{"id":"89dadb3d-e67c-4c9a-9a7c-dd50c317224a","arxiv_id":"2611.16411","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A flexible intelligent metasurface at the transmitter improves secrecy rate by up to 21% and saves 2.5 dB transmit power compared to fixed arrays, through joint optimization of transmit covariance and surface shape.","lead":"This paper proposes using a flexible intelligent metasurface (FIM) as a transmitter to improve the secrecy of wireless communications by dynamically reshaping the array geometry. Secure multiple-input single-output (MISO) transmission is considered in the presence of a multi-antenna eavesdropper, and joint optimization of the transmit signal and surface shape is shown to outperform conventional fixed arrays.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"Lemma 1's rank-one optimality citation may not cover multi-antenna eavesdropper, which is central to the optimization framework.","rationale":"The reader identified perfect CSI and independent morphing as the weakest assumptions, both of which the paper acknowledges. However, the more fundamental structural risk is that Lemma 1, which every algorithmic block (W update, BCA, closed-form for LoS) depends on, may not be valid for the multi-antenna eavesdropper case considered. The citation [11] is known to address only single-antenna Eve, and the paper does not provide a derivation for the general MIMO eavesdropper scenario. This is not an assumption that the paper acknowledges—it is a gap in the logical chain. If the lemma holds, the paper’s contributions stand; if it does not, the entire optimization framework collapses. Therefore, the verdict should be CONDITIONAL: accept only if the authors demonstrate that Lemma 1 applies to Ne>1 (by direct proof or appropriate citation), ensuring the theoretical foundation is sound. The concrete test is straightforward and can be satisfied without rerunning simulations.","tokens_in":12110,"tokens_out":3703,"duration_ms":29126,"concrete_test":"Check the content of [11] (Li & Ma, IEEE Trans. Signal Process., 2011) to confirm whether its proof of rank-one optimality covers the case of a multi-antenna eavesdropper (Ne>1). If it does not, provide a self-contained proof or cite an alternative source (e.g., Loyka & Charalambous 2011, or matrix analysis) that establishes this result for Ne>1. Without this, the central claim that the proposed design is optimal for the given problem (including multi-antenna Eve) is unsupported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Lemma 1 states that for a given FIM shape d, the optimal transmit covariance matrix W has rank at most one, citing [11] (Li & Ma 2011). However, [11] considers a MISO wiretap channel with a single-antenna eavesdropper, whereas the system model in this paper includes a multi-antenna eavesdropper (Ne≥1). The extension to Ne>1 is nontrivial because the eavesdropper's rate involves log det(I+G W G^H/σ²_e) rather than log(1+·). The paper provides no derivation and does not validate that the rank-one property holds for this general case. Since Lemma 1 is the foundation for the closed-form W update in the BCA algorithm and for the optimality claims in Proposition 1, this citation gap creates a risk that the core optimization is unsupported. The authors acknowledge perfect CSI and independent morphing as limitations, but this citation issue is not mentioned.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"This paper studies flexible intelligent metasurface (FIM)-aided secure MISO communications in the presence of a multi-antenna eavesdropper. A secrecy rate maximization problem is formulated by jointly optimizing the transmit covariance matrix and the FIM surface shape under power and morphing-range constraints. For a given FIM shape, the optimal transmit covariance is claimed to be rank-one (Lemma 1, citing [11]). For a fixed covariance, a projected gradient ascent method optimizes the shape, and a block-coordinate-ascent algorithm is proposed. For pure line-of-sight channels, a globally optimal closed-form design is derived under a sufficient morphing-range condition. Numerical results show secrecy-rate improvements over conventional rigid array benchmarks, including a 2.5 dB transmit-power saving at a given secrecy rate.","tokens_in":12425,"tokens_out":2972,"duration_ms":22869,"significance":"If the claims hold, this work would be the first application of flexible intelligent metasurfaces (FIM) to physical-layer security, introducing a new geometry-domain degree of freedom to combat multi-antenna eavesdropping. The closed-form solution for the LoS case is elegant and provides clear design insight. The numerical evidence of secrecy-rate improvement over fixed arrays is convincing for the considered scenarios, and the transmit-power saving of 2.5 dB is practically relevant.","major_comments":[{"comment":"Lemma 1 states that for a given FIM shape d, the optimal transmit covariance matrix W is rank-one, citing reference [11] (Li & Ma 2011). However, [11] considers a MISO wiretap channel with a single-antenna eavesdropper, whereas this paper's system model includes a multi-antenna eavesdropper (Ne ≥ 1). The eavesdropper's rate then involves log det(I + G W G^H/σ_e^2) rather than log(1+·), and the extension of the rank-one optimality property is nontrivial. The paper provides no derivation or additional justification for this extension. Since Lemma 1 is the foundation for the closed-form W update in the BCA algorithm and underpins the optimality claims in Proposition 1 (LoS), the core optimization framework is unsupported until the rank-one property is established for Ne ≥ 1. The authors should either prove this claim directly or cite a reference that covers the multi-antenna eavesdropper case.","section":"Section III-A, Lemma 1","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"Line 3: 'eavesdro pper' contains a spurious space; should be 'eavesdropper'.","section":"Abstract","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The expression uses 'sinθ_{tE} sinϕ_{tE} − sinθ_B sinϕ_B', but earlier in the paper the elevation angles for Eve are denoted ϕ_{tE} (or ϕ_{tE,ℓE}) and for Bob ϕ_B. Ensure consistent use of subscripts for clarity.","section":"Section IV, at definition of Δ_{b,e}","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The axis label in Fig. 2(b) is missing; state the unit (λ) explicitly in the caption or ensure the figure has the correct label.","section":"Figure 2 caption","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The beampattern figures are very informative, but the colorbar limits (−60 to 20 dB) and the (a), (b), (c) labels could be slightly larger for readability.","section":"Section V-C, Fig. 3 caption","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The factor 1/ln2 appears correctly, but the notation 'tr (Q^{-1}_e ∂Q_e/∂dn)' would benefit from a brief reminder that Q_e depends on d through G(d). The current derivation is otherwise clear.","section":"Appendix A, Eq. (15)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The parentheses in the complexity expression 'O(I_BCA[N_t^3 + N_e N_t^2 + I_PGA(N_e^3 + N_t N_e (N_t + N_e + L_E) + N_t L_B)])' are correctly matched, but the term 'N_t N_e (N_t + N_e + L_E)' could be broken into factors for easier parsing; this is a minor presentation point.","section":"Complexity expression, end of Section III-C","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"This paper does something new: it brings flexible intelligent metasurfaces into physical-layer security, formulating a joint transmit covariance and surface-shape optimization for a MISO link with a multi-antenna eavesdropper. The math is clean, the LoS closed-form (Proposition 1) is a genuine insight, and the 2.5 dB / 21% gains over a rigid array are credible. The BCA algorithm is standard but competently applied. I think the core idea is worthwhile and the execution is careful.\n\nThe main soft spot is Lemma 1. The paper cites [11] (Li & Ma 2011) for the rank-one optimality of the transmit covariance matrix under a fixed surface shape. But [11] treats a MISO wiretap channel with a single-antenna eavesdropper; here the eavesdropper has multiple antennas, so the second term is log det(I + G W G^H / σ_e^2) rather than log(1 + ·). Whether the rank-one property still holds in this general case is not obvious, and the paper provides no derivation or reference to support the extension. Since Lemma 1 is used in the BCA algorithm and is relied on by the optimality claims, this is a load-bearing gap. The authors acknowledge perfect CSI and independent morphing as limitations, which is good, but they don't flag this citation issue. The stress-test note is correct: this needs justification. If the property is known to hold, a citation to a proof for Ne ≥ 1 should be given; if not, a short proof is needed.\n\nEverything else looks fine. The perfect CSI assumption is stated and defended as a fundamental-secrecy study; the independent morphing model is a limitation but is honestly discussed. The simulation setup is standard and the results are internally consistent.\n\nWho is this for? Readers working on FIM applications, or on geometry-based PLS, will want to see it. It is not a breakthrough but a well-executed extension. I would send it to review, but I would ask the referees to check the Lemma 1 issue and request a clarification or a proof. If the authors can supply that, the paper is ready. If not, the core optimization is unsupported.","headline":"Applies FIM to physical layer security with solid optimization, but the rank-one citation for multi-antenna Eve needs a closer look.","tokens_in":13594,"tokens_out":686,"duration_ms":22302},"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"Morphable metasurfaces improve wireless secrecy and save 2.5 dB in transmit power.","keywords":["Flexible intelligent metasurface","Physical layer security","MISO","Secrecy rate maximization","Surface shape morphing","Beamforming","Line-of-sight","Block coordinate ascent"],"falsifier":"In a controlled LoS experiment with $N_t \\ge 2$, $d_{\\max} \\geq \\lambda/2|\\Delta_{b,e}|$, and perfect CSI, if the FIM-secure scheme does not achieve a secrecy rate within measurement uncertainty of $\\log_2(1 + P N_t |\\alpha_b|^2 / \\sigma_b^2)$ while the RAA benchmark shows significant leakage, then the decorrelation claim would be falsified.","tokens_in":13651,"feed_emoji":"🔒","tokens_out":3339,"duration_ms":25496,"temperature":0.7,"pith_summary":"This paper proposes using a flexible intelligent metasurface (FIM) as a transmitter to combat multi-antenna eavesdropping by dynamically reshaping its three-dimensional surface. The authors formulate a secrecy rate maximization problem that jointly optimizes the transmit covariance matrix and the FIM element displacements. They develop a block coordinate ascent algorithm that alternates between a closed-form rank-one covariance update and a projected gradient ascent for the surface shape. For the special case of line-of-sight propagation, they derive a globally optimal closed-form solution under a sufficient morphing-range condition. Numerical results demonstrate that the FIM architecture consistently outperforms a conventional rigid antenna array, achieving a transmit-power saving of about 2.5 dB at the same secrecy rate.","feed_headline":"Morphable array cuts power by 2.5 dB in secure wireless link","feed_subtitle":"Flexible intelligent metasurfaces dynamically reshape channels, outperforming rigid antennas in secrecy rate.","key_machinery":"The key mechanism is the FIM's morphing-induced phase response: each element's displacement $d_n$ along the y-axis adds a phase term $-k_c d_n \\sin\\theta\\sin\\phi$ to the array response, making the channel dependent on the surface shape vector $\\mathbf{d}$. The optimization framework alternates between two subproblems: (1) for fixed $\\mathbf{d}$, the optimal transmit covariance matrix $\\mathbf{W}^\\star$ is rank-one and given by the dominant generalized eigenvector of $(\\mathbf{I} + P\\sigma_b^{-2}\\mathbf{h}\\mathbf{h}^H,\\, \\mathbf{I} + P\\sigma_e^{-2}\\mathbf{G}^H\\mathbf{G})$; (2) for fixed $\\mathbf{W}$, the gradient of the secrecy rate with respect to $\\mathbf{d}$ is computed and used in a projected gradient ascent with Armijo backtracking. For the LoS scenario, a closed-form displacement pattern $d_n^\\star = (\\psi_n - \\delta_n + 2\\pi k_n)/(k_c \\Delta_{b,e})$ achieves orthogonality between the legitimate and eavesdropping array responses when the morphing range satisfies $d_{\\max} \\geq \\lambda / 2|\\Delta_{b,e}|$.","core_discovery":"The core discovery is that an FIM's ability to dynamically morph each radiating element along the y-axis introduces direction-dependent phase shifts that actively decorrelate the legitimate and eavesdropping channels. For a given surface shape, the optimal transmit covariance matrix is rank-one and obtained from the dominant generalized eigenvector. For a fixed covariance matrix, the surface shape is optimized via projected gradient ascent. Crucially, under pure line-of-sight conditions and a sufficient morphing range, the paper provides a closed-form joint design that achieves perfect nulling toward the eavesdropper while maintaining full beamforming gain toward the legitimate user, yielding the secrecy rate upper bound of $\\log_2(1 + P N_t |\\alpha_b|^2 / \\sigma_b^2)$.","pith_inferences":["The FIM concept could be extended to multi-user or multi-eavesdropper scenarios, where the surface shape is optimized to create distinct nulls toward several directions simultaneously.","Practical implementation must account for mechanical coupling between elements and actuator precision; the paper's independent-element assumption provides an upper bound that real systems may approach.","Combining FIM morphing with reconfigurable intelligent surfaces (RIS) on the receiver side might yield further hybrid gains in both security and link quality.","The derived morphing-range condition $d_{\\max} \\geq \\lambda / 2|\\Delta_{b,e}|$ could serve as a design guideline for physical FIM fabrication in LoS deployments."],"forward_implications":["FIM-aided transmitters can achieve higher secrecy rates than conventional rigid antenna arrays under the same total transmit power budget.","The closed-form LoS solution enables low-complexity, globally optimal secure transmission when the morphing range is sufficient, avoiding iterative numerical optimization.","A moderate morphing range (e.g., $0.5\\lambda$) captures most of the secrecy gain; further increasing the range yields diminishing returns after about 21% improvement.","The power saving of approximately 2.5 dB at equal secrecy rate suggests that FIMs can extend battery life or reduce interference in secure wireless systems."],"supporting_citations":[{"why":"Supplies the FIM array response model and morphing mechanism that the paper directly adopts for channel modeling.","marker":"[7]"},{"why":"Provides the optimal rank-one transmit covariance solution for MISO secrecy rate maximization, used in Lemma 1.","marker":"[11]"},{"why":"Provides background and foundational principles of flexible intelligent metasurfaces that motivate the system architecture.","marker":"[6]"},{"why":"Supplies the complex-valued matrix differentiation techniques used to derive the gradient for the FIM surface shape optimization.","marker":"[12]"}],"fun_headline_variants":["Morphable metasurface saves 2.5 dB power for secure MISO link","Flexible surface reshapes channels, cuts eavesdropper gain","Shape-shifting array nulls eavesdropper, saves power","Morphable antenna cuts 2.5 dB in secure wireless comms"],"cache_read_input_tokens":0,"weakest_assumption_plain":"Alice has perfect knowledge of both the legitimate channel $\\mathbf{h}(\\mathbf{d})$ and the eavesdropper channel $\\mathbf{G}(\\mathbf{d})$, which is idealistic but necessary for the optimization framework; in practice, channel estimation errors and limited feedback would degrade performance.","fun_headline_variants_meta":{"raw":{"variants":["Morphable metasurface saves 2.5 dB power for secure MISO link","Flexible surface reshapes channels, cuts eavesdropper gain","Shape-shifting array nulls eavesdropper, saves power","Morphable antenna cuts 2.5 dB in secure wireless comms"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":0.000265,"raw_usage":{"cost":0.000265328,"is_byok":false,"cost_details":{"upstream_inference_cost":0.000265328,"upstream_inference_prompt_cost":0.000100968,"upstream_inference_completions_cost":0.00016436},"total_tokens":1513,"prompt_tokens":926,"completion_tokens":587,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":256,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":507}},"tokens_in":926,"tokens_out":587,"cost_source":"openrouter_usage","duration_ms":5189,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":507,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-13T15:11:46.581810+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"In a controlled LoS experiment with $N_t \\ge 2$, $d_{\\max} \\geq \\lambda/2|\\Delta_{b,e}|$, and perfect CSI, if the FIM-secure scheme does not achieve a secrecy rate within measurement uncertainty of $\\log_2(1 + P N_t |\\alpha_b|^2 / \\sigma_b^2)$ while the RAA benchmark shows significant leakage, then the decorrelation claim would be falsified.","supporting_citations":[{"cited_title":"Flexible intelligent metasurfaces for downlink multiuser MISO communications,","cited_arxiv_id":null,"evidence_quote":"Supplies the FIM array response model and morphing mechanism that the paper directly adopts for channel modeling."},{"cited_title":"Optimal and robust transmit designs for MISO channel secrecy by semideﬁnite programming,","cited_arxiv_id":null,"evidence_quote":"Provides the optimal rank-one transmit covariance solution for MISO secrecy rate maximization, used in Lemma 1."},{"cited_title":"Emergi ng technologies in intelligent metasurfaces: Shaping the fut ure of wireless communications,","cited_arxiv_id":null,"evidence_quote":"Provides background and foundational principles of flexible intelligent metasurfaces that motivate the system architecture."},{"cited_title":"Complex-valued matrix d ifferentiation: Techniques and key results,","cited_arxiv_id":null,"evidence_quote":"Supplies the complex-valued matrix differentiation techniques used to derive the gradient for the FIM surface shape optimization."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}