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

• 77 exabytes of monthly mobile traffic were expected by 2022, motivating migration beyond sub-6 GHz bands toward millimeter- and sub-millimeter-wave frequencies.Source: A. Possible Migration to… · p. 1
• Millimeter- and sub-millimeter-wave signals are highly susceptible to blockage by large structures and severe attenuation from human bodies and foliage.Source: A. Possible Migration to… · p. 1
• Relays can create alternative propagation routes, but require dedicated power, reception, processing, and retransmission circuitry.Source: B. Relay-Aided Transmission · p. 1
• Half-duplex relaying reduces resource availability, while full-duplex relaying introduces self-interference, destination interference, complexity, and power consumption.Source: B. Relay-Aided Transmission · p. 1
• RISs are configurable after deployment and can support programmable smart radio environments, whereas non-reconfigurable reflectors cannot dynamically shape impinging waves.Source: C. Passive Non-Reconfigurable Reflectors · p. 1, D. Nearly-Passive Smart Surfaces · pp. 1–2
• The article introduces RIS applications at high frequencies and focuses on differences between RIS anomalous reflectors and relay-aided systems.Source: D. Nearly-Passive Smart Surfaces · p. 2, Figure 1 · p. 2

• An RIS is an artificial electromagnetic surface that customizes the propagation of incident radio waves.Source: A. What is an RIS? · p. 2
• Meta-surfaces are thin, sub-wavelength sheets made from sub-wavelength metallic or dielectric scattering particles called meta-atoms or unit-cells.Source: A. What is an RIS? · p. 2
• Non-reconfigurable meta-surfaces have fixed structural arrangements, whereas reconfigurable meta-surfaces can be modified and programmed using external stimuli.Source: B. Reconfigurable Meta-Surfaces · p. 2
• Electronic phase-changing components control switches or tunable elements, allowing the scattered wavefront to be manipulated and optimized.Source: B. Reconfigurable Meta-Surfaces · p. 2
• Reconfigurable surfaces are nearly passive: control circuitry consumes energy, but configured signal transmission requires no dedicated power supply.Source: B. Reconfigurable Meta-Surfaces · p. 2

• RIS applications include anomalous reflection or transmission, beamforming or focusing, joint transmitter/RIS encoding, and single-RF multi-stream transmission.Source: C. Uses of RISs in Wireless… · p. 2

Anomalous reflection/transmission

• Anomalous reflection steers waves toward specified directions independently of fading channels and receiver locations, but generally does not maximize signal-to-noise ratio.Source: C. Uses of RISs in Wireless… · p. 2

Beamforming/focusing

• Beamforming or focusing maximizes signal-to-noise ratio at selected locations, but its optimization depends on fading channels and receiver locations.Source: C. Uses of RISs in Wireless… · p. 2

Joint transmitter/RIS encoding

• Joint transmitter/RIS encoding uses meta-atom states to modulate additional data while requiring joint optimization of the transmitter and RIS.Source: C. Uses of RISs in Wireless… · p. 2

Single-RF multi-stream transmitter design

• A feeder-based RIS can reflect multiple data-modulated signals and realize multi-stream transmission with a limited number of RF chains.Source: C. Uses of RISs in Wireless… · p. 2

Scattering engineering

• RISs can increase the rank of multiple-antenna wireless channels by creating a rich-scattering environment, a function called scattering engineering.Source: C. Uses of RISs in Wireless… · pp. 2–3

• The paper primarily studies anomalous-reflector RISs as fundamental elements for manipulating waves around environmental objects and realizing smart radio environments.Source: C. Uses of RISs in Wireless… · p. 3

• Conventional communication design adapts transmitters, receivers, and protocols to a wireless environment modeled as uncontrollable.Source: A. From Adaptation to Control and… · p. 3
• RISs can shape wavefronts across the network, making the environment an optimizable parameter for rate, latency, reliability, energy efficiency, privacy, and connectivity.Source: A. From Adaptation to Control and… · p. 3

Signal engineering

• In signal engineering, an RIS redirects a blocked line-of-sight beam toward a mobile terminal to maximize received signal strength.Source: B. Illustrative Example of Smart Radio… · p. 3

Interference engineering

• In interference engineering, an RIS shapes an interfering wave so it destructively combines with another signal at the target terminal.Source: B. Illustrative Example of Smart Radio… · p. 3

Security engineering

• In security engineering, an RIS steers reflections toward the intended terminal and away from a malicious user while supporting diversity combining.Source: B. Illustrative Example of Smart Radio… · p. 3

Scattering engineering

• In scattering engineering, an RIS creates a high-rank channel from a low-scattering environment to support high-data-rate multiple-input multiple-output transmission.Source: B. Illustrative Example of Smart Radio… · p. 3

• Across these scenarios, RISs can engineer and optimize radio-wave propagation at low complexity in ways that benefit the network.Source: B. Illustrative Example of Smart Radio… · p. 3

• RISs and relays are compared qualitatively here, with quantitative performance comparisons presented in the following section.Source: RECONFIGURABLE INTELLIGENT SURFACES… · p. 3
• Relays use powered active components such as DACs, ADCs, mixers, power amplifiers, and low-noise amplifiers for reception and transmission.These components support decode-and-forward and amplify-and-forward relaying.Source: A. Hardware Complexity · p. 3
• RISs use metallic or dielectric patches with low-power switches or varactors, typically avoiding dedicated power amplifiers, mixers, and DACs/ADCs.The paper envisions lower complexity, particularly with inexpensive large-area electronics.Source: A. Hardware Complexity · p. 4

Noise

• Relay active components introduce additive noise; amplify-and-forward relays amplify noise, while decode-and-forward reduces it through regeneration at higher complexity and power consumption.Full-duplex relaying also suffers from residual loop-back self-interference.Source: B. Noise · p. 4
• RIS anomalous reflectors avoid additive noise but may experience phase noise and, when nearly passive, cannot amplify or regenerate signals.Source: B. Noise · p. 4

• RIS anomalous reflectors avoid half-duplex rate loss and loop-back self-interference, and their reflection coefficients can combine transmitter- and RIS-received signals.Source: C. Spectral Efficiency · p. 4

Power requirements

• Relays require an independent power source for signal transmission and electronic components.Source: D. Power Budget · p. 4
• RISs can use passive components with low-power switches or varactors, making energy harvesting suitable for close-to-passive implementations.Source: D. Power Budget · p. 4

RF power allocation

• RISs place the total RF power at the transmitter, while reflected power depends on an optimizable surface transmittance and equals impinging power ideally.Source: D. Power Budget · p. 4

Scaling with N

• With maximum ratio weighting and N relay antennas, average end-to-end signal-to-noise ratio increases linearly with N.Source: E. Average Signal-to-Noise Ratio vs.… · p. 4
• An RIS with N individually tunable antennas has average end-to-end signal-to-noise ratio increasing quadratically with N.The paper indicates N may reach a few thousands for inexpensive tunable antennas and ten thousands for meta-surfaces.Source: E. Average Signal-to-Noise Ratio vs.… · p. 4

Geometry and assumptions

• The analysis assumes a two-dimensional source-relay/RIS-receiver geometry with equal-distance notation based on dSR and dRD.A relay is at the origin, and a one-dimensional RIS of length 2L is centered there.Source: F. Average Signal-to-Noise Ratio vs.… · p. 4

Relay scaling

• AF-relay end-to-end received power scales with the reciprocal of the product of transmitter-to-relay and relay-to-receiver distances.Source: F. Average Signal-to-Noise Ratio vs.… · p. 4
• With noise, DF and AF relay signal-to-noise ratio scales with the reciprocal of the weaker of the two paths.Source: F. Average Signal-to-Noise Ratio vs.… · pp. 4–5

RIS regimes

• RIS distance scaling depends on its geometric size, wavelength, and relative transmitter-to-RIS and RIS-to-receiver distances.The paper analyzes electrically large and electrically small regimes.Source: F. Average Signal-to-Noise Ratio vs.… · p. 5

Electrically large RISs

• Electrically large RISs behave asymptotically as anomalous mirrors when their size is large relative to wavelength and transmission distances.Received power and average end-to-end signal-to-noise ratio scale as (αkdSR + βkdRD)^-1.Source: F. Average Signal-to-Noise Ratio vs.… · p. 5

Electrically small RISs

• Electrically small RISs behave asymptotically as diffusers and scale as 4L^2(dSRdRD)^-1, matching AF-relay received-power scaling.Their end-to-end signal-to-noise ratio also depends on RIS length 2L.Source: F. Average Signal-to-Noise Ratio vs.… · p. 5

Prototype scale

• A 1 m^2 RIS prototype operating at 10.5 GHz was reported to operate in the far field beyond approximately 70 m analytically and 28 m experimentally.Source: F. Average Signal-to-Noise Ratio vs.… · p. 5

Comparison setup

• The comparison assumes equal transmitter-to-device and device-to-receiver distances, with RIS length determined by Mma meta-atoms spaced at λ/D.Source: G. Takeaway Messages from the Comparison · p. 5

Short distances

• Relay-aided transmission and electrically large RISs offer similar distance scaling for short distances d0.Source: G. Takeaway Messages from the Comparison · p. 5
• RISs can potentially provide a better rate than relays at fixed RIS size when transmission distances are not too long.The stated comparison concerns anomalous-reflector RISs without half-duplex and loop-back self-interference constraints.Source: G. Takeaway Messages from the Comparison · p. 5

Long distances

• Electrically small RISs have less favorable distance scaling than relays for long distances d0.Source: G. Takeaway Messages from the Comparison · p. 5
• A sufficiently large electrically small RIS can still potentially outperform relay-aided transmission because its average end-to-end signal-to-noise ratio scales quadratically with size.The paper concludes that RIS-aided transmission may outperform relaying when RIS size is sufficiently large.Source: G. Takeaway Messages from the Comparison · p. 5

Evaluation setup

• The study considers one relay and one RIS, located equidistantly from the transmitter and receiver.DF relaying is used for the relay results, while ideal full-duplex performance is also reported.Source: NUMERICAL RESULTS · pp. 5–6

Evaluated comparisons

• Fig. 4 compares RIS and relay data rate as a function of transmission frequency.Source: Figure 4 · p. 6
• Fig. 5 compares RIS and relay data rate as a function of RIS size.Source: Figure 5 · p. 6

Evaluation setup

• A total power constraint splits the available power equally between the transmitter and the relay.RIS fields are computed using the analytical frameworks reported in the study.Source: NUMERICAL RESULTS · p. 6

Distance dependence

• At 28 GHz with a 2L = 1.5 m RIS, anomalous-reflector operation provides a rate similar to an ideal full-duplex relay without a power amplifier.The RIS length corresponds to 140λ; the result can require approximately 280–700 meta-atoms for spacings between λ/2 and λ/5.Source: A. RISs vs. Relays as a Function of… · p. 6
• The RIS behaves as an anomalous mirror for distances d0 up to approximately 25–50 m and as a diffuse scatterer beyond that range.Source: A. RISs vs. Relays as a Function of… · p. 6

Anomalous mirrors vs. focusing lenses

• A focusing-lens RIS generally outperforms an anomalous-reflector RIS and can achieve rates similar to the reflector’s long-distance approximation.The focusing lens requires estimating transmitter and receiver locations and adapting RIS phases to the wireless channels.Source: B. RISs: Anomalous Mirrors vs.… · p. 6

RIS-size dependence

• At d0 = 100 m, an RIS shorter than approximately L = 0.5–0.75 m is insufficient to provide rates similar to an ideal full-duplex relay at 28 GHz.Source: D. RISs vs. Relays as a Function of… · p. 7

Observed behavior

• Short-distance results exhibit oscillations caused by coherent summation of secondary waves reflected by the RIS with different phases.Source: D. RISs vs. Relays as a Function of… · p. 7

Physics-Based Modeling

• Current RIS models are simplified, motivating accurate but analytically tractable physics-based models grounded in electromagnetism.Existing local-structure models ignore spatial coupling among meta-atoms.Source: THE ROAD AHEAD · p. 7

Experimental Validation

• Equivalent RIS models require validation through hardware testbeds and empirical measurements, but only a few experiments have validated received-power scaling laws.The relevant variables include RIS size, transmission distance, and the wave transformation applied.Source: THE ROAD AHEAD · p. 7

Constrained System Design

• Nearly passive RIS implementation imposes stringent constraints on signal-processing algorithms and communication protocols.Without power amplifiers and channel-estimation units, RISs cannot perform channel estimation, creating a need for low-overhead acquisition protocols.Source: THE ROAD AHEAD · p. 7

Information and Communication Theory

• RIS-based smart radio environments challenge information-theoretic models that treat channel transition probabilities as fixed and non-optimizable.Including channel states among encoding and modulation degrees of freedom opens new system-optimization directions.Source: THE ROAD AHEAD · p. 7

Conclusions

• The paper concludes that sufficiently large anomalous-reflector RISs can outperform relay-aided systems in data rate while reducing implementation complexity.Source: CONCLUSIONS · p. 7