{"id":"aec6edaf-a051-4feb-8ded-5638594be4a8","arxiv_id":"2607.22141","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"LC-RIS phase shifts degrade with temperature; temperature-adaptive and temperature-robust designs keep the secrecy rate flat while designs that ignore temperature let it drop.","lead":"This paper models how temperature changes the phase shifts of liquid-crystal smart surface mirrors and designs anti-eavesdropping beam settings that hold as the temperature drifts. It matters because such mirrors are a leading low-power option for millimeter-wave networks, where temperature-induced phase errors can leak secret signals.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (16)'s multiplicative temperature scaling is used both as the design model and as the simulator's ground truth; the secrecy-rate gain is in-sample until validated with measured LC-RIS phase-vs-temperature data.","rationale":"The reader's weakest assumption is exactly the load-bearing concern I identify: the numerical validation uses Eq. (16) both for compensation and for simulating the true temperature effects, with no measured LC-RIS phase-versus-temperature data. This is the single biggest risk to the paper's central claim. I do not see an internal inconsistency or a fatal mathematical error; the optimization derivations are plausible, the code is public, and the reference-temperature voltage curve is adapted from experimental sources. However, the model's multiplicative temperature scaling is strong and unvalidated, and the performance claims rest on it. The SDP phase-uniformity relaxation (Lemma 1) is also self-admittedly heuristic, but it is less load-bearing than Eq. (16) because the scalable algorithm projects to feasible phases and the comparison in Fig. 8 could still be meaningful. Therefore, the appropriate verdict remains CONDITIONAL: accept only if the temperature model is validated against hardware data or an independent physical model. Since the reader already reached CONDITIONAL, I recommend no change to the verdict.","tokens_in":20446,"tokens_out":4488,"duration_ms":53285,"concrete_test":"In the Section VI simulator, replace the ground-truth temperature-affected phase response with measured phase-versus-voltage curves from a real LC-RIS phase shifter (or an independent LC electromagnetic model that does not assume Eq. (16)) at T = −20, −10, 10, 30, 40 °C, while keeping the algorithm's internal model as Eq. (16). Re-run Figs. 8–11. If the worst-case secrecy-rate gain over the 'neglected' design shrinks substantially, or if the robust design's secrecy rate is no longer flat across T, the central claim fails. As a direct check of Eq. (16), compute the ratio ω(V,T)/ω(V,Tr) from measured data: if it is not independent of V, the multiplicative scaling is falsified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that temperature-aware and temperature-robust LC-RIS phase-shift designs significantly improve worst-case secrecy rate over designs that neglect temperature. The entire optimization and the simulated 'actual' world rely on the same temperature model, Eq. (16): ω(V,T)=ω(V,Tr)·((Tc−T)/(Tc−Tr))^β. In Section VI, the temperature-dependent phase response used as ground truth is generated from this same Haller/Maier-Saupe scaling, with the same β, Tc, and Tr that the algorithms assume. Thus, the reported gains in Figs. 8–11 are in-sample: the compensator and the environment share the same multiplicative law. This matters because Eq. (16) is a strong structural assumption: it claims the normalized voltage-dependent phase profile is temperature-invariant, so all phase values compress by one scalar factor. Real LC cells have voltage-dependent director reorientation whose thermal behavior may not factor this way; the ordinary-index electrical length (the 'zero' baseline) also drifts with temperature and is not automatically calibrated out. The paper adapts experimental voltage data from [33], [34] at the reference temperature, but no measured LC-RIS phase-versus-temperature curves are reported to validate the multiplicative form. If the real temperature law differs—e.g., non-multiplicative, material-dependent, or with element-to-element variation—the claimed robustness and secrecy-rate gains are not established. This is a validation gap, not a proof error; the optimization machinery itself is reasonable and the code is available.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper addresses physical-layer security in an LC-RIS-aided mmWave downlink, where the RIS phase response depends on temperature. The authors first derive a temperature-dependent phase-shift model from the Maier–Saupe mean-field theory and the Haller empirical power law, yielding the multiplicative scaling in Eq. (16). They then formulate a worst-case secrecy-rate maximization problem in which only the spatial zones of the legitimate user and eavesdropper are known, not full CSI. Two phase-shift designs are proposed: a temperature-adaptive design (an SDP benchmark and a low-complexity scalable algorithm) and a temperature-robust design that operates without real-time temperature feedback. Simulations show that the temperature-aware and robust designs maintain a high worst-case secrecy rate as temperature varies, while the temperature-neglecting benchmark degrades away from the reference temperature.","tokens_in":20821,"tokens_out":6806,"duration_ms":80751,"significance":"If the underlying temperature model is trustworthy, the paper makes a useful contribution: it is, to my knowledge, the first treatment of temperature effects on LC-RIS phase shifts in a physical-layer-security context, and it addresses a practical scalability bottleneck by avoiding full CSI and by using a low-complexity, O(N) design. The public release of MATLAB code is a clear strength, as is the use of an SDP benchmark against which the scalable heuristic is compared. The physics-based derivation from the Maier–Saupe/Haller model is plausible and clearly presented. However, the quantitative central claim — that the proposed designs achieve large secrecy-rate gains over temperature-neglecting designs — is currently demonstrated only under the very model the algorithms assume. The paper would be significantly strengthened by measured LC-RIS phase-versus-temperature data or, failing that, by a systematic sensitivity analysis against alternative thermal laws. In its present form, the contribution is best read as a model-based design study whose experimental validation is still required.","major_comments":[{"comment":"The central load-bearing assumption is Eq. (16): ω(V,T)=ω(V,Tr)·((Tc−T)/(Tc−Tr))^β. In the simulations, the temperature-dependent phase response used as ground truth is generated from this same equation with the same β, Tc, and Tr that the algorithms assume. Consequently, Figs. 8–11 demonstrate in-sample performance: the compensator and the simulated environment share the same multiplicative law. The paper does not report any measured LC-RIS phase-versus-temperature curves, and the assumption that the normalized voltage-phase profile is temperature-invariant is not verified. A real LC cell may exhibit baseline (ordinary-index) drift, element-to-element variation, or a non-multiplicative temperature law. Please provide experimental validation, or at minimum a sensitivity analysis over alternative thermal models and over the stated parameter ranges (β=0.2–0.25, Tc around 95°C), before clai","section":"Section III-B, Eq. (16); Section VI-A"},{"comment":"The SDP method replaces the phase-range constraint C2 with the convex constraint cC2 using Lemma 1, which relies on the phases being approximately uniformly distributed over [0,ωmax] so that a law-of-large-numbers approximation holds. The authors concede in the text that \"a uniform distribution is not generally guaranteed\" and instead state, based on observations, that N≥50 is sufficient. Because the SDP solution is used as the high-performance benchmark against which the scalable algorithm and the neglected-temperature baseline are compared (Fig. 8), the validity of P5 is load-bearing. If the uniformity assumption is violated, the SDP solution need not satisfy the true constraint C2, and the benchmark may be invalid. Please provide a formal justification under near-field area illumination, or empirically verify the original C2 for the SDP solutions over many random channel realizations","section":"Section IV-B1, Lemma 1 and cC2"},{"comment":"The performance of the low-complexity algorithm and of the robust design is presented without a convergence guarantee. The authors note that because of the projection in Eq. (45), a strictly monotonic increase is not guaranteed; Fig. 6 shows that most initializations increase the secrecy rate but one should report the fraction of initializations that fail and the variance of the final rate. For the temperature-robust design in Section V, the number M of sampled temperatures is not reported, and no stopping rule or complexity analysis is given for the joint-spatial-thermal LSE iteration. Since the robust design is one of the paper’s two main contributions, please specify M, the convergence criterion, and the robust algorithm’s sensitivity to M.","section":"Section VI-B, Figs. 6 and 11"}],"minor_comments":[{"comment":"The output statement writes \"s⋆(T)=s(T)\"; this should be s⋆(T)←s(T) or similar, to avoid confusion between the optimal output and the current iterate.","section":"Algorithm 2, line 20"},{"comment":"The legend labels are difficult to parse in the caption (\"Optimized Neglected Robust, Scenario 1\"). Please use distinct markers/colors and a clearer legend entry for the robust design.","section":"Fig. 11"},{"comment":"The number of discretization points M for the temperature set in the robust design is not stated; please report it for reproducibility.","section":"Section VI-A"},{"comment":"The piecewise wrapping function is described as a projection, but it is not a Euclidean projection onto the feasible set. Please clarify that it is a heuristic feasibility-restoring map, not an optimal projection.","section":"Eq. (45)"},{"comment":"The text says the direct link is neglected in algorithm design but included in numerical evaluations, yet Section VI does not clearly state how the blocked-direct-link model is reconciled with the path-loss parameters for the BS-MU link. Please clarify.","section":"Section II-C"}],"recommendation":"major_revision","confidential_remarks":"The paper is technically competent and the optimization framework is sensible, but the central quantitative claim is currently validated only under the same model used by the algorithms. I recommend major revision rather than rejection because the deficiency is a validation gap, not a proof error: the authors could add measured temperature-dependent phase data or clearly reframe the results as predictions under the Haller model with a sensitivity analysis. The code release and the SDP-versus-scalable comparison are valuable and should be preserved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a solid, honest engineering paper. The genuinely new pieces are the temperature-robust static design in Section V — one nominal phase profile that keeps secrecy rate nearly flat across a temperature range without live thermal sensing — and the linear-complexity scalable algorithm that makes the adaptive design practical for large RIS. The SDP benchmark is standard but fine, and they ship MATLAB code, which earns real reproducibility credit.\n\nThe paper does well by being clear about what it assumes. Eq. (16) is stated as a model, baseline min-phase calibration is flagged, and Lemma 1's uniform-phase assumption is conceded rather than hidden. The complexity analysis is honest, and the LSE-style smoothing over the joint location–temperature uncertainty set is a sensible way to handle the non-smooth worst-case objective.\n\nThe load-bearing caveat is exactly what the stress-test note says: the simulated \"actual\" temperature response is generated from Eq. (16) — the same Haller/Maier-Saupe scaling, the same beta, Tc, Tr — that the algorithms use. So Figs. 8–11 show self-consistency, not hardware accuracy. If a real LC-RIS phase profile does not compress multiplicatively with temperature, or if the ordinary-index baseline drifts, the reported secrecy gains are not demonstrated. That is a validation gap, not a proof error. The fix is either measured phase-vs-temperature data or a clearly framed \"performance under the Haller model\" claim plus sensitivity analysis to beta, Tc, and non-multiplicative mismatch.\n\nA smaller soft spot: the C2-to-cC2 reformulation for the SDP benchmark rests on a phase-uniformity assumption that is not guaranteed. Since the scalable method does not use that reformulation, it mainly weakens the \"upper bound\" status of the SDP comparison. Minor to moderate. The relationship to [14] (same group, temperature-resilient LC-RIS) also needs clarification; as written it is unclear how much the robust idea overlaps.\n\nWho should read it: people working on RIS-assisted physical layer security and anyone prototyping LC-RIS control. It deserves a serious referee — the direction is plausible and the code is available. My recommendation: send to peer review; in revision, require at least one out-of-sample check (synthetic model mismatch or existing measured data from [33]/[34] at another temperature) and explicit error bars. With that, it becomes a solid contribution.","headline":"Competent extension of the authors' ICC 2025 work — the temperature-robust static design is a genuinely useful idea — but the simulations validate the algorithms against the same temperature model they assume, so the secrecy-rate gains are not yet tested against hardware.","tokens_in":21289,"tokens_out":2340,"would_cite":true,"duration_ms":27611,"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":"The paper claims that LC-RIS phase drift under temperature follows one multiplicative power law, and that optimizing against it—over location zones, not full CSI—keeps secrecy rates temperature-blind designs lose.","keywords":["liquid crystal","reconfigurable intelligent surface","temperature dependence","physical layer security","secrecy rate","phase-shift optimization","mmWave","robust design"],"falsifier":"Measure the phase shift of a real LC-RIS unit cell across several bias voltages at multiple temperatures (for example −20, 0, 10, 25, and 40 °C) and check whether the ratio ω(V,T)/ω(V,Tr) equals ((Tc−T)/(Tc−Tr))^β independent of V. A single fitted exponent across voltages would support the design; a V-dependent ratio, or a ratio that changes shape near the clearing temperature, would falsify the multiplicative model that the secrecy-rate gains rest on.","tokens_in":20347,"feed_emoji":"🌡️","tokens_out":10411,"duration_ms":95187,"temperature":0.7,"pith_summary":"The paper sets out to show that the main failure mode of liquid-crystal reconfigurable intelligent surfaces in secure wireless links is thermal: as temperature rises, the achievable phase-shift range compresses below 2π, and any phase configuration computed at a reference temperature silently degrades. It proposes a physics-based model in which the entire voltage-to-phase response at any temperature is the reference-temperature response scaled by a single power-law factor, ((Tc−T)/(Tc−Tr))^β. On top of that model it builds two phase-shift designs—one that adapts to a known temperature and one that requires no temperature knowledge at all—optimized over the possible location zones of the legitimate user and eavesdropper rather than full channel state information. The central claim is that these designs keep the worst-case secrecy rate (the excess of the legitimate user's rate over the eavesdropper's) nearly flat across temperature, while designs that ignore temperature leak signal to eavesdroppers as the temperature deviates from the reference. A sympathetic reader would care because this gives LC-RIS hardware—otherwise energy-efficient and scalable—a tractable thermal behavior that can be engineered around in secure mmWave systems.","feed_headline":"Heat-aware LC-RIS tuning preserves secrecy; ignoring heat loses it","feed_subtitle":"A static phase setting built from the temperature law holds secrecy from -20 to 40 °C with no live temperature data.","key_machinery":"The central object is the multiplicative temperature-scaling law of Eq. (16), obtained from the standard empirical power-law approximation to the mean-field order parameter of nematic liquid crystals; it does the work of turning the physics of thermal disorder into a parameter-free engineering rule: take the reference-temperature phase response and scale it by ((Tc−T)/(Tc−Tr))^β. On the algorithmic side, the key mechanism is the low-complexity design's log-sum-exp (LSE) surrogate for the worst-case secrecy rate over the joint spatial/temperature uncertainty set, together with a piecewise wrapping function that projects unconstrained phase angles back into the temperature-limited range. This","core_discovery":"The load-bearing result is the paper's thermal model, Eq. (16): the phase shift of an LC-RIS element at bias voltage V and temperature T is ω(V,T) = ω(V,Tr)·((Tc−T)/(Tc−Tr))^β, where Tr is the reference temperature at which a full 2π range is available and Tc is the clearing temperature at which the liquid crystal becomes isotropic. The model says thermal change does not add a constant phase error; it multiplicatively compresses the whole phase-shift profile, so the maximum tunable range falls below 2π as soon as the temperature exceeds the reference. The paper then formulates secure communication as a worst-case optimization over the possible locations of the legitimate user and the eavesdr","pith_inferences":["An extension the paper leaves implicit: replacing the uniform worst-case treatment of the user/eavesdropper zones with a probability distribution would let the same LSE surrogate weight high-threat locations more heavily.","The multiplicative form of Eq. (16) suggests temperature compensation could be applied as a post-hoc remapping of any existing LC-RIS phase configuration, rather than a full re-optimization; the paper does not test this transfer.","If the model is confirmed by hardware measurements, the temperature-robust configuration implies an architectural shortcut: configure the RIS once for the local climate range and let base-station beamforming handle fast adaptation, eliminating per-element thermal sensing."],"forward_implications":["Designing LC-RIS phase shifts with the Eq. (16) model keeps the worst-case secrecy rate roughly constant as the operating temperature moves from −20 °C to 40 °C, while temperature-blind designs lose secrecy at both extremes (Figs. 9–11).","The temperature-robust static configuration, computed without any live temperature reading, still achieves a nearly flat secrecy rate across the whole tested range—so thermal sensing hardware and feedback overhead are not mandatory.","The scalable low-complexity algorithm runs in time linear in the number of RIS elements (seconds for N=400) versus cubic for the SDP benchmark (hours), making real-time reconfiguration of very large surfaces feasible.","Coverage by zone rather than by point means the same phase configuration serves a moving legitimate user and resists eavesdropper location uncertainty, which directly reduces CSI acquisition overhead in mmWave systems.","The requirement π<ωmax<2π for the convex reformulation of the phase-range constraint is satisfied by the experimental temperature range considered, so the math aligns with the physics in the operating regime."],"fun_headline_variants":["LC-RIS phase shifts compress with heat; new design stays secure","No live temp needed: LC-RIS design holds secrecy from -20 to 40°C","Thermal law for LC-RIS: phase range shrinks, secrecy protected","Temperature-adaptive LC-RIS: robust secrecy without real-time heat data","Heat-aware LC-RIS tuning: security preserved without live temperature"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire numerical demonstration assumes that real LC-RIS hardware follows the empirical power-law temperature model of Eq. (16) with the same exponent and clearing temperature used by both the algorithm and the 'ground-truth' simulator; no measured phase-versus-temperature data from a real cell are presented, so if the thermal response differs—especially if the phase profile does not scale multiplicatively across all bias voltages—the reported secrecy-rate gains are not de","fun_headline_variants_meta":{"raw":{"variants":["LC-RIS phase shifts compress with heat; new design stays secure","No live temp needed: LC-RIS design holds secrecy from -20 to 40°C","Thermal law for LC-RIS: phase range shrinks, secrecy protected","Temperature-adaptive LC-RIS: robust secrecy without real-time heat data","Heat-aware LC-RIS tuning: security preserved without live temperature"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000424,"raw_usage":{"total_tokens":2060,"prompt_tokens":838,"completion_tokens":1222,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":582,"completion_tokens_details":{"reasoning_tokens":1123}},"tokens_in":582,"tokens_out":1222,"duration_ms":9319,"temperature":1.0,"reasoning_tokens":1123,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T05:39:46.983010+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the phase shift of a real LC-RIS unit cell across several bias voltages at multiple temperatures (for example −20, 0, 10, 25, and 40 °C) and check whether the ratio ω(V,T)/ω(V,Tr) equals ((Tc−T)/(Tc−Tr))^β independent of V. A single fitted exponent across voltages would support the design; a V-dependent ratio, or a ratio that changes shape near the clearing temperature, would falsify the multiplicative model that the secrecy-rate gains rest on.","supporting_citations":[],"review_version":1}