{"id":"fc515182-79df-42de-bba5-746cc0f0f460","arxiv_id":"1908.08747","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"For a sufficiently large anomalous-reflector RIS, the paper's model predicts spectral efficiencies comparable to an ideal full-duplex relay, without a power amplifier or a half-duplex penalty.","lead":"This article compares two ways to route wireless signals around obstacles: passive reconfigurable surfaces that redirect radio waves, and relays that actively receive and retransmit them. It argues that if the surface is large enough, it can match or beat an ideal relay's data rate with far less hardware.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The conclusion that a sufficiently large RIS can match an ideal FD relay rests entirely on the unvalidated path-loss scaling of companion paper [11]; Sec. VI concedes coupling is ignored and experimental validation is open.","rationale":"The reader's weakest-assumption identification is correct and is the single most load-bearing issue. Every quantitative figure in Sec. V imports the companion paper's path-loss model wholesale. The paper is otherwise a tutorial; the qualitative comparison of hardware, noise, duplexing, and scaling laws is plausible and internally consistent. The phrase 'Exact' in Fig. 3 refers to exact evaluation of [11, Eq. (3)], not to an experimentally verified model, so it cannot serve as independent support. The paper's own Sec. VI openly lists the missing pieces: a physics-based model that accounts for meta-atom coupling, and experimental validation of the scaling laws. These are not rhetorical caveats; they directly gate the headline. A full-wave simulation with realistic coupling is a concrete and decisive check because it bypasses the unresolved debate between product-law and sum-law path-loss models (cf. refs [11] and [12]) and tests the actual configuration used in Fig. 3. One secondary observation: Table I defines the relay self-interference as IS=10N0PR, which is dimensionally inconsistent since N0 and PR are both powers. This appears to be a typographical error and does not affect the comparison against the ideal FD relay, so it should be corrected but does not change the verdict. Overall, the paper's own caution indicates that the quantitative claim should be accepted only conditionally on validation of [11]; the reader's CONDITIONAL verdict remains appropriate, and my analysis does not move it.","tokens_in":11078,"tokens_out":13072,"duration_ms":125986,"concrete_test":"Run a full-wave electromagnetic simulation (e.g., method of moments or FDTD) of the exact Fig. 3 configuration: a 2L=1.5 m, 140λ strip at 28 GHz, phase gradient for 45° incidence and 60° reflection, line source at distance d0, receiver at distance d0 on the reflected path, for d0 = 10, 25, 50, 100, 150 m. Compare the simulated received power against [11, Eq. (3)] and the approximations [11, Eqs. (10), (11)]. If the simulated received power at d0=50 m is more than 3 dB below (αkd_SR+βkd_RD)^-1, or if the crossover distance where the RIS rate drops below the ideal FD relay shifts by more than 20%, the conclusion of Sec. IV-G is not supported independent of [11]. The simulation should include realistic meta-atom mutual coupling or a surface impedance boundary condition, since Sec. VI explicitly identifies coupling as an omitted effect.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. IV-F and Figs. 3-5 quantify the RIS rate using [11, Eq. (3)] and the asymptotic laws [11, Eqs. (10), (11)] without derivation or independent validation. The central claim in Sec. IV-G—'sufficiently large RISs can outperform relay-aided systems'—follows from the electrically large law (αkd_SR+βkd_RD)^-1, which makes a finite anomalous-mirror strip behave like a line-of-sight link of length d_SR+d_RD. If the true received power of a 140λ strip at 28 GHz carries an extra aperture- or coupling-dependent loss relative to this ideal sum-distance law, the crossover distances in Figs. 3-5 shift and the 'similar rate as an ideal FD relay' at d0=10–50 m can fail. The paper itself flags exactly this vulnerability: Sec. VI (Physics-Based Modeling) states that spatial coupling among the meta-atoms is ignored, and Sec. VI (Experimental Validation) states that only a few experimental results validate the scaling laws. Hence the quantitative comparison, and therefore the headline, is model-dependent until [11]'s scaling is independently confirmed.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a qualitative and quantitative comparison between reconfigurable intelligent surfaces (RISs) configured as anomalous reflectors and relay-aided transmission. The authors identify differences in hardware complexity, noise, spectral efficiency, power budget, and SNR scaling, and they introduce scaling laws taken from the companion paper [11]. They report numerical results at 28 GHz and sweeps over frequency and RIS size, concluding that a sufficiently large RIS can outperform an ideal full-duplex decode-and-forward relay in terms of data rate. The paper also lists open issues: physics-based modeling, experimental validation, constrained system design, and information-theoretic aspects.","tokens_in":11340,"tokens_out":4659,"duration_ms":49891,"significance":"If the imported path-loss scaling is correct, the central conclusion is significant: a nearly passive surface could rival or exceed a full-duplex relay at millimeter-wave frequencies while reducing implementation complexity. The qualitative taxonomy of differences between RISs and relays is a useful contribution for the community, and the paper is explicit about several open issues. However, the headline quantitative claim is conditional on an unvalidated external model, so the significance is not yet established; the paper's own Section VI acknowledges this vulnerability.","major_comments":[{"comment":"The quantitative comparison rests entirely on the RIS path-loss model of [11], specifically Eqs. (10) and (11), which is neither derived nor independently measured in this manuscript. Section VI explicitly concedes that current models ignore spatial coupling among meta-atoms and that only a few experimental results have validated the scaling laws. Because the crossover distances in Figs. 3-5 and the conclusions in Sec. IV-G and the Conclusions all follow from this model, the central claim should be presented as conditional on the model of [11] unless the authors include a self-contained derivation and a statement of the model's domain of validity. As written, the headline claim is load-bearing on an external, unvalidated model.","section":"Sec. IV-F, IV-G, Table I, Figs. 3-5"},{"comment":"The asymptotic law (αkd_SR+βkd_RD)^{-1} asserts that a finite anomalous-mirror strip provides received power equivalent to a line-of-sight path of length d_SR+d_RD, independent of the strip length 2L. This is a strong physical claim, yet the paper does not state a quantitative condition for when a given RIS of length 2L is 'large enough' for this regime; the threshold d0≈25-50 m in Fig. 3 is inferred visually from the plots. Since the regime boundary is part of the imported model from [11], the authors should either provide its derivation or explicitly state that this regime boundary has the same validation status as the rest of the model.","section":"Sec. IV-F and Fig. 3"}],"minor_comments":[{"comment":"The phrase 'without the need of using a power amplifier' should be qualified: in the RIS case the source radiates the full power P, whereas in the relay case the power is split between the source and the relay; the quantitative comparison is fair, but the hardware-complexity discussion should acknowledge that the source in the RIS case still requires a power amplifier.","section":"Sec. V.A"},{"comment":"The self-interference value I_S = 10 N0 P_R appears to be an arbitrary representative choice; please state explicitly that this is a fixed representative value and consider including a sensitivity check with respect to this parameter.","section":"Table I"},{"comment":"The x-axis label in the typeset version of Fig. 4 appears garbled as '0 1 53 04 56 07 59 0 1 0 0'; it should read 'Frequency [GHz]'.","section":"Fig. 4 caption"},{"comment":"The statement that the SNR of an RIS 'increases quadratically with N' could be misunderstood; this scaling assumes the energy-conservation and no-coupling model that the paper later acknowledges as an open issue, and the statement would benefit from explicitly invoking those assumptions.","section":"Sec. IV-E"},{"comment":"The final conclusion is stated in absolute terms; consider leading with a conditional such as 'Under the analytical model of [11]' to align the conclusion with the acknowledged limitations in Section VI.","section":"Conclusions"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a hybrid of a tutorial comparison and a quantitative claim. The reliance on the companion paper [11] for the central scaling law is transparent, but the conclusions are stated more strongly than the evidence supports. If the journal typically publishes viewpoint or magazine-style papers, the authors may only need to qualify the claims; if the journal expects self-contained technical validation, the current dependence on [11] is a substantive gap. I recommend major revision so that the authors either add the missing derivation/validation context or soften the headline, rather than rejection, since the qualitative contribution is useful and the paper is well written."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one if you want a concise map of the RIS-vs-relay debate at mmWave frequencies. The paper does what it sets out to do: it lays out the qualitative differences (hardware, noise, duplexing, power budget), derives the relevant scaling laws in a few lines, and then runs a numerical comparison of an anomalous-reflector RIS against DF relays under a total power constraint. The exposition is genuinely good, especially the discussion of why RIS signal-to-noise grows quadratically with the number of elements while relays grow linearly, and why the path-loss dependence on distance can erase that advantage.\n\nWhat is actually new is limited but real: a systematic side-by-side of an anomalous mirror RIS with half- and full-duplex DF relays, including an ideal FD relay for reference. No new equations are introduced—the RIS field expressions are taken from the companion paper [11] and the relay formulas are textbook—but the comparison and the figures are useful for intuition. The paper's own Sec. VI explicitly flags the two soft spots: current RIS models ignore spatial coupling among meta-atoms, and only a few experiments have validated the scaling laws. That is honest, and it is the right place to put the caveat.\n\nThe load-bearing claim is in Sec. IV-G: a sufficiently large RIS can outperform an ideal FD relay. That claim follows directly from the electrically-large-RIS asymptote (αkd_SR + βkd_RD)^{-1} taken from [11]. If that model is optimistic—say, if a real 140λ strip at 28 GHz has extra aperture or coupling loss—then the crossover distances in Figs. 3–5 shift and the headline rate comparisons become weaker. This is a genuine dependency, not a manufactured flaw, but the paper never hides it. The numerical results are self-consistent once you accept [11], and the self-citation is not abusive; it is just a strict inheritance of the model's validity. No code or data are provided, but for a tutorial-level comparison that is a minor omission, not a fatal one.\n\nBottom line: this is a well-written, honest tutorial and parameter study, not a breakthrough. The right reader is someone entering the RIS field who wants a crisp summary of where the technology stands relative to relays. A serious referee should engage with it—ideally pushing the authors to state more prominently that the quantitative results are conditional on [11] being correct, and maybe to include a sensitivity check under a less optimistic RIS path-loss model. It deserves peer review, not desk rejection.\n\nBest,\n[Your name]","headline":"A clear, honest tutorial-level comparison of RISs and relays whose quantitative conclusion rests on the companion path-loss model [11] rather than on new analysis or measurements.","tokens_in":11919,"tokens_out":1294,"would_cite":true,"duration_ms":14567,"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":"The paper claims that sufficiently large reconfigurable intelligent surfaces can match or exceed relay data rates with lower hardware complexity.","keywords":["reconfigurable intelligent surfaces","relaying","anomalous reflection","path loss model","spectral efficiency","millimeter wave","smart radio environments","meta-surfaces"],"falsifier":"Measure the received power at 28 GHz for a fixed RIS of length 1.5 m placed equidistant from a transmitter and receiver, for $d_{0}$ from 10 m to 200 m, and compare the measured distance exponent with $(\\alpha k d_{\\mathrm{SR}} + \\beta k d_{\\mathrm{RD}})^{-1}$ in the short-range regime and $4L^{2}(d_{\\mathrm{SR}} d_{\\mathrm{RD}})^{-1}$ in the long-range regime; also measure how the SNR grows as the surface length is doubled. A distance exponent worse than these laws, or a sub-quadratic gain in length, would falsify the claimed crossover against an ideal full-duplex relay.","tokens_in":10903,"feed_emoji":"📡","tokens_out":8364,"duration_ms":71848,"temperature":0.7,"pith_summary":"Reconfigurable intelligent surfaces are nearly passive metasurfaces that redirect radio waves, while relays are active devices that receive, amplify or decode, and retransmit. The paper argues that the decisive factor in comparing them at millimeter-wave frequencies is the electrical size of the surface. When the RIS is large relative to the wavelength, its received-power scaling becomes comparable to a relay's, and because it avoids half-duplex loss, loop-back self-interference, and a dedicated power amplifier, it can deliver equal or higher spectral efficiency. The reported numerical comparisons at 28 GHz show a 1.5 m RIS matching an ideal full-duplex relay over typical indoor and outdoor distances. This matters because it identifies a concrete regime where relaying hardware could be replaced by lower-complexity nearly passive structures.","feed_headline":"Large passive surfaces can match or beat relays in data rate","feed_subtitle":"A sufficiently large RIS avoids half-duplex loss and power amplifiers, matching an ideal full-duplex relay.","key_machinery":"The load-bearing machinery is a pair of distance-scaling laws for the received power of an RIS, taken from the companion analysis [11]. For an electrically large RIS (geometric size large compared with wavelength and distances), the received power behaves like that of an anomalous mirror, $\\propto (\\alpha k d_{\\mathrm{SR}} + \\beta k d_{\\mathrm{RD}})^{-1}$ with $\\alpha$ and $\\beta$ fixed by the incidence and reflection angles [11, Eq. (10)]. For an electrically small RIS, it behaves like a diffuser, $\\propto 4L^{2}(d_{\\mathrm{SR}} d_{\\mathrm{RD}})^{-1}$ [11, Eq. (11)]. These are set against the relay baselines: half-duplex decode-and-forward halves the rate, full-duplex adds residual self-interference, and the end-to-end SNR scales with the weaker hop; for a multi-antenna relay the SNR grows linearly with $N$, whereas for an RIS with $N$ individually tunable elements it grows quadratically with $N$. The length $2L$ of the RIS is linked to the number of meta-atoms by $2L = M_{\\mathrm{ma}}\\lambda/D$, which is what turns the $4L^{2}$ factor into a quadratic gain in element count.","core_discovery":"The paper's central claim is that a reconfigurable intelligent surface, when made sufficiently large in terms of wavelengths, can deliver end-to-end data rates at least as high as an ideal full-duplex decode-and-forward relay, while needing no power amplifier, no reception chain, and no half-duplex scheduling at the surface. The argument is carried by path-loss scaling laws: an electrically large RIS behaves as an anomalous mirror whose received power scales as $(\\alpha k d_{\\mathrm{SR}} + \\beta k d_{\\mathrm{RD}})^{-1}$, while a relay's end-to-end signal-to-noise ratio is set by the weaker of the two hops, $\\min\\{(kd_{\\mathrm{SR}})^{-1}, (kd_{\\mathrm{RD}})^{-1}\\}$; an electrically small RIS behaves as a diffuser with received power scaling as $4L^{2}(d_{\\mathrm{SR}} d_{\\mathrm{RD}})^{-1}$, the same distance dependence as amplify-and-forward relaying but with a factor proportional to the square of the surface length. Because the RIS uses the full transmit power, adds no noise, and suffers no duplexing loss, the comparison favors the RIS exactly when its aperture is large enough. The 28 GHz numerical examples show a 1.5 m RIS (140 wavelengths) tracking an ideal full-duplex relay out to roughly 150 m, and a focusing-lens configuration doing better still.","pith_inferences":["The aperture-versus-power tradeoff suggests a rule for system designers: at fixed transmit power, rate can be bought with surface area rather than amplifier power, so the relevant comparison metric is cost per bit, which this paper does not compute.","Because electrical size is measured in wavelengths, the same physical surface becomes more favorable as carrier frequency rises; the frequency sweeps hint that sub-terahertz deployments could favor RISs even more, although the path-loss model's validity at those frequencies remains untested.","The single-surface, free-space setup is a best-case comparison; in a multi-cell environment, relay noise and RIS configuration overhead enter differently, so the crossover distances should be re-derived for stochastic deployments and for surfaces with mutual coupling between meta-atoms."],"forward_implications":["A sufficiently large RIS can deliver data rates comparable to an ideal full-duplex relay without a transmit amplifier at the surface, because it uses the full transmit power and adds no receiver noise.","At short ranges, where the RIS is electrically large, its distance scaling matches a relay's, so avoiding half-duplex and self-interference losses lets it win.","At long ranges, an electrically small RIS suffers a steeper distance loss, but that loss can be offset by increasing the surface size because the SNR grows with the square of the number of meta-atoms.","A focusing-lens RIS outperforms an anomalous-reflector RIS, but requires knowing receiver positions and channel-adaptive phases, whereas a long phase-gradient surface alone can already beat an ideal full-duplex relay.","At higher carrier frequencies, a fixed-size RIS becomes electrically larger, which is why at 100 m and frequencies above roughly 20 GHz the RIS matches an ideal full-duplex relay in the paper's example."],"supporting_citations":[{"why":"Supplies the RIS path-loss formulas, including Eqs. (10) and (11), used for every RIS rate curve in the comparison.","marker":"[11]"},{"why":"Provides the relay signal-to-noise ratio scaling and duplexing constraints used as the baseline.","marker":"[2]"},{"why":"Gives the prior energy-efficiency comparison of DF relays and focusing-lens RISs that this paper extends to anomalous reflectors.","marker":"[12]"},{"why":"Documents an RIS prototype with thousands of inexpensive antennas and the quadratic SNR scaling with element count.","marker":"[8]"},{"why":"Reports an RIS prototype and path-loss measurements that the paper cites as the main experimental evidence for the size-dependent scaling laws.","marker":"[13]"},{"why":"Establishes the generalized laws of reflection from phase-gradient metasurfaces that underlie the anomalous-reflector operation of RISs.","marker":"[7]"}],"fun_headline_variants":["Electrically large surfaces mimic ideal relays","Big passive mirrors relay data without noise","No amplifier needed: big RIS matches relays","Oversized RIS outpaces relays when large","Size matters: large RIS beats relay limits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire quantitative comparison presumes the companion path-loss model for RISs, especially Eqs. (10) and (11), which specify how received power scales with distance and surface size; the article cites that model without deriving or independently measuring it, and it explicitly lists experimental validation of these scaling laws as an open issue.","fun_headline_variants_meta":{"raw":{"variants":["Electrically large surfaces mimic ideal relays","Big passive mirrors relay data without noise","No amplifier needed: big RIS matches relays","Oversized RIS outpaces relays when large","Size matters: large RIS beats relay limits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000354,"raw_usage":{"total_tokens":1943,"prompt_tokens":981,"completion_tokens":962,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":597,"completion_tokens_details":{"reasoning_tokens":897}},"tokens_in":597,"tokens_out":962,"duration_ms":11003,"temperature":1.0,"reasoning_tokens":897,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:30:46.531130+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the received power at 28 GHz for a fixed RIS of length 1.5 m placed equidistant from a transmitter and receiver, for $d_{0}$ from 10 m to 200 m, and compare the measured distance exponent with $(\\alpha k d_{\\mathrm{SR}} + \\beta k d_{\\mathrm{RD}})^{-1}$ in the short-range regime and $4L^{2}(d_{\\mathrm{SR}} d_{\\mathrm{RD}})^{-1}$ in the long-range regime; also measure how the SNR grows as the surface length is doubled. A distance exponent worse than these laws, or a sub-quadratic gain in length, would falsify the claimed crossover against an ideal full-duplex relay.","supporting_citations":[{"cited_title":"Analytical Modeling of the Path-Loss for Reconfigurable Intelligent Surfaces -- Anomalous Mirror or Scatterer ?","cited_arxiv_id":"2001.10862","evidence_quote":"Supplies the RIS path-loss formulas, including Eqs. (10) and (11), used for every RIS rate curve in the comparison."},{"cited_title":"Stochastic geometry modeling and system- level analysis & optimization of relay-aided downlink cellular net- works","cited_arxiv_id":null,"evidence_quote":"Provides the relay signal-to-noise ratio scaling and duplexing constraints used as the baseline."},{"cited_title":"Intelligent reflecting surface vs. decode-and- forward: How large surfaces are needed to beat relaying?","cited_arxiv_id":null,"evidence_quote":"Gives the prior energy-efficiency comparison of DF relays and focusing-lens RISs that this paper extends to anomalous reflectors."},{"cited_title":"RFocus: Practical beamforming for small devices","cited_arxiv_id":null,"evidence_quote":"Documents an RIS prototype with thousands of inexpensive antennas and the quadratic SNR scaling with element count."},{"cited_title":"Wireless Communications with Reconfigurable Intelligent Surface: Path Loss Modeling and Experimental Measurement","cited_arxiv_id":"1911.05326","evidence_quote":"Reports an RIS prototype and path-loss measurements that the paper cites as the main experimental evidence for the size-dependent scaling laws."},{"cited_title":"Light propagation with phase discontinuities: Generalized laws of reflection and refraction","cited_arxiv_id":null,"evidence_quote":"Establishes the generalized laws of reflection from phase-gradient metasurfaces that underlie the anomalous-reflector operation of RISs."}],"review_version":1}