{"id":"607eaefd-d20d-4353-bbcf-30fb45ae469b","arxiv_id":"2411.09235","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A fluid-antenna transmitter that jointly optimizes antenna positions and beamforming improves secrecy rate under a covertness constraint compared with fixed-position antennas.","lead":"Fluid antennas can move within a small region, and this paper uses that movement to send messages that resist both eavesdropping and detection by a warden. The authors propose a joint antenna-position and beamforming optimization and show simulations where it beats fixed-antenna baselines.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The position subproblem uses Φ_k=Σ_k^HΣ_k in Eq. (18), but the paper's own channel model requires Φ_k=Σ_k^H11^HΣ_k; the MM bounds do not bound Tr(H_kV), so the claimed FAS gains are not backed by the stated algorithm.","rationale":"The reader's conditional verdict identifies the phase-only far-field model as the weakest assumption and also flags Eq. (29) as a clear error. I find Eq. (29) correct: for a=t_n^(m)-t_v, the first-order lower bound of the convex norm ‖x‖ is (a^T x)/‖a‖. The more serious issue is internal to the derivation: the expansion of Tr(H_kV) in Eq. (18) uses Φ_k=Σ_k^HΣ_k, but with h_k^H=1^HΣ_kF_k the correct quadratic form is F_k^HΣ_k^H11^HΣ_kF_k. This is not a modeling choice; it is an algebraic mismatch with the paper's own channel equation. Because every lower and upper bound in the position subproblem is built from the incorrect Φ_k, the constraints (26)-(28) are not valid surrogates for (9), (11b), and (11c). The algorithm, as written, may solve a different problem, so the simulated superiority over FPA is not attributable to a feasible solution of Problem (10). This defect is checkable and likely fixable, so I would not escalate to REJECT: the paper's idea remains coherent, but the derivations and simulations must be corrected and ideally accompanied by code. Since the final verdict remains CONDITIONAL, I leave the reader's verdict unchanged, though my reason differs from the reader's stated weakest assumption.","tokens_in":8647,"tokens_out":12626,"duration_ms":142628,"concrete_test":"With L_r=L_t=4 and a single non-zero beamforming entry v_n, compute the true value |1^HΣ_kf_k(t_n)|^2|v_n|^2 and the paper's β_k(t_n)=|v_n|^2 f_k(t_n)^HΣ_k^HΣ_kf_k(t_n) for a random Σ_k. If they differ, replace Φ_k by Σ_k^H11^HΣ_k throughout Section III-B and rerun the exact Figs. 1-2 simulations; verify that the corrected sequence still satisfies (9) at every iteration and that the secrecy-rate gap over FPA remains. If it does, the error is a typo; if not, the FAS gains are an artifact of the incorrect surrogate.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing defect is in the algebra of Section III-B, not in the far-field phase-only assumption or in Eq. (29), which is a valid first-order lower bound for the convex norm. The channel model gives h_k^H=1^HΣ_kF_k(¯t), so Tr(H_kV)=Tr(VF_k^HΣ_k^H11^HΣ_kF_k). Equation (18) and all subsequent MM bounds instead use Φ_k=Σ_k^HΣ_k. These two forms agree only when L_r=1; the simulations use L_r=L_t=4. Thus α_k, β_k(t_n), Ω_k, Ψ_k, and the bounds (19)-(25) are not bounds on the actual Tr(H_kV). Consequently, the convex surrogate constraints (26)-(28) do not enforce the covertness constraint (9) or the auxiliary constraints (11b)-(11c). Solving the position subproblem (30) as written can produce positions that do not satisfy Problem (10). Since the claimed FAS gains in Figs. 1-2 come from this AO loop, the algorithmic evidence for the central claim is invalid as written unless this is corrected.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This letter studies a fluid-antenna-system (FAS)-aided transmitter that serves a legitimate receiver while an eavesdropper overhears and a warden tries to detect the transmission. The authors formulate a secrecy-rate maximization problem subject to a covertness constraint derived from Pinsker's inequality, a transmit power constraint, antenna minimum-distance constraints, and a rank-one beamforming constraint. They propose an alternating optimization algorithm that uses a penalty-based convex update for the beamforming matrix and a majorization-minimization (MM) update for the antenna positions. Simulation results in Figs. 1 and 2 compare the proposed scheme with fixed-position-antenna (FPA), random-position-antenna (RPA), and exhaustive-antenna-selection (EAS) benchmarks and report significant gains from the fluid antenna system.","tokens_in":8887,"tokens_out":14739,"duration_ms":161227,"significance":"If the algebraic derivations were correct, the paper would provide a reasonable first extension of FAS to the joint secure-and-covert setting, with a transparent coordinate-descent structure and comparisons against external benchmarks rather than fitted data. The MM and penalty elements are standard, and the problem formulation is sensible. However, the central identity used for the position subproblem in Section III-B is incorrect as written: the matrix Φ_k used throughout the MM bounds is not the matrix that appears in Tr(H_kV) under the paper's own channel model. This error invalidates the feasibility guarantees of the surrogate constraints (26)-(28) and therefore the simulation evidence for the claimed FAS gains. The error is localized and appears fixable by replacing Φ_k with the correct Gram-type matrix, but the numerical results would need to be regenerated before the central claim can be accepted.","major_comments":[{"comment":"Eq. (18) sets Φ_k=Σ_k^HΣ_k and expands Tr(H_kV) as α_k+β_k(t_n)+2Re{f_k^H(t_n)Ω_k}. With the channel h_k^H=1^HΣ_kF_k(¯t) from Eq. (3), however, Tr(H_kV)=||1^HΣ_kF_k(¯t)v||^2=v^H F_k^H(¯t)Σ_k^H 1 1^H Σ_k F_k(¯t)v. The matrix that must appear in the expansion is therefore Φ_k=Σ_k^H11^HΣ_k, not Σ_k^HΣ_k. These two choices agree only when L_r^k=1, whereas Section IV explicitly sets L_r^k=L_t^k=4. Consequently, α_k, β_k(t_n), Ω_k, Ψ_k, and the bounds in (19)-(25) are not bounds on the actual Tr(H_kV), and the surrogate constraints (26)-(28) do not enforce the covertness constraint (9) or the auxiliary constraints (11b)-(11c). Solving the position subproblem (30) as written can produce antenna positions that violate Problem (10). This is a load-bearing error for the claimed FAS gains in Figs. 1 and 2; the definition of Φ_k should be corrected and the simulations rerun.","section":"III-B, Eq. (18)"},{"comment":"The partial derivative with respect to y_t^n is printed with a factor cos φ^t_{k,l}; from the definition ρ_{k,l}(t_n)=x_t^n sinθ^t_{k,l} cosφ^t_{k,l}+y_t^n cosθ^t_{k,l} in Section II, the y-derivative of the phase is proportional to cosθ^t_{k,l}, not cosφ^t_{k,l}. Since ∇β_k(t_n) is used in both the lower bound (20) and the upper bound (24), the printed gradient would invalidate the MM inequalities even after the Φ_k correction. Please correct the formula and verify the signs and coefficients in (32) and (33).","section":"Appendix A, Eq. (33)"}],"minor_comments":[{"comment":"In the middle term of the right-hand side of Eq. (18), the subscript is printed as β_b(t_n); it should be β_k(t_n) for consistency with the surrounding notation.","section":"III-B, Eq. (18)"},{"comment":"The notation v(n) and v_n is used interchangeably for the elements of the beamforming vector; please unify the notation.","section":"III-B"},{"comment":"Eq. (29) is dimensionally consistent as printed: the denominator ||t_n^m−t_v|| makes the left-hand side a scalar distance, which is then compared with D. No correction is needed here, but the sentence could clarify that this is the standard first-order lower bound of the convex norm ||t_n−t_v||.","section":"III-B, Eq. (29)"},{"comment":"The y-axis labels in Figs. 1 and 2 are given as 'Secrecy rate' without units; bits/s/Hz would be the conventional unit for the secrecy rate.","section":"IV"},{"comment":"The paper states only the convergence accuracy 10^{-4} and does not describe the stopping criterion for the AO loop or the penalty update rule. A brief statement on the convergence criterion would improve reproducibility.","section":"III"}],"recommendation":"major_revision","confidential_remarks":"The issue in Eq. (18) is a genuine algebraic error in a load-bearing part of the derivation, not a matter of presentation or scope. The framework appears repairable, but the numerical claims must be regenerated after replacing Φ_k with Σ_k^H11^HΣ_k; if the corrected simulations no longer show consistent FAS gains, the paper's central claim would need to be weakened accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper asks a genuinely good question—can fluid antennas help with secure and covert transmission—and the overall AO structure is sensible. But the position-subproblem algebra in Section III-B contains a wrong quadratic form that breaks the MM bounds, so the simulation results as presented do not support the central claim.\n\nWhat is new and good: combining FAS with a warden's detection constraint under a secrecy-rate objective is a natural and timely problem. The DEP bound via Pinsker's inequality is standard but correctly applied. The AO split into a penalty-based beamforming update and an MM-based position update is reasonable, and the benchmarks (FPA, RPA, EAS) are appropriate. The paper is clearly written and the literature is cited fairly.\n\nThe serious issue is in Eq. (18). The channel model gives h_k^H = 1^H Σ_k F_k, so Tr(H_k V) = Tr(V F_k^H Σ_k^H 1 1^H Σ_k F_k). The derivation instead uses Φ_k = Σ_k^H Σ_k in α_k, β_k, and Ω_k. These two forms differ unless L_r = 1, and the simulations set L_r = 4. Consequently, the bounds (19)–(25) and the surrogate constraints (26)–(28) are not bounds on the actual Tr(H_k V). Solving the position subproblem (30) as written can violate the covertness constraint and produce positions that do not solve Problem (10). The reported FAS gains in Figs. 1–2 come from this loop, so the numerical evidence is invalid as stated.\n\nThe reader's concern about Eq. (29) is a false alarm: that first-order linearization of the convex distance constraint is dimensionally consistent and mathematically correct. The Appendix A typos (\"dose not\", \"respre-sented\", the y_tm subscript) are minor and easily fixed. Lack of simulation code is a smaller issue but would help reproducibility.\n\nThe far-field phase-only model is an idealization, but it is a standard FAS assumption and not a flaw by itself.\n\nBottom line: the paper deserves a serious referee, but only after the authors fix the Φ_k error and rerun the experiments. If the corrected results still show gains over FPA/EAS, this would be a useful letter. In its current form, I would not trust the numbers.\n\nRecommendation: send to peer review with a clear request for major revision; the idea is worth engaging with, and the fix is straightforward.","headline":"A promising FAS secrecy/covertness formulation is undermined by a wrong quadratic form in the position subproblem; the reported gains are not yet supported, but the idea is sound and fixable.","tokens_in":9425,"tokens_out":4070,"would_cite":false,"duration_ms":42012,"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":"Fluid antennas can raise the secrecy rate of a covert wireless link by jointly optimizing beamforming and antenna positions, outperforming fixed-position arrays in simulation.","keywords":["fluid antenna system","covert communication","physical layer security","beamforming","alternating optimization","majorization-minimization","secrecy rate","position optimization"],"falsifier":"Measure the actual channel response of a physical fluid antenna as a function of position and compare the achieved secrecy rate under the optimized positions with the phase-only model's prediction; if the realized channels differ enough to break the assumed phase-only relationship, the algorithm's performance advantage over FPA would not reproduce.","tokens_in":8440,"feed_emoji":"📡","tokens_out":4410,"duration_ms":42025,"temperature":0.7,"pith_summary":"The paper proposes a fluid antenna system (FAS) transmitter whose multiple movable antennas are jointly optimized with the transmit beamforming to maximize the secrecy rate against an eavesdropper while keeping the transmission undetectable by a warden. The central claim is that adjusting antenna positions adds a spatial degree of freedom that significantly improves secure and covert communication performance compared to fixed-position antennas (FPA). The paper develops an alternating optimization algorithm that alternates between a penalty-based beamforming update and a majorization-minimization position update, and simulations show higher secrecy rates than FPA, random-position, and exhaustive-search benchmarks under equal power and covertness constraints. If correct, this establishes FAS as a practical physical-layer technique for communication that must remain both secret and hidden.","feed_headline":"Moving antennas lift secrecy while dodging detection","feed_subtitle":"Joint beamforming and antenna-position design keeps hidden transmissions faster than fixed-antenna setups.","key_machinery":"The central object is the transmit field response vector $f_k(t_n) = [e^{jrac{2\\pi}{\\lambda}\\rho_{k,1}(t_n)}, \\ldots]^T$ with $\\rho_{k,l}(t_n) = x_n\\sin\\theta^t_{k,l}\\cos\\phi^t_{k,l} + y_n\\cos\\theta^t_{k,l}$, which makes each antenna position affect the channel only through path phases. On this basis the algorithm alternates: a penalty-based semidefinite program relaxes the rank-one beamforming constraint, and an MM-based position update uses first-order and second-order Taylor expansions to build concave lower bounds for the legitimate link and convex upper bounds for the eavesdropper and warden links, ensuring each subproblem is convex and solvable by CVX.","core_discovery":"Under a planar far-field response model in which moving a fluid antenna changes only the phase of each multipath component, the paper shows that jointly designing the transmit beamforming vector and the fluid antennas' positions can maximize the secrecy rate subject to a covertness constraint expressed as a bound on the Kullback-Leibler divergence at the warden. The optimization problem is non-convex, so the paper decomposes it into subproblems: a penalty-based rank-one approximation handles beamforming, while majorization-minimization constructs concave lower bounds and convex upper bounds for the channel terms as functions of antenna position. Simulation results demonstrate that the proposed scheme outperforms FPA, RPA, and EAS benchmarks, with the gain growing in maximum transmit power and tolerated detection coefficient.","pith_inferences":["A natural testable extension is to replace the simulated channel model with measured or full-wave-computed channel responses from a physical fluid antenna prototype; the algorithm's structure would remain unchanged, but the phase-only assumption could be validated or refuted.","Because the covertness guarantee relies on a lower bound from Pinsker's inequality, a tighter bound or an exact detection-error-probability expression could change the feasible set and the optimal positions, so the design's margin under exact detection is open.","The same alternating optimization framework could apply to other reconfigurable geometries, such as movable antenna arrays at the receiver or intelligent reflecting surfaces, wherever channel phase is a tunable function of position.","The paper uses a local search, so the reported gains may depend on initialization; testing random restarts or a coarse grid of starting positions would reveal the robustness of the advantage over FPA."],"forward_implications":["For a fixed covertness requirement, adding movable antennas at the transmitter yields a higher secrecy rate than an equal-power fixed-position array, because position optimization creates channel diversity that can be steered against the eavesdropper.","The secrecy-rate gain of FAS over FPA grows with the maximum transmit power and with the tolerated detection coefficient, since both factors relax the constraints that position optimization exploits.","The proposed algorithm provides a concrete, converging design procedure: each iteration solves two convex programs, so system designers can compute locally optimal beamforming vectors and antenna positions without global search.","The covertness constraint is handled analytically through Pinsker's inequality and the Lambert W function, giving a closed-form upper bound on the warden's received power that any feasible solution must satisfy."],"supporting_citations":[{"why":"Supplies Pinsker's inequality used to convert the warden's detection error probability into a tractable KL-divergence constraint.","marker":"[4]"},{"why":"Gives the condition D(p0||p1) ≤ 2ε² that defines the feasible covertness region.","marker":"[5]"},{"why":"Provides the phase-only planar far-field model and the majorization-minimization bound used to solve the antenna position subproblems.","marker":"[19]"},{"why":"Introduces the fluid antenna system concept that the transmitter design is built on.","marker":"[9]"},{"why":"Is the convex solver used to compute both the beamforming update and the position update.","marker":"[21]"}],"fun_headline_variants":["Move antennas to beat eavesdroppers and wardens","Fluid antennas boost secrecy and covertness","Joint beam and position design for covert links","Adaptive antenna positions enhance covert rates"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The design assumes that moving a fluid antenna changes only the phase of each multipath component, leaving amplitudes and angles unchanged; if real antennas also change amplitude, mutual coupling, or near-field behavior, the secrecy and covertness gains shown in simulation may shrink.","fun_headline_variants_meta":{"raw":{"variants":["Move antennas to beat eavesdroppers and wardens","Fluid antennas boost secrecy and covertness","Joint beam and position design for covert links","Adaptive antenna positions enhance covert rates"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000321,"raw_usage":{"total_tokens":1742,"prompt_tokens":818,"completion_tokens":924,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":434,"completion_tokens_details":{"reasoning_tokens":868}},"tokens_in":434,"tokens_out":924,"duration_ms":95220,"temperature":1.0,"reasoning_tokens":868,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:53:13.166876+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual channel response of a physical fluid antenna as a function of position and compare the achieved secrecy rate under the optimized positions with the phase-only model's prediction; if the realized channels differ enough to break the assumed phase-only relationship, the algorithm's performance advantage over FPA would not reproduce.","supporting_citations":[{"cited_title":"Covert communic ation in cognitive radio networks with poisson distributed jamme rs,","cited_arxiv_id":null,"evidence_quote":"Supplies Pinsker's inequality used to convert the warden's detection error probability into a tractable KL-divergence constraint."},{"cited_title":"Intelligent reflecting surface-aided full-duplex cover t communications: Information freshness optimization,","cited_arxiv_id":null,"evidence_quote":"Gives the condition D(p0||p1) ≤ 2ε² that defines the feasible covertness region."},{"cited_title":"Flui d antenna system,","cited_arxiv_id":null,"evidence_quote":"Introduces the fluid antenna system concept that the transmitter design is built on."},{"cited_title":"CVX: MA TLAB software for discipli ned convex programming,","cited_arxiv_id":null,"evidence_quote":"Is the convex solver used to compute both the beamforming update and the position update."}],"review_version":1}