Control Signaling for Reconfigurable Intelligent Surfaces: How Many Bits are Needed?

Anders Enqvist; Cicek Cavdar; Emil Bj\"ornson; \"Ozlem Tu\u{g}fe Demir

REVIEW 2 major objections 4 minor 15 references

Control Signaling for Reconfigurable Intelligent Surfaces: How Many Bits are Needed?

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper proves that a line-of-sight RIS can be configured with O(log N) feedback bits while keeping array gain above half its maximum, and gives the exact bit count.

desk verdict A solid, usable result: about log2 N + 4 feedback bits suffice for an LoS RIS with a bounded 3 dB loss, though the guarantee assumes continuous per-element phase control. read the letter →

arxiv 2506.03929 v1 pith:WFF5LNOB submitted 2025-06-04 cs.IT eess.SPmath.IT

Anders Enqvist , Özlem Tuğfe Demir , Cicek Cavdar , Emil Björnson This is my paper

classification cs.ITeess.SPmath.IT

keywords Reconfigurableintelligentsurfacescontrolsignalingfeedbackquantizationcodebookdesignline-of-sightchannelsarraygainhalf-powerbeamwidthRicianfading

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

This paper asks how many control bits a base station must send to a reconfigurable intelligent surface (RIS) to set its phase shifts, and proposes an answer: for line-of-sight channels, $l = \lceil 1.1746 + \log_2(N) \rceil$ bits suffice to keep the RIS array gain above $N^2/2$ for any angles, with a rule of thumb of $\log_2(N) + 4$ bits total when a static path is present. The key move is to notice that in pure LoS the optimal phase profile is a linear ramp determined by one real number, $\sin\theta_1 + \sin\theta_2$, so the feedback problem reduces to quantizing that number on a uniform grid. The authors contrast this with naive element-wise quantization, whose $N b$ bits scale linearly with the number of elements and quickly become impractical for surfaces with hundreds or thousands of elements. The practical stakes: the control channel is a scarce wireless resource, and this result says a LoS-deployed RIS can be reconfigured at logarithmic cost with a guaranteed bounded SNR loss.

What carries the argument

The load-bearing object is the uniform-grid scalar codebook for $\Theta = \sin\theta_1 + \sin\theta_2$, with entries $D_i = \mathrm{diag}(1, e^{j\pi\Theta_i}, \ldots, e^{j\pi(N-1)\Theta_i})$. Since the LoS-optimal RIS phase shift at element $n$ is $\psi_n = \pi(n-1)\Theta$, one quantized scalar determines the whole configuration. The analysis uses the standard approximation of the array factor as $N\,\mathrm{sinc}(N(\Theta - \hat{\Theta})/2)$, whose half-power beamwidth $\mathrm{HPBW}(N) = 1.772/N$ converts the requirement that every angle lie within the main lobe into the inequality $4/2^l \leq 1.772/N$, which yields $l \geq \log_2(4N/1.772)$. Lemma 1 characterizes element-wise quantization via $\mathbb{E}\{G_A\} = N^2\,\mathrm{sinc}^2(1/2^b) + N(1-\mathrm{sinc}^2(1/2^b))$, and Lemma 2 quantifies the static-path alignment term with a $\mathrm{sinc}(1/2^d)$ factor.

What would settle it

Take $N=128$, pick the worst-case angle where the chosen codebook entry deviates by half a cell width, then compute the realized array gain after rounding each phase shift to the nearest of four discrete phase levels; if the gain falls below $N^2/2$, the guarantee in Eq. (15) does not survive discrete phase hardware.

Watch

Extended reading notes

Core claim

The paper's central claim is that codebook-based feedback can realize near-optimal RIS beamforming in LoS channels at a cost that grows only logarithmically in the number $N$ of RIS elements. Concretely, if the BS and RIS share a codebook of $2^l$ configurations $D_i = \mathrm{diag}(1, e^{j\pi\Theta_i}, \ldots, e^{j\pi(N-1)\Theta_i})$ with $\Theta_i$ the midpoints of $2^l$ equal subintervals of $(-2,2)$, then choosing $l = \lceil 1.1746 + \log_2(N) \rceil$ guarantees that for every pair of LoS angles $\theta_1,\theta_2$ the array gain satisfies $G_A \geq N^2/2$. The reason is that the optimal phase shifts depend only on $\Theta = \sin\theta_1 + \sin\theta_2$, and the array gain as a function of quantization error follows the sinc-squared pattern whose half-power beamwidth is about $1.772/N$; a uniform grid with cell width no larger than that beamwidth keeps any operating point inside the main lobe. The paper further shows the same codebook works with minor loss under Rician fading when the $\kappa$ factor is large, and that aligning a static BS–UE path only adds $d = 1$ or $2$ extra bits for a common phase rotation, giving the $t = \log_2(N) + 4$ rule of thumb.

Load-bearing premise

The guarantee assumes the RIS elements can realize the exact continuous phase shifts $\psi_n = \pi(n-1)\Theta_i$ specified by the codebook, and that the sinc-based half-power beamwidth approximation remains accurate at the worst-case operating points.

Editorial extensions

If this is right

If the claim is correct, a base station can reconfigure a LoS-deployed RIS once per coherence time with $O(\log_2 N)$ control bits, so feedback overhead no longer scales with the surface size.
A 128-element RIS needs $l = 9$ bits for the LoS codebook, versus 384 bits for near-optimal element-wise quantization with $b = 3$ bits per element.
For Rician channels with $\kappa = 10$ dB, the LoS-tuned codebook still achieves near-maximum SNR at the bit count satisfying the HPBW condition.
When a static BS–UE path is present, allocating one or two of the total bits to the common phase rotation improves SNR even before the LoS codebook condition is fully met.
The rule of thumb $t = \log_2(N) + 4$ bits keeps the receiving BS inside the half-power beamwidth of the reflected beam.

Reading between the lines

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

Ours: the logarithmic guarantee presumes ideal continuous phase control; with 1–2 bit phase-resolution hardware, a natural follow-up is a codebook whose entries are robust to that quantization, or an analysis that folds the hardware resolution into the bit count.
Ours: the uniform grid is a simple quantizer; a nonuniform grid could likely reduce the constant 1.1746 by placing more codewords near the extremes of $\sin\theta_1+\sin\theta_2$, but the paper does not claim optimality.
Ours: the bit count counts the codebook index only; adding channel coding on the control channel would multiply the over-the-air cost by a constant factor, which the paper explicitly sets aside by assuming error-free reception.
Ours: extending the same reasoning to a planar RIS in three dimensions would replace the scalar $\Theta$ with a two-parameter angle pair, so the logarithmic-in-$N$ scaling is plausible but unproven.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Worth knowing: this paper gives a clean closed-form answer to a practical question. For an RIS in pure LoS, with l = ceil(1.1746 + log2 N) feedback bits to convey the quantized slope sin θ1 + sin θ2, the array gain stays above N^2/2 for any pair of angles. Add one or two bits for aligning a static path and you get the rule of thumb t = log2 N + 4. The derivation is short and essentially correct: uniform quantization over (−2, 2), compare cell half-width against the 1.772/N HPBW, get the bit count. The optimal LoS configuration being a linear phase ramp is textbook; the new piece is the quantization analysis and the explicit scaling law, which I don't think is in the cited control-signaling papers. Simulations match the analysis, and the Rician/static-path extensions are handled honestly.

Soft spots, in order of importance. First, the guarantee in (15) presumes the RIS can realize the continuous phases in (16). If the elements have discrete phase resolution (1–3 bits is common), per-element quantization error can push the array gain below N^2/2 near codebook tile boundaries. The paper never states a hardware model for this. That is a genuine gap, but it is fixable: either state the continuous-phase assumption clearly or analyze the robustness. It does not collapse the main claim about feedback-bit scaling, since that is about bits on the control channel, not phase-shifter resolution.

Second, “minimal SNR loss” is a bit strong for a worst-case 3 dB loss. A wording change would fix it.

Third, Lemma 1’s Eq. (20) does not actually describe the element-wise curve in Fig. 2 for the LoS case. The proof assumes independent quantization errors across elements, but for a deterministic linear phase ramp the errors are fully correlated functions of the angle, so the independence premise fails. The lemma is fine for uncorrelated/Rayleigh channels; it just shouldn’t be presented as the analytical model for the LoS comparison. The qualitative point, that element-wise feedback costs N b bits and is wasteful here, still stands.

No parameter fitting, no circularity: the HPBW and channel models come from the authors’ own textbook but are standard external benchmark values. The citation pattern is fine.

For whom: anyone doing system-level feasibility of RIS control, or designing feedback protocols. It deserves a serious referee; the revisions are modest. I’d send it out.

Referee Report

2 major / 4 minor

Summary. This paper considers the uplink of an RIS-aided system and studies how many feedback bits are required to convey the RIS phase configuration from the BS to the RIS. For pure LoS channels, the proposed codebook quantizes the scalar Θ = sinθ1 + sinθ2 on (−2,2); with l = ⌈1.1746 + log2 N⌉ bits the array gain is guaranteed to remain above N^2/2, yielding O(log N) scaling. The design is extended to Rician fading and to the presence of a static BS–UE path, where d additional bits align the common phase of the RIS with the static channel; the paper concludes with the rule of thumb t = log2 N + 4. The paper also analyzes naive element-wise quantization and compares the two approaches via Monte Carlo simulations.

Significance. The logarithmic scaling law is clean, practically relevant, and derived from standard array-factor and HPBW results; the analytical steps in Eqs. (14), (19), (20), and (29) are transparent and check out under their stated assumptions. The paper gives explicit design formulas and a simple rule of thumb, which is valuable for system-level studies of RIS control overhead. The main limitation is that the guarantee in Eq. (15) presumes continuous per-element phase control, an assumption that is never stated; this must be clarified or extended before the headline claim can be accepted as written.

major comments (2)

[Section III-A, Eq. (16)] The N^2/2 guarantee in Eq. (15) presumes the RIS can realize the continuous phase shifts ψ_n = π(n−1)Θ_i for every element. This hardware assumption is never stated in the manuscript. If the RIS has discrete per-element phase resolution (e.g., 1–3 bits), the implemented phases are Q_b(ψ_n) = ψ_n + ε_n, and the array factor becomes Σ e^{j(ε_n − π(n−1)(sinθ1+sinθ2−Θ_i))}; the per-element errors ε_n are not accounted for, and the array gain can fall below N^2/2 near tile boundaries. Since the paper asks 'how many bits are needed?', the distinction between feedback bits and the RIS's internal phase-resolution bits is load-bearing. Please state the continuous-phase assumption explicitly or extend the analysis to discrete-phase RISs and quantify the additional bits required.
[Section III-C, Lemma 1] The independence assumption on the quantization errors is not satisfied by the LoS model of Section III-A, because all optimal phase shifts ψ_n are deterministic functions of the single scalar Θ = sinθ1 + sinθ2. Thus Eq. (20) does not describe element-wise quantization of the LoS-optimal profile, and the analytical comparison with the proposed codebook is not based on the same channel model. Please clarify that Lemma 1 applies to unstructured channels (e.g., Rayleigh fading) where the optimal phases are effectively independent, or provide a LoS/Rician-specific element-wise analysis.

minor comments (4)

[Section IV, paragraph after Eq. (25)] The sentence containing 'l for Dψ and t for ϕ' should read 'l for Dψ and d for ϕ'.
[Conclusion] The rule of thumb t = log2(N) + 4 should be stated more precisely, since t must be an integer and l is a ceiling; for example, t = ⌈log2 N⌉ + 4 is a safe upper bound for all N.
[Eqs. (12), (22), (29)] The function sinc(·) is used without a definition; please state the convention sinc(x) = sin(πx)/(πx) used throughout.
[Fig. 2] The x-axis scaling is not described, although the text refers to operating points 'much further to the right' in a linearly scaled axis; please clarify the axis scaling or replot the figure.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the O(log N) feedback-bit bound is derived from standard HPBW analysis and is not a restatement of its inputs.

full rationale

The derivation chain is self-contained given standard LoS array modeling. The array factor in Eqs. (7)-(12) is obtained from the geometric channel model in Eqs. (4)-(6), and the uniform quantization of Theta over (-2,2) is defined in the paper. The HPBW value 1.772/N is cited from [14], but that formula is a standard parameter-free beamwidth result for the sinc-squared mainlobe and is not the target claim. The condition in Eqs. (13)-(15) is a design inequality: choose l so that the worst-case midpoint quantization error stays inside the half-power width; the guarantee GA >= N^2/2 then follows from the definition of HPBW. This is a design criterion rather than a tautology, because the nontrivial content is the explicit bit count l = ceil(1.1746 + log2(N)) and its logarithmic scaling. Lemma 1 and Lemma 2 are proved from quantization-error distributions; no fitted parameters are renamed as predictions. The Rician and static-path extensions are evaluated against 3GPP parameters and Monte Carlo simulations, not against the derived bounds' own outputs. The only self-citations ([2], [13], [14]) supply standard channel models and beamforming facts, and they are not the load-bearing justification for the bit-scaling claim. The unmodeled discrete per-element phase resolution is a modeling limitation and a correctness/scope risk, not circularity, because Eq. (16) explicitly presumes continuous phase shifts and the guarantee is conditional on that stated model.

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

The central claim is derived rather than fitted: no free parameters enter the codebook design or the bit-count formula. The main pulled-in items are standard beamforming identities (HPBW, ray-based far-field channel model) and the channel/control assumptions. The most fragile unstated item is the continuous-phase realizability of RIS elements, axiom 2.

assumptions (6)

domain assumption The RIS and BS are ULAs with half-wavelength spacing, and the RIS is in the far field of BS and UE, so channels are array-response vectors with a common gain (Eqs. 4-6).
Used throughout Section III to express the channel as βrβt aK(φ) aN^T(θ1) Dψ aN(θ2) and to write the array gain as a squared array factor.
domain assumption The optimal RIS phase shifts for LoS are the linear ramp ψn = π(n−1)(sin θ1 + sin θ2), and the surface can set each element to exactly these continuous values (Eqs. 8-9, 16).
This is the structural basis of the codebook. The continuous realizability is never stated; if the surface has discrete phase resolution, the bit-count guarantee can fail.
standard math The half-power beamwidth of the array factor is HPBW(N) = 1.772/N and the array gain is approximated by N²sinc²(NΔ/2) (Eqs. 11-13, from [14]).
These textbook results are loaded in from [14] (authors' own book) and are asymptotically valid for large N; they underpin the bound in (14)-(15).
domain assumption The control channel is error-free and consumes orthogonal resources; the BS has perfect channel knowledge and computes the configuration before quantization (Section II).
Explicitly assumed in Section II; the paper deliberately excludes channel estimation and control-channel errors from the analysis.
ad hoc to paper For element-wise quantization, the per-element quantization errors are mutually independent (Lemma 1).
Stated in Lemma 1; this independence does not hold for LoS channels where all errors derive from quantizing one common angle parameter, limiting the lemma's use in the comparison.
domain assumption The static path is Rayleigh with hs ~ CN(0, ρIK) and the common rotation ejφ does not change the RIS path gain (Section IV, Eqs. 25-26).
Supports the computation of E||h||² and Lemma 2; the Rayleigh assumption makes the phase of aK^H(φ)hs uniform, enabling the d-bit quantizer analysis.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Control Signaling for Reconfigurable Intelligent Surfaces: How Many Bits are Needed?." pith.science (2026). https://pith.science/paper/WFF5LNOB

@misc{pith2026250603929,
  author       = {Pith},
  title        = {Pith review of: Control Signaling for Reconfigurable Intelligent Surfaces: How Many Bits are Needed?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WFF5LNOB}},
  note         = {Machine review of arXiv:2506.03929}
}

read the original abstract

Reconfigurable intelligent surfaces (RISs) can greatly improve the signal quality of future communication systems by reflecting transmitted signals toward the receiver. However, even when the base station (BS) has perfect channel knowledge and can compute the optimal RIS phase-shift configuration, implementing this configuration requires feedback signaling over a control channel from the BS to the RIS. This feedback must be kept minimal, as it is transmitted wirelessly every time the channel changes. In this paper, we examine how the feedback load, measured in bits, affects the performance of an RIS-aided system. Specifically, we investigate the trade-offs between codebook-based and element-wise feedback schemes, and how these influence the signal-to-noise ratio (SNR). We propose a novel quantization codebook tailored for line-of-sight (LoS) that guarantees a minimal SNR loss using a number of feedback bits that scale logarithmically with the number of RIS elements. We demonstrate the codebook's usefulness over Rician fading channels and how to extend it to handle a non-zero static path. Numerical simulations and analytical analysis are performed to quantify the performance degradation that results from a reduced feedback load, shedding light on how efficiently RIS configurations can be fed back in practical systems.

Figures

Figures reproduced from arXiv: 2506.03929 by the authors.

**Figure 2.** The uplink SNR that is achieved with an RIS with 128 elements under [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** The uplink SNR that is achieved with an RIS with 128 elements under [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

15 extracted references · 13 canonical work pages

[1]

Reconfigurable intelligent surfaces for energy efficiency in wireless communication,

C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in wireless communication,” IEEE Transactions on Wireless Communica- tions, vol. 18, no. 8, pp. 4157–4170, 2019

2019
[2]

Reconfigurable intelligent surfaces: A signal pro- cessing perspective with wireless applications,

E. Bj ¨ornson, H. Wymeersch, B. Matthiesen, P. Popovski, L. Sanguinetti, and E. De Carvalho, “Reconfigurable intelligent surfaces: A signal pro- cessing perspective with wireless applications,” IEEE Signal Processing Magazine, vol. 39, no. 2, pp. 135–158, 2022

work page 2022
[3]

Smart radio environments empowered by reconfigurable intelligent surfaces: How it works, state of research, and the road ahead,

M. Di Renzo, A. Zappone, M. Debbah, M.-S. Alouini, C. Yuen, J. de Rosny, and S. Tretyakov, “Smart radio environments empowered by reconfigurable intelligent surfaces: How it works, state of research, and the road ahead,” IEEE Journal on Selected Areas in Communications , vol. 38, no. 11, pp. 2450–2525, 2020

work page 2020
[4]

Dahlman, S

E. Dahlman, S. Parkvall, and J. Sk ¨old, 5G/5G-advanced: The new generation wireless access technology . Elsevier Science & Technology, 2023

work page 2023
[5]

Multi-Branch Attention Convolutional Neural Network for Online RIS Configuration with Discrete Responses: A Neuroevolution Approach

G. Stamatelis, K. Stylianopoulos, and G. C. Alexandropoulos, “Multi- branch attention convolutional neural network for online RIS configura- tion with discrete responses: A neuroevolution approach,” arXiv preprint arXiv:2409.01765, 2024

work page Pith review arXiv 2024
[6]

On the impact of control signaling in RIS- empowered wireless communications,

F. Saggese, V . Croisfelt, R. Kotaba, K. Stylianopoulos, G. C. Alexan- dropoulos, and P. Popovski, “On the impact of control signaling in RIS- empowered wireless communications,” to appear in IEEE Open Journal of the Communications Society , 2024

work page 2024
[7]

Overhead-aware design of reconfigurable intelligent surfaces in smart radio environments,

A. Zappone, M. Di Renzo, F. Shams, X. Qian, and M. Debbah, “Overhead-aware design of reconfigurable intelligent surfaces in smart radio environments,” IEEE Transactions on Wireless Communications , vol. 20, no. 1, pp. 126–141, 2020

work page 2020
[8]

Control aspects for using RIS in latency-constrained mobile edge computing,

F. Saggese, V . Croisfelt, F. Costanzo, J. Shiraishi, R. Kotaba, P. Di Lorenzo, and P. Popovski, “Control aspects for using RIS in latency-constrained mobile edge computing,” in 57th Asilomar Confer- ence on Signals, Systems, and Computers , 2023, pp. 174–181

work page 2023

Show all 15 references

[9]

Reconfigurable intelligent surfaces control sig- naling,

E. ´A. Miranda Piera, “Reconfigurable intelligent surfaces control sig- naling,” B.S. thesis, Universitat Polit `ecnica de Catalunya, 2024

2024
[10]

Reconfigurable intelligent surfaces for wireless communi- cations: Overview of hardware designs, channel models, and estimation techniques,

M. Jian, G. C. Alexandropoulos, E. Basar, C. Huang, R. Liu, Y . Liu, and C. Yuen, “Reconfigurable intelligent surfaces for wireless communi- cations: Overview of hardware designs, channel models, and estimation techniques,” Intelligent and Converged Networks, vol. 3, no. 1, pp...

2022
[11]

NR; Multiplexing and channel coding (Release 17),

3rd Generation Partnership Project (3GPP), “NR; Multiplexing and channel coding (Release 17),” V17.5.0, 3GPP, Tech. Rep. TS 38.212, 2022, available at: https://www.3gpp.org/ftp/Specs/archive/

2022
[12]

Capacity characterization for intelligent reflect- ing surface aided MIMO communication,

S. Zhang and R. Zhang, “Capacity characterization for intelligent reflect- ing surface aided MIMO communication,” IEEE Journal on Selected Areas in Communications , vol. 38, no. 8, pp. 1823–1838, 2020

2020
[13]

RIS-aided wireless communications: Prototyping, adaptive beamforming, and indoor/outdoor field trials,

X. Pei, H. Yin, L. Tan, L. Cao, Z. Li, K. Wang, K. Zhang, and E. Bj¨ornson, “RIS-aided wireless communications: Prototyping, adaptive beamforming, and indoor/outdoor field trials,” IEEE Transactions on Communications, vol. 69, no. 12, pp. 8627–8640, 2021

2021
[14]

Introduction to multiple antenna com- munications and reconfigurable surfaces,

E. Bj ¨ornson and ¨O. T. Demir, “Introduction to multiple antenna com- munications and reconfigurable surfaces,” Now Publishers, Inc. , 2024

2024
[15]

Spatial channel model for MIMO simulations,

3rd Generation Partnership Project (3GPP), “Spatial channel model for MIMO simulations,” V14.0.0, 3GPP, Tech. Rep. TS 25.996, 2017, available at: https://www.3gpp.org/ftp/Specs/archive/

2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Reconfigurable intelligent surfaces for energy efficiency in wireless communication,

C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in wireless communication,” IEEE Transactions on Wireless Communica- tions, vol. 18, no. 8, pp. 4157–4170, 2019

2019

[2] [2]

Reconfigurable intelligent surfaces: A signal pro- cessing perspective with wireless applications,

E. Bj ¨ornson, H. Wymeersch, B. Matthiesen, P. Popovski, L. Sanguinetti, and E. De Carvalho, “Reconfigurable intelligent surfaces: A signal pro- cessing perspective with wireless applications,” IEEE Signal Processing Magazine, vol. 39, no. 2, pp. 135–158, 2022

work page 2022

[3] [3]

Smart radio environments empowered by reconfigurable intelligent surfaces: How it works, state of research, and the road ahead,

M. Di Renzo, A. Zappone, M. Debbah, M.-S. Alouini, C. Yuen, J. de Rosny, and S. Tretyakov, “Smart radio environments empowered by reconfigurable intelligent surfaces: How it works, state of research, and the road ahead,” IEEE Journal on Selected Areas in Communications , vol. 38, no. 11, pp. 2450–2525, 2020

work page 2020

[4] [4]

Dahlman, S

E. Dahlman, S. Parkvall, and J. Sk ¨old, 5G/5G-advanced: The new generation wireless access technology . Elsevier Science & Technology, 2023

work page 2023

[5] [5]

Multi-Branch Attention Convolutional Neural Network for Online RIS Configuration with Discrete Responses: A Neuroevolution Approach

G. Stamatelis, K. Stylianopoulos, and G. C. Alexandropoulos, “Multi- branch attention convolutional neural network for online RIS configura- tion with discrete responses: A neuroevolution approach,” arXiv preprint arXiv:2409.01765, 2024

work page Pith review arXiv 2024

[6] [6]

On the impact of control signaling in RIS- empowered wireless communications,

F. Saggese, V . Croisfelt, R. Kotaba, K. Stylianopoulos, G. C. Alexan- dropoulos, and P. Popovski, “On the impact of control signaling in RIS- empowered wireless communications,” to appear in IEEE Open Journal of the Communications Society , 2024

work page 2024

[7] [7]

Overhead-aware design of reconfigurable intelligent surfaces in smart radio environments,

A. Zappone, M. Di Renzo, F. Shams, X. Qian, and M. Debbah, “Overhead-aware design of reconfigurable intelligent surfaces in smart radio environments,” IEEE Transactions on Wireless Communications , vol. 20, no. 1, pp. 126–141, 2020

work page 2020

[8] [8]

Control aspects for using RIS in latency-constrained mobile edge computing,

F. Saggese, V . Croisfelt, F. Costanzo, J. Shiraishi, R. Kotaba, P. Di Lorenzo, and P. Popovski, “Control aspects for using RIS in latency-constrained mobile edge computing,” in 57th Asilomar Confer- ence on Signals, Systems, and Computers , 2023, pp. 174–181

work page 2023