^{j(2π/λ)y_b sinθ sinφ}. Because each y_b is an independent continuous variable, the optimization in Eq. (35) has the same structure and the same number of degrees of freedom as optimizing per-element phase shifters in a conventional array. The numerical comparisons in Figs. 2-6 are only y=0 versus random or optimized y; there is no baseline with the same number of phase-only degrees of freedom, such as a conventional phase-controlled array or an RIS with independent phase shifts. The reported gains of about 2.5 dB for random FIMs, the additional 2 dB for optimized FIMs, and the MUSIC improvements are therefore not attributable to the physical morphing mechanism; they may simply demonstrate the benefit of adding phase degrees of freedom. This is load-bearing for the paper's central claim.","section":"Section II.A, Eq. (3), and Section IV.B, Eq. (35)"},{"comment":"The optimization treats each y_b as an independent variable under only box constraints and ignores the mechanical coupling of a continuous elastic surface, as well as changes in radiation patterns and mutual coupling. Moreover, the update rule in Eq. (46) is an unconstrained gradient step; no projection onto [ymin, ymax] is described, so the final y values may violate the stated bounds. Without either physical shape constraints or a projection/barrier mechanism, the optimized y values may not be realizable by a flexible metasurface.","section":"Section II.A, Eqs. (1c)-(2), and Algorithm 1"},{"comment":"The optimization and evaluation use the same idealized model, and Algorithm 1 requires perfect knowledge of all path parameters (gains, delays, Doppler shifts, and angles) while the transmit covariance is fixed to T=I_Nds. In addition, the sensing constraint in Eq. (31c) is a total received-power proxy, not a detection or estimation metric, so the MUSIC results in Figs. 4-6 are not directly optimized by the algorithm. These issues do not invalidate the derivations, but they mean that the practical gains in a real FIM/ISAC system are not yet established.","section":"Section IV.A-B, Eq. (31c), and Section V"}],"minor_comments":[{"comment":"The text sets ymin=0 after Eq. (1c), whereas Table II sets ymin=-λ m; Eq. (2) defines y as nonnegative, which also excludes the negative values permitted by Table II. Please reconcile these definitions.","section":"Section II.A, Eq. (2)"},{"comment":"The row 'Sampling Frequency λ 20 MHz' should read 'Sampling Frequency F_S 20 MHz'; the symbol λ is used for wavelength elsewhere.","section":"Table II"},{"comment":"'Hardamard product' should be 'Hadamard product'.","section":"Section IV.B, Eq. (37)"},{"comment":"The captions in Figs. 2-3 contain 'NT=NT=4', which should be 'NT=NR=4', and the axis labels in Figs. 4-7 contain garbled LaTeX such as '?inp' and '3inp'. Please fix these rendering issues.","section":"Figure captions"},{"comment":"The sentence 'the achievable rates for the different waveforms across all the cases are almost identical...' is repeated verbatim after Fig. 3; the duplicate should be removed.","section":"Section V.A"},{"comment":"Reference [11] appears incomplete: 'D. W. Bliss and S. Govindasamy. Cambridge University Press, 2013' is missing the book title.","section":"References"},{"comment":"The approximation T≈I_Nds is justified only for large N·ds, but the simulations use N=16 or 64 and ds=4; a sensitivity check or a footnote assessing this approximation would be useful.","section":"Section IV.B"}],"recommendation":"major_revision","confidential_remarks":"The central difficulty is that the FIM-specific contribution rests on comparisons that do not include a same-DoF phase-control baseline. The paper also builds heavily on the authors' prior metasurface channel models (e.g., [39]-[41]), and the novelty relative to those works should be stated more explicitly. The authors should be encouraged to add the missing baseline and, if the phase-control equivalence holds, to reframe the contribution accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a look if you track the FIM/DD channel line. The genuinely new piece is a channel model that puts each FIM element's vertical position into the array phase, then runs gradient ascent on those positions for OFDM/OTFS/AFDM ISAC rate with a sensing penalty. The math is clean: equations (36)-(45) give correct closed-form gradients, and the unified Kronecker treatment of the three waveforms is useful. The comparative I/O structures in (18), (24), and (30) are a nice didactic contribution.\n\nThe stress-test note lands. Since y_b enters only via e^{j2π y_b sinθ sinφ/λ} and y_b ranges over ±λ, each element's phase is effectively independently adjustable. The optimization is mathematically identical to choosing arbitrary per-element phase shifts on a conventional array. The paper compares only 'no FIM' and 'random FIM', never a phase-controlled array with the same number of phase-only DoFs. So the reported 2.5 dB and 2 dB rate gains and the sharper MUSIC peaks are real, but they are not evidence for FIM-specific physics. A same-DoF phase-shifter baseline would settle this. Footnote 3 covers delay and Doppler variations only; mutual coupling, pattern distortion, and mechanical coupling of a continuous elastic surface remain unmodeled, so the physical realizability of independently optimized y values is open. The reader's CONDITIONAL verdict is the right one.\n\nOther soft spots are real but minor. The gradient computation implicitly assumes perfect knowledge of all path parameters; the transmit covariance is fixed to identity; the sensing constraint is a total received power proxy; there are no error bars; and the paper itself admits the achievable rates are nearly identical across waveforms, which undercuts the waveform comparison but not the FIM optimization itself. The heavy self-citation is acceptable here because the prior DD and SIM results are the actual substrate, not padding.\n\nWho is this for? Researchers working on metasurface-aided MIMO or delay-Doppler ISAC models. It is a useful incremental contribution with honest limitations. I would not cite it in my own work, but I would bring it to a reading group discussion about what FIM modeling actually adds. A serious referee should engage: require a same-DoF phase-array baseline or a reframed claim, and ask for a paragraph on physical realizability. That is addressable without throwing out the paper.","headline":"Solid incremental extension of the authors' DD-metasurface line, but the headline FIM gains are not isolated from generic per-element phase control.","tokens_in":19682,"tokens_out":1476,"would_cite":false,"duration_ms":16129,"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 claims that treating the vertical positions of flexible-metasurface elements as design variables improves both communication rate and sensing accuracy in high-mobility MIMO integrated sensing and communications.","keywords":["flexible intelligent metasurfaces","doubly dispersive channel","MIMO","integrated sensing and communications","OFDM","OTFS","AFDM","achievable rate"],"falsifier":"Take a small morphing element or a full-wave model of one, move its position over the range $\\pm\\lambda$, and compare the measured or simulated complex channel against the phase-only prediction $e^{j2\\pi y_b\\sin\\theta\\sin\\phi/\\lambda}$. If amplitude variations, delay shifts, or pattern changes appear at a level that affects the channel, the reported rate and sensing gains from shape optimization will not transfer to hardware.","tokens_in":18711,"feed_emoji":"📡","tokens_out":6280,"duration_ms":64553,"temperature":0.7,"pith_summary":"Flexible intelligent metasurfaces (FIMs) can bend their elements out of the array plane, and this paper argues that those element positions are a usable design resource rather than a mechanical nuisance. It builds a doubly dispersive MIMO channel model in which each FIM element's vertical coordinate $y_b$ enters only through a phase factor in the array response, and it derives input-output relations for OFDM, OTFS, and AFDM under that model. The paper then optimizes the vertical positions of transmitter and receiver FIMs with gradient ascent, subject to a sensing power constraint, and reports about 2.5 dB of achievable-rate gain from random FIM shapes and roughly another 2 dB from optimized shapes. The same optimization sharpens MUSIC angle-of-arrival spectra, so a single FIM deformation serves both communication and sensing.","feed_headline":"Morphing metasurface positions add ~4.5 dB and sharpen sensing","feed_subtitle":"A shape-aware channel model turns a flexible array's bends into rate and angle-estimation gains for OFDM, OTFS, and AFDM.","key_machinery":"The load-bearing object is the FIM array response vector $\\mathbf{b}(y,\\phi,\\theta)$, whose phase depends on each element's position $y$ along the morphing axis, combined with the effective MIMO channel $\\bar{\\mathbf{H}}(y_T,y_R)=\\sum_p \\check{\\mathbf{H}}_p\\otimes \\mathbf{G}_p$, where $\\check{\\mathbf{H}}_p$ carries the FIM-dependent array outer products and $\\mathbf{G}_p$ is the waveform-specific delay-Doppler matrix for OFDM, OTFS, or AFDM. This separable structure lets the gradient of the achievable rate with respect to each $y$ coordinate be written in closed form, so a simple gradient-ascent update with a line search can deform the surfaces. The morphing range $y_{\\min}\\le y_b\\le y_{\\max}$ is the resource being allocated.","core_discovery":"The central claim is that the geometric shape of a FIM is itself an optimization variable that materially changes MIMO performance in fast, frequency-selective channels. The paper's FPDD channel model writes the channel as a sum over paths of outer products of FIM array responses, with the $b$-th element's response carrying the factor $e^{j2\\pi y_b \\sin\\theta \\sin\\phi/\\lambda}$; because this factor depends on the vertical position $y_b$, the surface shape enters every waveform's effective channel through the same array response. Solving the achievable-rate maximization with the sensing constraint $\\operatorname{tr}(\\bar{H}(y_T,y_R)T\\bar{H}^H(y_T,y_R))\\ge \\Psi$ by gradient ascent yields the reported rate gains and cleaner MUSIC peaks for all three waveforms.","pith_inferences":["If the phase-only model is optimistic, real FIM prototypes may show smaller gains; a full-wave simulation or measured prototype that compares the modeled and actual channel as $y$ moves would settle this.","The shape optimization acts as a mechanically reconfigurable phase taper, so combining it with digital precoding or power allocation could push the rate further than the identity-covariance assumption used here.","Rate parity across waveforms is a mutual-information statement; practical bit-error-rate and inter-carrier-interference behavior would probably still favor OTFS and AFDM at high Doppler.","Because the paper does not quantify angular resolution or Cramér-Rao bounds, the sensing improvement is demonstrated qualitatively via spectra; a resolution-bound analysis would be a direct next step."],"forward_implications":["Any of the three waveforms inherits the FIM gain, since the shape enters through the physical array response rather than through the modulation.","Optimized FIM shapes can be computed from one channel realization and then reused when delays, Doppler shifts, or the waveform change.","A sensing power constraint can be enforced alongside the rate objective through a penalty term without abandoning closed-form gradients.","Larger morphing ranges and more elements give more degrees of freedom, so the gain scales with the FIM's flexibility and size.","MUSIC direction finding benefits even though the optimization objective is communication rate, because the optimized response separates scatterer peaks."],"supporting_citations":[{"why":"supplies the doubly dispersive channel input-output relations for OFDM, OTFS, and AFDM that the FPDD model extends.","marker":"[13]"},{"why":"defines the FIM element geometry, morphing range, and the phase-only array response used in Eq. (3).","marker":"[43]"},{"why":"establishes the FIM MIMO capacity gains in flat fading that motivate the rate optimization.","marker":"[37]"},{"why":"demonstrates FIM shape optimization gains for wireless sensing, motivating the sensing QoS constraint.","marker":"[38]"},{"why":"prior metasurface-parameterized doubly dispersive MIMO model that this paper extends to flexible metasurfaces.","marker":"[39]"},{"why":"companion doubly dispersive MIMO model with stacked intelligent metasurfaces that supplies the parametrization approach.","marker":"[40]"},{"why":"provides the MUSIC procedure used to evaluate angle-of-arrival sensing performance.","marker":"[53]"},{"why":"defines OTFS modulation and its Zak-transform structure used in the OTFS effective channel derivation.","marker":"[14]"},{"why":"defines the AFDM chirp parameter model used in the AFDM input-output relation.","marker":"[49]"}],"fun_headline_variants":["Bend the metasurface to improve MIMO ISAC rate and sensing","Shape-aware FIM model boosts rate and sensing for ISAC waveforms","Flexible metasurface geometry as a knob for high-mobility MIMO","Tuning surface bends lifts rate and angle estimation in DD channels","Metasurface shape optimization yields gains for OFDM, OTFS, AFDM"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument stands on the premise that moving a FIM element out of the plane changes only the phase of that element's contribution, leaving delays, Doppler shifts, radiation patterns, mutual coupling, and near-field behavior exactly as they were.","fun_headline_variants_meta":{"raw":{"variants":["Bend the metasurface to improve MIMO ISAC rate and sensing","Shape-aware FIM model boosts rate and sensing for ISAC waveforms","Flexible metasurface geometry as a knob for high-mobility MIMO","Tuning surface bends lifts rate and angle estimation in DD channels","Metasurface shape optimization yields gains for OFDM, OTFS, AFDM"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000328,"raw_usage":{"total_tokens":1807,"prompt_tokens":895,"completion_tokens":912,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":511,"completion_tokens_details":{"reasoning_tokens":817}},"tokens_in":511,"tokens_out":912,"duration_ms":9390,"temperature":1.0,"reasoning_tokens":817,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:09:17.830706+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a small morphing element or a full-wave model of one, move its position over the range $\\pm\\lambda$, and compare the measured or simulated complex channel against the phase-only prediction $e^{j2\\pi y_b\\sin\\theta\\sin\\phi/\\lambda}$. If amplitude variations, delay shifts, or pattern changes appear at a level that affects the channel, the reported rate and sensing gains from shape optimization will not transfer to hardware.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the doubly dispersive channel input-output relations for OFDM, OTFS, and AFDM that the FPDD model extends."},{"cited_title":"Flexible intelligent metasurfaces for downlink multiuser miso communications,","cited_arxiv_id":null,"evidence_quote":"defines the FIM element geometry, morphing range, and the phase-only array response used in Eq. (3)."},{"cited_title":"Flexible intelligent metasurfaces for enhancing MIMO communications,","cited_arxiv_id":null,"evidence_quote":"establishes the FIM MIMO capacity gains in flat fading that motivate the rate optimization."},{"cited_title":"Flexible intelligent metasurface for enhancing multi- target wireless sensing,","cited_arxiv_id":null,"evidence_quote":"demonstrates FIM shape optimization gains for wireless sensing, motivating the sensing QoS constraint."},{"cited_title":"Metasurfaces-Integrated Doubly-Dispersive MIMO: Channel Modeling and Optimization","cited_arxiv_id":"2506.14985","evidence_quote":"prior metasurface-parameterized doubly dispersive MIMO model that this paper extends to flexible metasurfaces."},{"cited_title":"Fast and efficient sequential radar parameter estimation in MIMO-OTFS systems,","cited_arxiv_id":null,"evidence_quote":"provides the MUSIC procedure used to evaluate angle-of-arrival sensing performance."},{"cited_title":"Orthogonal time frequency space modulation,","cited_arxiv_id":null,"evidence_quote":"defines OTFS modulation and its Zak-transform structure used in the OTFS effective channel derivation."}],"review_version":2}