Pith. sign in

REVIEW 2 major objections 5 minor 1 cited by

Reconfigurable Intelligent Surfaces vs. Relaying: Differences, Similarities, and Performance Comparison

T0 review · 2 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that sufficiently large reconfigurable intelligent surfaces can match or exceed relay data rates with lower hardware complexity.

desk verdict 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. read the letter →

arxiv 1908.08747 v2 pith:ZKSELVOR submitted 2019-08-23 eess.SP cs.ITmath.IT

M. Di Renzo , K. Ntontin , J. Song , F. H. Danufane , X. Qian , F. Lazarakis , J. de Rosny , D.-T. Phan-Huy
show 7 more authors
O. Simeone R. Zhang M. Debbah G. Lerosey M. Fink S. Tretyakov S. Shamai (Shitz)
This is my paper · ORCID
classification eess.SPcs.ITmath.IT
keywords reconfigurableintelligentsurfacesrelayinganomalousreflectionpathlossmodelspectralefficiencymillimeterwavesmartradioenvironmentsmeta-surfaces
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

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.

What carries the argument

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.

What would settle it

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.

Watch

Extended reading notes

Core claim

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.

Load-bearing premise

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.

Editorial extensions

If this is right

  • 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.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • 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.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

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.

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 (2)
  1. [Sec. IV-F, IV-G, Table I, Figs. 3-5] 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.
  2. [Sec. IV-F and Fig. 3] 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.
minor comments (5)
  1. [Sec. V.A] 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.
  2. [Table I] 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.
  3. [Fig. 4 caption] 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]'.
  4. [Sec. IV-E] 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.
  5. [Conclusions] 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.

Circularity Check

1 steps flagged · score 4.0 of 10

The quantitative RIS-vs-relay comparison is built on the self-cited companion path-loss model [11], whose empirical validation the paper admits is still open; the headline conclusion therefore inherits that model rather than being independently demonstrated.

  1. self citation load bearing [Sec. IV-F (Average Signal-to-Noise Ratio vs. Transmission Distance), Table I, and Numerical Results in Sec. V]
    "Based on [11, Sec. III-B], two notable regimes are worth of analysis. ... Electrically large RISs: If the geometric size of the RIS is large enough ... the power received from the RIS and the end-to-end average signal-to-noise ratio at the receiver scale ... as (αkdSR + βkdRD)−1 ... [11, Eq. (10)]; Electrically small RISs: ... 4L2(dSRdRD)−1 ... [11, Eq. (11)]. ... The intensity of the electric field is obtained from the analytical frameworks in [11], as reported in Table I."

    The central conclusion — 'RIS-aided transmission may outperform relay-aided transmission provided that the size of the RIS is sufficiently large' — is obtained from the electrically-large and electrically-small scaling laws given in [11, Eqs. (3), (10), (11)]. Those laws are not derived in this paper; they are imported from a companion paper by the same research group and are used directly in Table I and Figs. 3-5. The paper itself concedes in Sec. VI that spatial coupling among meta-atoms is ignored and that only 'a few experimental results' have validated the scaling laws. Thus the quantitative comparison, and the headline, reduce to a self-cited analytical model whose empirical status is explicitly admitted to be open.

full rationale

The paper's relay analysis and qualitative differences (hardware complexity, noise, power budget, duplexing constraints) rest on standard communication theory and are not circular. The RIS curves, however, are obtained entirely from the self-cited companion model [11] (exact field expression and large/small distance asymptotics), with no derivation of that model inside this paper. Because the paper itself flags physics-based modeling (coupling ignored) and experimental validation as open issues, the claim that a sufficiently large RIS can match or exceed an ideal full-duplex relay is conditional on [11]'s scaling laws. That is load-bearing self-citation, but it is not the construction-equivalence or fitted-prediction form of circularity; no data are fitted and no uniqueness claim is imported. On the 0-10 scale this warrants a 4 rather than 0, since the central quantitative claim inherits an unvalidated self-cited input, but it is not fully circular.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

All quantitative results rest on the companion path-loss model [11] and on hand-chosen system parameters. No new physical entities, particles, or forces are introduced, and no parameters are fitted to experimental data.

free parameters (5)
  • Reference SNR P/N0 at 1 m = 114 dB
    Case-study value used in all numerical results; the crossover distances in Figs. 3-5 scale with this value.
  • Full-duplex relay residual self-interference I_S = 10 N0 P_R
    Hand-chosen residual self-interference level in Table I; not based on measurement, but it affects the FD relay rates.
  • RIS total length 2L = 1.5 m (140 lambda at 28 GHz)
    Illustrative size used in Figs. 3 and 4; the 'sufficiently large' conclusion depends on the chosen size.
  • RIS incidence and reflection angles = 45 and 60 degrees relative to the normal
    Geometry for the RIS in Figs. 3-5; these angles affect alpha and beta in the anomalous-mirror formula.
  • Relay transmit power split = P_R = P/2
    Under the total power constraint, the source and relay each get half the power, while the RIS uses the full source power P; this fairness assumption favors the RIS.
assumptions (6)
  • domain assumption Free-space cylindrical wave model with |E(d)|^2 proportional to (kd)^-1
    Invoked in Sec. IV-F and Table I; a 2D model that may not match all 3D deployment geometries.
  • domain assumption Electrically large RIS acts as an anomalous mirror with received power scaling (alpha*k*d_SR + beta*k*d_RD)^-1 from [11, Eq. (10)]
    Core RIS model used for the mirror curves in Fig. 3; imported from the companion paper without derivation here.
  • domain assumption Electrically small RIS acts as a diffuser with received power scaling 4L^2(d_SR*d_RD)^-1 from [11, Eq. (11)]
    Used for the long-distance approximation; same scaling as AF relaying; imported from the companion paper.
  • domain assumption A passive or nearly passive RIS reflects the total incident power and adds no noise
    Used in Sec. IV-D and Table I to avoid noise and power-split penalties; an idealization not experimentally validated.
  • standard math Standard relay formulas for half-duplex and full-duplex decode-and-forward relaying
    Used in Table I and Sec. IV; these are standard results from the cited relaying literature, independent of the RIS model.
  • domain assumption Residual self-interference model I_S = 10 N0 P_R for full-duplex relays
    Used in Table I and Fig. 3; a representative but hand-chosen model of relay impairment.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reconfigurable Intelligent Surfaces vs. Relaying: Differences, Similarities, and Performance Comparison." pith.science (2026). https://pith.science/paper/ZKSELVOR

@misc{pith2026190808747,
  author       = {Pith},
  title        = {Pith review of: Reconfigurable Intelligent Surfaces vs. Relaying: Differences, Similarities, and Performance Comparison},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZKSELVOR}},
  note         = {Machine review of arXiv:1908.08747}
}
read the original abstract

Reconfigurable intelligent surfaces (RISs) have the potential of realizing the emerging concept of smart radio environments by leveraging the unique properties of meta-surfaces. In this article, we discuss the potential applications of RISs in wireless networks that operate at high-frequency bands, e.g., millimeter wave (30-100 GHz) and sub-millimeter wave (greater than 100 GHz) frequencies. When used in wireless networks, RISs may operate in a manner similar to relays. This paper elaborates on the key differences and similarities between RISs that are configured to operate as anomalous reflectors and relays. In particular, we illustrate numerical results that highlight the spectral efficiency gains of RISs when their size is sufficiently large as compared with the wavelength of the radio waves. In addition, we discuss key open issues that need to be addressed for unlocking the potential benefits of RISs.

Figures

Figures reproduced from arXiv: 1908.08747 by the authors.

Figure 1
Figure 1. Possible uses of reconfigurable intelligent surfaces. (i) Anomalous reflection: a radio wave incident at an angle of 90 degrees is reflected towards an angle of 45 degrees. (ii) Focusing lens: a radio wave incident at an angle of 90 degrees is focused (beamforming) towards a specified location in order to maximize the energy at that point. (iii) An RIS illuminated by a feeder reflects two phase-modulated signals by … view at source ↗
Figure 2
Figure 2. , where four application scenarios are identified. Signal engineering: Assume that small cell 1 wishes to communicate with mobile terminal (MT) 1, but the LOS link is blocked by an object. In this case, small cell 1 redirects the transmitted beam towards RIS 1 that coats object 1, and assists the communication by shaping the incident wave towards MT 1 so that the received signal strength is maximized. Interference e… view at source ↗
Figure 3
Figure 3. Data rate of RISs and relays versus the transmission distance. the potential of providing a better rate than relays if, for a fixed size of the RIS, the distances are not too long; • Compared with relays, electrically small RISs (i.e., with a slight abuse of terminology, for long distances d0) offer a less favorable scaling law as a function of the distance. However, the average end-to-end signal-to-noise ratio of e… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Data rate of RISs and relays versus the transmission frequency. 0 0.25 0.5 0.75 1 1.25 1.5 1.75 L [m] 8 10 12 14 16 18 20 22 Rate [bit/s/Hz] d0 = 10 m HD DF FD DF FD DF - Ideal RIS - Exact 0 0.25 0.5 0.75 1 1.25 1.5 1.75 L [m] 8 10 12 14 16 18 20 22 Rate [bit/s/Hz] d0 …
Figure 5
Figure 5. Figure 5: Data rate of RISs and relays versus the size of the RIS. rate of an ideal FD relay, in which the residual loop-back self￾interference is assumed to be zero. A total power constraint is assumed and, therefore, the total power is equally split between the transmitter and…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Reconfigurable Intelligent Surface Assisted UAV Communication: Joint Trajectory Design and Passive Beamforming

    cs.IT 2019-08 conditional novelty 5.0 of 10

    A UAV and a reconfigurable intelligent surface can be jointly optimized, with a closed-form phase alignment and a successive-convex-approximation trajectory design, to raise the average downlink rate above trajectory-...

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages · cited by 1 Pith paper

  1. [11]

    Analytical Modeling of the Path-Loss for Reconfigurable Intelligent Surfaces -- Anomalous Mirror or Scatterer ?

    M. Di Renzo et al., “Analytical modeling of the path-loss for re- configurable intelligent surfaces – Anomalous mirror or scatterer?”, arXiv:2001.10862

    work page Pith review arXiv 2001

  2. [1]

    Wireless communications and applications above 100 GHz: Opportunities and challenges for 6G and beyond

    T. S. Rappaport et al., “Wireless communications and applications above 100 GHz: Opportunities and challenges for 6G and beyond”, IEEE Access, vol. 7, 2019

    work page 2019

  3. [2]

    Stochastic geometry modeling and system- level analysis & optimization of relay-aided downlink cellular net- works

    W. Lu and M. Di Renzo, “Stochastic geometry modeling and system- level analysis & optimization of relay-aided downlink cellular net- works”, IEEE Trans. Commun. , vol. 63, Nov. 2015

    work page 2015

  4. [3]

    Coverage Enhancement for NLOS mmWave Links Using Passive Reflectors

    W. Khawaja et al., “Coverage enhancement for NLOS mmWave links using passive reflectors”, arXiv:1905.04794

    work page Pith review arXiv 1905

  5. [4]

    A new wireless communication paradigm through software-controlled metasurfaces

    C. Liaskos et al., “A new wireless communication paradigm through software-controlled metasurfaces”, IEEE Commun. Mag. , Sep. 2018

    work page 2018

  6. [5]

    Wireless communications through reconfigurable intelligent surfaces

    E. Basar et al., “Wireless communications through reconfigurable intelligent surfaces”, IEEE Access, vol. 7, Aug. 2019

    work page 2019

  7. [6]

    Smart radio environments empowered by recon- figurable AI meta-surfaces: An idea whose time has come

    M. Di Renzo et al., “Smart radio environments empowered by recon- figurable AI meta-surfaces: An idea whose time has come”, EURASIP J. Wireless Commun. Net. , vol. 129, May 2019

    work page 2019

  8. [7]

    Light propagation with phase discontinuities: Generalized laws of reflection and refraction

    N. Yu et al., “Light propagation with phase discontinuities: Generalized laws of reflection and refraction”, Science, vol. 334, Oct. 2011

    work page 2011

Show all 15 references
  1. [8]

    RFocus: Practical beamforming for small devices

    V . Arun and H. Balakrishnan, “RFocus: Practical beamforming for small devices”, USENIX, 12 pages, Feb. 2020

    2020

  2. [9]

    Beyond max-SNR: Joint encoding for reconfigurable intelligent surfaces

    R. Karasik et al., “Beyond max-SNR: Joint encoding for reconfigurable intelligent surfaces”, IEEE ISIT, arXiv:1911.09443

    1911 arXiv

  3. [10]

    Wireless communications with programmable meta- surface: New paradigms, opportunities, and challenges on transceiver design

    W. Tang et al., “Wireless communications with programmable meta- surface: New paradigms, opportunities, and challenges on transceiver design”, IEEE Wireless Commun., arXiv:1907.01956

    1907 arXiv

  4. [12]

    Intelligent reflecting surface vs. decode-and- forward: How large surfaces are needed to beat relaying?

    E. Bjornson et al., “Intelligent reflecting surface vs. decode-and- forward: How large surfaces are needed to beat relaying?”, IEEE Wireless Commun. Lett., vol. 9, Feb. 2020

    2020

  5. [13]

    Wireless communications with reconfigurable intel- ligent surface: Path loss modeling and experimental measurement

    W. Tang et al., “Wireless communications with reconfigurable intel- ligent surface: Path loss modeling and experimental measurement”, arXiv:1911.05326

    1911 arXiv

  6. [14]

    A communication model for large intelligent surfaces

    R. J. Williams et al., “A communication model for large intelligent surfaces”, arXiv:1912.06644

    1912 arXiv

  7. [15]

    Cascaded channel estimation for large intel- ligent metasurface assisted massive MIMO

    Z.-Q. He and X. Yuan, “Cascaded channel estimation for large intel- ligent metasurface assisted massive MIMO”, IEEE Wireless Commun. Lett., vol. 9, Feb. 2020

    2020

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.