{"id":"376fd489-f2e8-40c9-8096-ac884cc9c9aa","arxiv_id":"2501.12752","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A 100x100 two-bit RIS at 304 GHz can be modeled as three discrete rays, and its far-field approximation holds at roughly one fifth of the textbook far-field distance.","lead":"This paper uses full-wave simulations of a 100x100 reconfigurable surface at 304 GHz to show that its reflected field can be captured by three dominant rays, and that a far-field approximation works at distances much closer than theory suggests. The result could make ray-tracing tools fast enough to model THz indoor links that use extremely large surfaces.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The three-ray model is extracted under normal plane-wave illumination only; its transfer to arbitrary indoor TX positions is untested, so the ray-tracing shortcut may fail for oblique incidence.","rationale":"I read the full text in good faith. The paper is internally consistent: the CST full-wave RCS patterns of Section III-B clearly show a dominant lobe at 30 degrees, a secondary lobe at -30 degrees, and a specular component at 0 degrees, and the three-ray compression is a reasonable reading of those data. The far-field amplitude check in Fig. 5 is also a legitimate piece of evidence. My concern is not that the fit is wrong for the particular simulated configuration; it is that the paper moves from a normal-incidence plane-wave characterization to an indoor channel model without testing the parameter that most plausibly breaks the three-ray assumption: incidence angle. In an indoor room the TX will generally not be at broadside, and the RIS phase profile is explicitly designed for normal plane waves. Because RIS scattering is governed by a generalized Snell's law, the number and direction of grating lobes change with incidence angle, so the three-ray decomposition is not expected to be invariant. The reader identified exactly this issue in the weakest_assumption field, and I agree it is the load-bearing gap. I would keep the verdict CONDITIONAL: the paper's own results support the three-ray approximation for the simulated normal-incidence case, but the generalization to indoor NLOS scenarios requires at least one oblique-incidence simulation or measurement. I do not see grounds to reject the paper; the stated scope could be narrowed, or the missing simulation added. No ad hominem intended; the critique targets the argument's scope.","tokens_in":6031,"tokens_out":4598,"duration_ms":50759,"concrete_test":"Run a CST full-wave RCS simulation of the same 100x100 2-bit RIS with a plane wave incident at an oblique angle, e.g. 20 degrees (and optionally 45 degrees), at 304 GHz. From the resulting pattern, extract the directions and relative amplitudes of the dominant lobes using the same 2 degree cone criterion as Section IV-B. If the main beam direction obeys the predicted anomalous-reflection law but the spurious -30 degree lobe and 0 degree specular lobe change in relative level by more than about 3 dB, or if additional significant lobes appear, the three-ray model is illumination-specific and cannot be transferred to general indoor TX positions. Also repeat with a point source at a representative distance (e.g., 5 m) to test wavefront curvature.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that a full-wave RCS simulation of the 100x100 2-bit RIS, under normal plane-wave illumination, yields a three-ray decomposition (main beam at +30 degrees, symmetric spurious lobe at -30 degrees, specular leftover at 0 degrees) that can be plugged into ray tracing for an indoor NLOS scenario. For this transfer to be valid, the three-ray structure and its amplitudes must be stable as the TX moves: real indoor links illuminate the RIS from oblique angles and with spherical wavefronts, not as a normally incident plane wave. The phase profile in Section III is optimized for normal incidence; the generalized anomalous-reflection law implies that for oblique incidence the main beam direction, the quantization-induced grating lobe, and the specular component all shift in angle and relative power. Since no oblique-incidence or finite-distance source simulation is reported, the observed three-ray fit may be an artifact of the single illumination condition tested. This does not make the paper internally inconsistent; it makes the headline claim about 'indoor channel characterization' broader than the evidence. A second, related gap is that the far-field approximation in Fig. 5 is validated only for amplitude at the 30 degree observation angle, not for phase, which is needed for coherent multipath interference; however, the incidence-angle question is the more fundamental limitation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a 2-bit static RIS design at 304 GHz with a 100×100 element, 5×5 cm aperture intended to redirect normally incident plane waves to θout = 30°. Full-wave RCS simulations of this RIS show three dominant scattering directions: the main beam at +30°, a spurious symmetric beam at −30°, and a specular leftover at 0°. Based on these observations, the authors propose approximating the RIS behavior in ray tracing by a three-ray model and claim that the far-field approximation is valid at RIS-RX distances as small as 2 m. They also derive a single-ray received-power estimate via the bistatic radar equation and discuss the frequency dependence of the RIS beam direction.","tokens_in":6279,"tokens_out":4197,"duration_ms":39802,"significance":"If fully validated, the proposed three-ray approximation would be a useful, computationally efficient means of incorporating extremely large RISs into ray tracing for THz indoor channels. The paper's strengths include a full-wave simulation of an electrically large 100×100 RIS, a 2-bit unit-cell design with a 20 µm fabrication-tolerance check, RCS patterns showing narrow beams, use of the standard bistatic radar equation, and an amplitude-based far-field comparison at the main-beam angle. However, the central transfer from a single normal-incidence full-wave RCS characterization to arbitrary indoor TX/RX positions remains unverified, and the coherent (phase) behavior needed for ray interference is not validated. The significance of the result is therefore conditional on additional checks.","major_comments":[{"comment":"The three-ray model is extracted only for normal plane-wave illumination of the RIS. The indoor scenario in Fig. 1, however, involves a TX whose position implies generally oblique incidence on the RIS, and the phase profile in Section III is explicitly designed for normal incidence. For oblique incidence the main beam direction, the quantization-induced grating lobe, and the specular component all change in angle and relative power. Since no oblique-incidence or finite-distance source simulation is reported, the proposed ray-tracing shortcut is not yet supported. A concrete test would be to simulate at least two oblique incidence angles (e.g., θin = 20° and 40°) and show that three rays with the same structure and stable amplitudes/directions emerge, or to provide an incidence-angle-dependent parameterization of the rays.","section":"§IV-B and Figs. 4, 8"},{"comment":"The three-ray model parameters are read from the same full-wave RCS simulation that is used to demonstrate the approximation, so the 'agreement' is a measure of self-consistency rather than an external prediction. No validation is shown in an actual channel: there is no comparison of the three-ray model against full-wave fields for a TX-RX link with the geometry of Fig. 1, and no ray-tracing implementation is run. Please provide a quantitative error metric (e.g., integrated angular field error or a sample NLOS channel impulse response computed both with the full-wave RIS data and with the three-ray model).","section":"§IV-B and Fig. 8"},{"comment":"The far-field approximation is validated only for the amplitude of the E-field at the fixed observation angle of 30°. Ray tracing for interference requires the complex field, i.e., both amplitude and phase, because the relative phases of the three rays determine their coherent summation. Since Fig. 8 explicitly reports the relative phase between the rays, the claim that the far-field approximation is valid at short distances should be checked for the phase as well, for example by comparing complex E-field versus distance at several angles.","section":"§III-C and Fig. 5"},{"comment":"The frequency-shift formula Δθ(f) = θout − arcsin(f0/f) is dimensionally inconsistent: at f = f0 it gives a nonzero angle shift (arcsin(1) = 90°), and for f < f0 the argument of arcsin exceeds unity. For a fixed phase gradient, the correct relation is sin θ(f) = (f0/f) sin θ0 (or, for small angles, Δθ ≈ (1 − f0/f) θ0). Please correct Eq. (2) and re-derive any frequency-dependence claims that rely on it.","section":"§IV-A, Eq. (2)"}],"minor_comments":[{"comment":"The phrase 'An 100×100' should be 'A 100×100'; the same grammatical issue appears at the beginning of Section II.","section":"Abstract and §I"},{"comment":"The acronym 'FTTD' in 'the Finite Difference Time Domain (FTTD) solver' should be 'FDTD'.","section":"§III-B"},{"comment":"The sentence 'It can observed that' is missing 'be'; it should read 'It can be observed that'.","section":"§III-B"},{"comment":"The phrase 'with and amplitude variation below 0.4 dB' should be 'with an amplitude variation below 0.4 dB'.","section":"§III-C"},{"comment":"The caption contains a duplicated definite article: 'to facilitate the the TX-RX link' should read 'to facilitate the TX-RX link'.","section":"Fig. 1 caption"},{"comment":"The text states that 'the RCS can vary by 3 dB' within a 10 GHz bandwidth, whereas Fig. 6 shows E-field values; please clarify whether the 3 dB variation refers to the RCS or to the received E-field, since the two differ by a square-root relation.","section":"§IV-A and Fig. 6"}],"recommendation":"major_revision","confidential_remarks":"This is a plausible engineering contribution, but the headline claim is broader than the validation. The core gaps are the lack of oblique-incidence simulations and the absence of an actual ray-tracing or channel-level comparison. If the authors can add those checks and correct Eq. (2), the paper would be suitable for publication. I would also encourage them to relate their results to their own experimental work in reference [20], which could provide the missing external validation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is worth your time, but with the same caveat the reader flagged: the three-ray model is derived and verified only for normal plane-wave illumination, and the title and abstract claim a lot more than that. The specific contributions are concrete and credible for what they simulate: a 100x100 two-bit RIS at 304 GHz with a designed +30° anomalous reflection, full-wave CST RCS patterns showing a main beam at +30°, a specular leftover at 0°, and a spurious symmetric lobe at -30°, and a claim that the far-field approximation holds within 0.4 dB beyond 2 m of the RIS. The phase analysis in Fig. 8 is a nice touch, because coherent multipath models need phase, not just amplitude.\n\nWhat is genuinely new here is the complete chain from unit-cell design to a reduced-order three-ray description that a ray tracer could actually use. Most prior work either idealizes the RIS or stops at the RCS pattern. The paper also checks robustness to ±20 µm fabrication variations, which is practical. The bistatic radar equation application is standard, and the frequency-dispersion formula (Eq. 2) is correctly identified as something ray tracers often ignore.\n\nNow the soft spots, in proportion. First, the model is self-referential: the three rays are read off the same full-wave simulation they are supposed to approximate, so the agreement is not an independent prediction. That is acceptable as a modeling step, but it should be stated more honestly. Second, and more importantly, only normal incidence is simulated. In the indoor scenario of Fig. 1, the TX can be anywhere, so waves hit the RIS at oblique angles and with spherical wavefronts. The generalized anomalous reflection law means the main beam, grating lobe, and specular component all shift with incidence angle; the paper does not show that the three-ray structure survives that change. Without oblique-incidence or finite-distance TX simulations, the \"indoor channel characterization\" claim overshoots the evidence. Third, the far-field validation in Fig. 5 is amplitude-only at one observation angle; phase errors could matter for coherent interference, though the phase-flatness finding partially mitigates this. No measurements or code/data are included, so the numbers are not independently checkable.\n\nThe citation pattern is fine; the authors cite their own prior work on spurious rays as background, which is appropriate. My overall take: the central result is credible for the condition it actually tests, and the paper would be a solid contribution if the claims were scaled to normal incidence or augmented with at least one oblique-incidence simulation. As is, it is a useful design-and-characterization study that a serious ray tracing or THz RIS researcher should read. I'd send it to peer review, with a recommendation to add oblique-incidence validation or temper the title/abstract.","headline":"Credible three-ray characterization of a specific large THz RIS under normal incidence, but the indoor-generalization claims outrun the simulations.","tokens_in":6842,"tokens_out":2568,"would_cite":true,"duration_ms":25217,"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":"A 100x100, 2-bit RIS at 304 GHz can be represented in ray tracing as three equivalent rays rather than a full near-field aperture.","keywords":["reconfigurable intelligent surface","THz communications","indoor channel characterization","ray tracing","radar cross section","near-field propagation","anomalous reflection","2-bit unit cell"],"falsifier":"Simulate or measure the same 100x100 RIS under plane-wave illumination at oblique incidence angles such as $10^\\circ$, $20^\\circ$, or $30^\\circ$. If a spurious lobe at $-\\theta_{\\mathrm{out}}$ disappears, a new strong lobe appears, or the relative amplitudes of the three dominant directions change by more than a few dB, then the three-ray model as stated fails for non-normal TX positions.","tokens_in":5818,"feed_emoji":"📡","tokens_out":4750,"duration_ms":45788,"temperature":0.7,"pith_summary":"The paper seeks a cheap way to put extremely large reconfigurable intelligent surfaces into indoor channel simulations at sub-THz frequencies, where full-wave modeling of the surface is too heavy. It designs a 100x100, 2-bit RIS on a 5x5 cm aperture operating at 304 GHz, aimed at tilting a normally incident plane wave to +30 degrees. Full-wave radar cross section simulations show that the surface sends energy mainly into three directions: the intended +30 degree beam, a spurious symmetric beam at -30 degrees, and a leftover specular reflection at 0 degrees. The paper argues that these three rays can be inserted into existing ray tracing tools, and that the far-field approximation, despite a theoretical far-field distance of 10 m, is accurate to within 0.4 dB for RIS-to-receiver distances beyond 2 m. If true, indoor RIS-assisted THz channel characterization becomes far lighter while still capturing the main multipath behavior.","feed_headline":"A 100x100 THz metasurface shrinks to three rays","feed_subtitle":"Full-wave simulations at 304 GHz show energy lands in just three directions, letting standard ray tracing model the surface.","key_machinery":"The machinery is the three-ray decomposition of the full-wave RCS pattern of the entire metasurface. Full-wave RCS simulations replace the coupled near-field problem by an equivalent set of plane-wave rays: one for the designed anomalous beam, one for the spurious symmetric beam, and one for the specular leftover; each ray carries the amplitude and phase read from the simulated pattern. The 2-bit unit cell with four phase states, here $0^\\circ$, $96^\\circ$, $184^\\circ$, and $273^\\circ$, provides the discrete phase profile that creates the anomalous reflection, and the bistatic radar equation supplies the path-loss scaling for the dominant ray.","core_discovery":"The paper claims that an extremely large RIS does not need to be modeled as a full near-field aperture in ray tracing; its scattering can be compressed to three equivalent rays. For the designed 100x100 element, 2-bit RIS at 304 GHz, the full-wave RCS pattern has a main lobe at $\\theta_{\\mathrm{out}}=30^\\circ$ with RCS 15.6 dBm$^2$, a specular lobe at $0^\\circ$ with 6.1 dBm$^2$, and a spurious symmetric lobe at $-30^\\circ$ with 3.9 dBm$^2$, with a main-beam half-power width of only $1^\\circ$. The phase of the main lobe is nearly constant over the central 12 dB angular range, so a single ray with the bistatic radar equation captures that lobe, and the two smaller lobes are added as interfering rays. In addition, the paper shows that the far-field approximation is adequate after 2 m even though the textbook far-field distance of the 5 cm aperture at 304 GHz is about 10 m.","pith_inferences":["If the three-ray compression holds for oblique incidence, something the paper does not simulate, the same shortcut would apply to many TX positions in a room, since real links rarely hit the RIS at exactly normal incidence.","The 65% efficiency relative to ideal specular reflection suggests the energy in the spurious and specular lobes is not negligible; a designer could deliberately reuse those rays for coverage diversity rather than treating them only as interference.","A natural extension is to recompute the three-ray parameters for other 2-bit phase distributions: the model structure may persist, but the lobe angles and relative amplitudes must be re-extracted for each configuration."],"forward_implications":["Ray tracing tools can model this 304 GHz RIS by launching three rays instead of embedding the full surface, which removes the need for near-field integration over 10,000 elements.","For indoor rooms with TX and RX distances above roughly 2 m, the far-field formula can be used for RIS-to-RX links, so coverage calculations stay simple.","A spurious symmetric beam and a specular leftover will appear in the channel as extra interference paths, so wideband indoor channel models should include these rays rather than assuming a single ideal reflection.","Because the beam direction shifts with frequency according to $\\Delta\\theta(f) = \\theta_{\\mathrm{out}} - \\arcsin(f_0/f)$, simulations across a 10 GHz bandwidth need to update the ray directions; the RCS varies by about 3 dB over that band."],"supporting_citations":[{"why":"Supplies the 2-bit unit-cell design rationale and the earlier observation that spurious rays are inevitable in RIS scattering.","marker":"[17]"},{"why":"Provides the hybrid ray-optical channel modeling approach that the paper wants RISs to plug into.","marker":"[24]"},{"why":"Documents experimental validation of ray tracing at THz frequencies, motivating the three-ray simplification.","marker":"[25]"},{"why":"Gives the frequency-dependent beam-pointing formula that the paper uses to express ray shifts across bandwidth.","marker":"[26]"}],"fun_headline_variants":["Three rays replace a 100x100 THz smart surface","Huge RIS at 300 GHz modeled as just 3 rays","Far-field at 2m for 100x100 RIS? Yes at 304 GHz","Ray tracing tamed: 100x100 RIS collapses to 3 rays","Full-wave RIS shrinks to 3 rays for fast indoor models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything rests on the RCS pattern measured under a normally incident plane wave staying representative when waves arrive at the RIS from the oblique angles that occur in real indoor links; the paper only simulates normal incidence.","fun_headline_variants_meta":{"raw":{"variants":["Three rays replace a 100x100 THz smart surface","Huge RIS at 300 GHz modeled as just 3 rays","Far-field at 2m for 100x100 RIS? Yes at 304 GHz","Ray tracing tamed: 100x100 RIS collapses to 3 rays","Full-wave RIS shrinks to 3 rays for fast indoor models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000622,"raw_usage":{"total_tokens":2897,"prompt_tokens":978,"completion_tokens":1919,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":594,"completion_tokens_details":{"reasoning_tokens":1836}},"tokens_in":594,"tokens_out":1919,"duration_ms":12803,"temperature":1.0,"reasoning_tokens":1836,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T16:50:02.193436+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or measure the same 100x100 RIS under plane-wave illumination at oblique incidence angles such as $10^\\circ$, $20^\\circ$, or $30^\\circ$. If a spurious lobe at $-\\theta_{\\mathrm{out}}$ disappears, a new strong lobe appears, or the relative amplitudes of the three dominant directions change by more than a few dB, then the three-ray model as stated fails for non-normal TX positions.","supporting_citations":[{"cited_title":"Reconfigurable intelligent surfaces for THz: Signal processing and hardware design challenges,","cited_arxiv_id":null,"evidence_quote":"Supplies the 2-bit unit-cell design rationale and the earlier observation that spurious rays are inevitable in RIS scattering."},{"cited_title":"Hybrid channel model for low terahertz links in a data center,","cited_arxiv_id":null,"evidence_quote":"Provides the hybrid ray-optical channel modeling approach that the paper wants RISs to plug into."},{"cited_title":"Terahertz wireless channels: A holistic survey on measurement, modeling, and analysis,","cited_arxiv_id":null,"evidence_quote":"Documents experimental validation of ray tracing at THz frequencies, motivating the three-ray simplification."},{"cited_title":"True time-delay beamsteering for radar,","cited_arxiv_id":null,"evidence_quote":"Gives the frequency-dependent beam-pointing formula that the paper uses to express ray shifts across bandwidth."}],"review_version":1}