{"id":"2e02f7c9-a8b8-4459-9204-3e540515c684","arxiv_id":"2607.17475","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"New beamforming algorithms maximize energy efficiency of a TRIS transceiver ISAC system under perfect and worst-case imperfect CSI, using FP+MM and S-Procedure-based convex surrogates.","lead":"Engineers design transmit beamforming for a transmissive reconfigurable intelligent surface that both serves users and senses a target, maximizing energy efficiency under perfect and imperfect channel knowledge. Two iterative algorithms are proposed and simulated to show energy-efficiency gains over a conventional multi-antenna base station.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (1) treats the TRIS as a fully digital multi-stream array; a single-feed TMA TRIS has a rank-one common-feed signal space, so the optimized EE is an unrealized upper bound unless hardware feasibility is shown.","rationale":"I read the paper in good faith. The FP/MM/S-Procedure machinery is applied in a standard way; Theorem 1's monotonicity argument, while terse, is coherent, and no circularity or data fitting is evident. The weakest link is the transition from the described single-feed TRIS architecture to the fully flexible signal model in Eq. (1). This is the same assumption the reader flagged, and I agree it is load-bearing: the EE gain over a conventional N-RF-chain BS is precisely the paper's headline result, and it depends on the TRIS achieving arbitrary beamforming vectors with one RF chain. The paper's footnote citing [31] does not supply a hardware derivation, and the omitted Algorithm 2 convergence proof compounds the uncertainty but is secondary. I therefore recommend no change to the reader's conditional verdict: the condition should be a concrete demonstration or correction of the hardware model.","tokens_in":23491,"tokens_out":16238,"duration_ms":162730,"concrete_test":"Analytical rank test: compute the rank of the instantaneous transmit covariance E{x x^H} for Eq. (1); with independent x_c,k and x_r,n it is min(N,K+N). For the single-feed TMA TRIS model x=s_feed(t)*t(t), the covariance has rank one. If the ranks differ for any K,N >= 2, the signal model in Eq. (1) cannot describe the hardware and the optimized weights cannot be realized; re-running Fig. 4 under the rank-one common-feed constraint is the quantitative follow-up.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Central claim requires that Algorithm 1/2 output beamforming vectors that a TRIS transceiver can physically synthesize. In Section II-A, Eqs. (1)-(4), w_c,k and w_r,n are arbitrary complex vectors with only per-antenna power constraints. But the TRIS architecture in the Introduction has one horn-feed antenna and passive time-modulated unit cells. With a single feed, the transmitted vector is x(t)=s_feed(t)*t(t), where t(t) is the vector of unit-cell transmission coefficients; all spatial components are driven by the same scalar feed waveform. Eq. (1) instead transmits K independent user symbols and N independent radar waveforms, each with its own beamforming vector, i.e., K+N spatial streams. Realizing it requires at least K+N RF chains (or equivalent baseband controls), not the single RF chain used in the power model (13). TMA harmonic constraints further couple each element's amplitude and phase (fundamental coefficient c_n=(tau_n/T)sinc(pi*tau_n/T)e^{-j pi tau_n/T}) and bound its magnitude, so the feasible set in (P0)/(P1) is a strict superset of physically achievable TRIS weights. Consequently, the EE values in Figs. 4 and 8 and the claimed advantage over Tra. BS are not shown to be attainable.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers a downlink ISAC system in which a single-feed, time-modulated-array (TMA) transmissive RIS (TRIS) transceiver serves K single-antenna users and senses a point target. It formulates two energy-efficiency (EE) maximization problems, (P0) under perfect CSI and (P1) under bounded-norm imperfect CSI, with per-user rate constraints, a beampattern-gain constraint on the sensing target, and per-antenna power constraints. For the perfect-CSI case, the authors use fractional programming (FP) and majorization-minimization (MM) to obtain convex subproblems and state a monotonic-convergence theorem (Theorem 1). For the imperfect-CSI case, they use the S-Procedure and a sign-definiteness lemma to convert semi-infinite constraints into LMIs and propose an MM-based algorithm (Algorithm 2), whose convergence proof is explicitly omitted. Numerical experiments report convergence and claim EE gains over a conventional N-RF-chain base station.","tokens_in":23797,"tokens_out":12054,"duration_ms":121009,"significance":"If the system model were physically realizable, the paper would be a useful contribution to ISAC optimization: the perfect-CSI algorithm has a written convergence proof, all subproblems are convex, complexity estimates are provided, and the robust reformulation uses standard tools. However, the central premise — that a single-feed TMA TRIS can transmit K+N independent spatial streams with arbitrary beamforming vectors while consuming only one RF chain — is not justified and is contradicted by the architecture described in the Introduction. The reported EE values in Figs. 4 and 8 and the claimed advantage over the conventional base station are therefore not shown to be attainable on the modeled hardware. The algorithmic machinery may be of independent interest for an abstract per-antenna-constrained multi-antenna transceiver, but it does not establish the paper's TRIS-specific claims.","major_comments":[{"comment":"The transmitted signal in Eq. (1) is modeled as K independent communication streams and N independent radar streams, each with an arbitrary beamforming vector w_c,k or w_r,n, subject only to the per-antenna power constraints in (2). This is a full-DoF, multi-stream array model. The TRIS architecture described in the Introduction, however, has a single horn-feed antenna illuminating passive TMA unit cells. With a single feed, the spatial transmit vector is a common feed waveform multiplied element-wise by the unit-cell transmission coefficients; at each TMA harmonic the spatial vector is rank-one and the coefficients are coupled in amplitude and phase by the duty cycle. Realizing the K+N streams of Eq. (1) would require K+N RF chains or equivalent baseband controls, not the single RF chain used in the power model (13). Thus the feasible sets of (P0) and (P1) are strict supersets of physic","section":"Section II-A, Eqs. (1)-(4)"},{"comment":"The convergence analysis for Algorithm 2 is omitted with the statement that it 'follows that for (P0).' This is not a cosmetic gap. The imperfect-CSI algorithm introduces additional slack variables μ, ν, the S-Procedure multipliers, and the MM linearization of log(ν_k). The feasibility and monotonicity of this more complex iteration — especially the possible interaction between the upper-bounding tangent for log(ν_k), the conservative robust constraints, and the fractional objective — are not immediate consequences of Theorem 1. Since the paper claims the same convergence behavior for Algorithm 2, a proof or a precise statement of what is guaranteed is required.","section":"Section IV-B1"},{"comment":"The numerical section does not specify several parameters on which the central EE comparison depends: P_c,TRIS (and its components P_s,TRIS and P_RF), ξ_TRIS, the rate threshold R_th, the beampattern threshold P_r, the uncertainty radii ξ_c,k and ξ_r, and the power-consumption model of the 'Tra. BS' benchmark. No carrier frequency, bandwidth, or number of random channel realizations is reported, and the figures contain no error bars or confidence intervals. Because the claimed EE advantage is driven largely by the difference between one RF chain and N RF chains, omitting these values makes the quantitative results non-reproducible and weakens the validation of the central claim.","section":"Section V"}],"minor_comments":[{"comment":"The theorem statement says the objective value of problem (P5) increases monotonically, but the proof shows monotonic increase of the original EE objective through the surrogate bounds. The statement should be reworded to match the proof.","section":"Theorem 1"},{"comment":"After the MM lower bound, the paper notes that (60) is a conservative sufficient condition for (35c), but the subsequent S-Procedure step is written with '⇒'. Clarify that the LMI condition is sufficient (and equivalent to the lower-bound condition) rather than equivalent to the original constraint.","section":"Appendix B, Eq. (61)"},{"comment":"The x-axis is labeled 'Maximum distance' and the text varies it from 50 to 200 m, but the system setup in Section V states user distances are 'randomly generated within a sector region with distances ranging from 20 to 50 m.' Clarify how the maximum-distance experiment is generated.","section":"Fig. 5"},{"comment":"The initialization includes ν(0) but not μ(0); for reproducibility specify initializations for all blocks, including the multipliers.","section":"Algorithm 2"},{"comment":"The heavy reliance on the authors' own prior TRIS papers ([12], [14]–[18], [22]–[23]) is acceptable, but the statement in footnote 1 should cite a source that actually establishes the full-DoF signal model for a TRIS transceiver, or justify it from first principles.","section":"References"}],"recommendation":"reject","confidential_remarks":"The hardware-feasibility issue in Section II-A is fundamental. The paper optimizes over a fully digital multi-stream array while claiming the hardware cost of a single-feed passive TRIS. This is not a local fix: incorporating TMA harmonic constraints, rank-one feed structure, and the associated power model changes the optimization problem and likely the conclusions. Even if the omitted convergence proof for Algorithm 2 and the missing simulation parameters were supplied, the TRIS-specific contribution would remain unsupported. The authors might re-frame the work as an EE-optimal ISAC design for a generic multi-antenna transceiver with per-antenna power constraints, but that would be a substantial rewrite rather than a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The optimization machinery is solid, but the system model is not the TRIS transceiver the paper describes. Eq. (1) transmits K user symbols and N radar waveforms, each with its own arbitrary N-dimensional beamforming vector, constrained only by per-antenna power. That is a fully digital array with K+N independent spatial streams. The TRIS described in the introduction has a single horn feed and time-modulated unit cells, so all transmitted signals share the same scalar feed waveform; the spatial response is a single vector (or at most a set of coupled harmonic responses), not K+N free beamformers. The power model in (13) then charges only one RF chain, which makes the EE comparison against the N-RF-chain 'Tra. BS' an unfair fight. The EE gains in Figs. 4 and 8 are upper bounds for the actual hardware, and the central claim is unsubstantiated. The stress-test note is correct.\n\nWhat the paper does well: the FP+MM reformulations are executed carefully, Theorem 1's monotone convergence argument is written out and plausible, and the S-Procedure / sign-definiteness transformations for the robust problem are standard but handled in enough detail to be checkable. The complexity analysis is thorough, and the simulation curves are internally consistent. As a solution to a generic per-antenna power-constrained multiuser beamforming problem with a fractional objective, the paper is competent.\n\nThe soft spots beyond the model flaw: Algorithm 2's convergence proof is explicitly omitted, and the numerics lack error bars and a full parameter list. These are minor relative to the load-bearing issue.\n\nWho is this for? A reader working on TRIS hardware modeling might use it as a cautionary example of the gap between idealized array models and single-feed/TMA constraints. But the paper's own contribution is not what it claims to be.\n\nRecommendation: desk-reject in current form. The fundamental mismatch between Eq. (1) and the described hardware would be the first thing a referee finds; there is no need to spend referee time on it. The authors could rework the model to include the single-feed/TMA degrees-of-freedom constraints, and the algorithmic machinery might then be reusable, but that would be a substantially different paper.","headline":"Competent optimization, but Eq. (1) models a fully digital array and charges one RF chain, so the claimed TRIS EE gains are not realizable.","tokens_in":24282,"tokens_out":4481,"would_cite":false,"duration_ms":48504,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper designs beamforming for a transmissive-RIS transceiver so that one integrated sensing and communication system can maximize energy efficiency while guaranteeing user rates and target sensing, under both perfect and imperfect chan","keywords":["transmissive RIS transceiver","integrated sensing and communication","energy efficiency optimization","transmit beamforming","robust beamforming","imperfect CSI","majorization-minimization","S-Procedure"],"falsifier":"Build or simulate a time-modulated TRIS transceiver and try to synthesize an optimized beamforming vector from Algorithm 1 using only the available harmonic weight controls; measure the achieved per-user SINR, target beampattern gain, and total consumed power and compare to the predicted values. A significant shortfall would falsify the claim that the optimized beamformers are realizable under the stated per-antenna power model.","tokens_in":23396,"feed_emoji":"📡","tokens_out":2167,"duration_ms":25671,"temperature":0.7,"pith_summary":"The paper studies an integrated sensing and communication (ISAC) downlink system built around a transmissive reconfigurable intelligent surface (TRIS) transceiver, which uses a single feed antenna and one RF chain to shape transmitted signals. It tries to prove that the transmit beamformers can be optimized to maximize energy efficiency while enforcing a minimum rate per user, a minimum beampattern gain toward a sensing target, and a per-antenna power limit. For perfect channel state information, the authors propose an iterative algorithm based on fractional programming and majorization-minimization that they claim converges monotonically and always returns a feasible solution. For imperfect channel state information, they use the S-Procedure to turn semi-infinite robust constraints into linear matrix inequalities and develop a second MM-based algorithm. If the claims hold, a TRIS transceiver can deliver higher energy efficiency than a conventional fully active multi-antenna base station because it needs only one RF chain.","feed_headline":"One-RF-chain TRIS transceiver boosts ISAC energy efficiency","feed_subtitle":"Algorithms optimize beamforming under perfect and imperfect CSI, beating conventional base stations in EE.","key_machinery":"The load-bearing mechanism is a reformulation chain: fractional programming transforms the logarithmic rate into a quadratic surrogate with auxiliary variables, then majorization-minimization linearizes the non-convex fractional objective and non-convex rate and beampattern constraints. For robustness, the S-Procedure converts bounded-norm channel uncertainties into finite-dimensional LMIs. The key modeling object is the TRIS transceiver, where beamforming vectors are arbitrary complex weight vectors subject only to per-antenna power constraints, and total power consumption is modeled as the amplified transmit power plus a static circuit power and one RF-chain power.","core_discovery":"The central claim is that the energy-efficiency maximization problem for a TRIS transceiver-enabled ISAC system can be reformulated and solved by iterative algorithms in both perfect and imperfect CSI settings. In the perfect CSI case, the rate functions are recast through Lagrangian dual and quadratic transforms, then the non-convex fractional objective and constraints are convexified using MM-based surrogate functions; Algorithm 1 is claimed to guarantee a feasible solution and monotonic increase of the objective. In the imperfect CSI case, the worst-case semi-infinite constraints are converted into LMIs via the S-Procedure, slack variables and further MM surrogates are introduced, and Alg","pith_inferences":["The paper's modeling of TRIS beamformers as arbitrary complex vectors likely idealizes real time-modulated arrays; actual harmonic constraints, unit-cell phase/amplitude limits, and mutual coupling may restrict the achievable weights, so the reported EE gains should be tested against a hardware-accurate TRIS model.","The EE advantage over a conventional base station depends on the power model that assigns circuit power to N RF chains for the conventional array; a power model that accounts for TRIS controller and switching overhead could narrow the gap.","A natural testable extension is to incorporate TMA harmonic spectral constraints into the optimization, converting the beamforming variables into limited sets of Fourier weights and measuring the resulting EE loss.","The robust design could be extended to angular uncertainty in target direction directly, rather than bounding the channel vector error, which may yield tighter robust sensing beampattern guarantees."],"forward_implications":["If the algorithms converge as claimed, TRIS transceivers can be used as energy-efficient ISAC transmitters while enforcing per-user rate guarantees and sensing gain constraints.","The single-RF-chain architecture would make TRIS transceivers attractive for power-constrained 6G deployments, yielding higher EE than conventional fully active arrays under the same per-antenna power budget.","The robust formulation provides worst-case EE guarantees when channel estimation errors are bounded, making the design applicable to practical imperfect-CSI scenarios.","Larger TRIS sizes (more unit cells) improve EE due to extra beamforming degrees of freedom, suggesting a favorable scaling path for the architecture.","The proposed framework extends to other TRIS-enabled applications needing simultaneous communication and sensing with power budgets."],"fun_headline_variants":["TRIS transceiver cuts ISAC energy cost with smart beamforming","Energy-efficient ISAC via TRIS transceiver under imperfect CSI","MM-based algorithms maximize ISAC energy efficiency with TRIS","Robust TRIS beamforming for energy-efficient ISAC with uncertain CSI"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The central assumption is that the TRIS transceiver can realize any complex optimized beamforming vector subject only to per-antenna power limits; if the time-modulated array hardware imposes additional constraints (phase quantization, harmonic control, coupling), the computed EE and the comparison to a conventional base station may not hold.","fun_headline_variants_meta":{"raw":{"variants":["TRIS transceiver cuts ISAC energy cost with smart beamforming","Energy-efficient ISAC via TRIS transceiver under imperfect CSI","MM-based algorithms maximize ISAC energy efficiency with TRIS","Robust TRIS beamforming for energy-efficient ISAC with uncertain CSI"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000753,"raw_usage":{"total_tokens":3189,"prompt_tokens":749,"completion_tokens":2440,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":493,"completion_tokens_details":{"reasoning_tokens":2376}},"tokens_in":493,"tokens_out":2440,"duration_ms":17439,"temperature":1.0,"reasoning_tokens":2376,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T17:50:36.324926+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build or simulate a time-modulated TRIS transceiver and try to synthesize an optimized beamforming vector from Algorithm 1 using only the available harmonic weight controls; measure the achieved per-user SINR, target beampattern gain, and total consumed power and compare to the predicted values. A significant shortfall would falsify the claim that the optimized beamformers are realizable under the stated per-antenna power model.","supporting_citations":[],"review_version":1}