IRS Physical Architectures Overview

Updated 12 March 2026

IRS physical architectures are engineered frameworks that use programmable metasurfaces to control electromagnetic waves via per-element phase shifts and amplitude modulation.
They are classified into fully-passive, semi-passive, and active designs, each with distinct hardware features, power requirements, and system-level implications.
IRS architectures support advanced applications such as integrated sensing and communication, multi-IRS deployments, and significant beamforming gains for enhanced spatial multiplexing.

Intelligent Reflecting Surface (IRS) physical architectures are engineered frameworks that leverage reconfigurable metasurfaces—planar arrays of subwavelength elements—capable of manipulating incident electromagnetic waves via programmable per-element phase shifts and, in some cases, amplitude modulation. These architectures underpin next-generation wireless systems, including integrated sensing and communication (ISAC) networks and reconfigurable radio environments, by enabling adaptive control of wireless propagation through compact, energy-efficient hardware. IRS physical architectures encompass a wide design space, including fully-passive, semi-passive, and active realizations, as well as topological variants such as single-panel, cascaded (multi-IRS), and hybrid deployments integrating additional functionalities such as receive sensors or local amplifiers. Architectures are typically categorized by degree of element actuation, channel configuration, array geometry, hardware complexity, and intended system-level functionality (Song et al., 2024).

1. Taxonomy of IRS Physical Architectures

IRS architectures can be rigorously classified based on the actuation mechanism of the reflecting elements, the presence of on-site signal processing, and topological deployment (Song et al., 2024, Wu et al., 15 Jan 2025).

1. Fully-Passive IRS:

Consists solely of phase-tunable, unit-amplitude reflecting elements, realized as 2D metasurfaces patterned with subwavelength meta-atoms (e.g., patches, dipoles, split-ring resonators). Phase modulation is achieved via integrated tunable components such as PIN diodes, varactor diodes, or MEMS switches, controlled by a low-power processor. No RF chains or amplifiers are present.

2. Semi-Passive IRS:

Enhances the passive panel with a small number of co-located receive RF chains or sensors, enabling direct sampling of echoes or environmental signals. The overall structure remains dominated by passive elements, but the local sensing chain facilitates more efficient target parameter estimation in ISAC settings (Chen et al., 27 Jun 2025, Song et al., 2024).

3. Active IRS:

Integrates low-noise amplifiers (LNAs) at each reflecting element, supporting both phase and amplitude modulation ( $a_n e^{j\phi_n}$ , $a_n > 1$ ). This confers the ability to overcome severe path losses but incurs additional design complexity, introduces amplifier noise, and requires distributed power supply (Song et al., 2024, Wu et al., 15 Jan 2025).

4. Multi-IRS Topologies:

Architectural variants involving multiple IRS panels—either in cascaded (e.g., double-hop, multi-hop) or parallel (e.g., distributed, multi-region) configurations—expand the spatial degrees of freedom (DoF), enable fine-grained spatial multiplexing, and can facilitate sophisticated beam routing or cooperative sensing (Chen et al., 27 Jun 2025, Han et al., 2020, Song et al., 2024).

Architecture	Element Actuation	On-site Sensing	Per-element Gain	Power Model
Fully-Passive	Phase only	No	$a_n = 1$	Control only
Semi-Passive	Phase + sensors	Yes	$a_n = 1$	Sensors+control
Active	Phase + amplitude	Yes	$a_n \geq 1$	High (amps)
Multi-IRS	Any	Possible	Topology-specific	Aggregated

2. Core Hardware Realizations and Geometric Configurations

An IRS panel comprises four core modules: (a) metasurface array, (b) per-element phase-shifter network, (c) control processor, and (d) optional RF-feed/calibration circuitry (Wu et al., 17 Jun 2025). The metasurface is typically a thin PCB or film with patterned meta-atoms atop a control layer. Each element is spatially spaced at $d \sim \lambda/2$ to avoid grating lobes. The control layer contains phase-shifter ICs (PIN/varactor/MEMS), digital control lines (SPI/I $^2$ C/parallel), power regulation, and, when present, receive sensor interfaces or amplifier biasing (Wu et al., 15 Jan 2025, Wu et al., 17 Jun 2025).

IRS panels are further categorized by geometric configuration:

Planar (rectangular, square) arrays: Most common, scalable via tiling.
Modular tiles: Deployable as 0.5 m $\times$ 0.5 m units for large-area coverage.
Inkjet-printed, paper-based IRS: Ultra-low-cost, mass-producible, and easily tileable for indoor use (Takimoto et al., 2024).

Example: The printer-friendly IRS by Takimoto et al. realizes a 10-element supercell using inkjet-deposited silver nanoparticle patches, with variable patch lengths encoding a discrete linear phase ramp (step $\Delta \Phi = 2\pi/10$ ), on a paper substrate with MDF spacer and copper ground plane (Takimoto et al., 2024).

Layer	Function
Metasurface	Wave manipulation (phase/amplitude)
Control circuitry	Phase tuning, calibration, power supply
Mounting framework	Orientation, weather-sealing, support

Multi-IRS deployments enforce precise geometric constraints. For double-IRS, optimal placement aligns one IRS near transmitter, one near receiver, with far-line-of-sight separation such that the inter-IRS channel is low-rank (preferably rank-1), maximizing joint beamforming (Han et al., 2020, Chen et al., 27 Jun 2025).

3. Channel Modeling and System-Level Equations

The electromagnetic behavior and system-level performance of IRS architectures are described by composite channel models integrating per-element coefficients, array responses, and network topology.

Single-IRS End-to-End Channel:

For a BS-IRS-user link with phase-shift matrix $\Theta = \mathrm{diag}(e^{j\theta_n})$ ,

$a_n > 1$ 0

with mean received power

$a_n > 1$ 1

under ideal phase coherence (Wu et al., 15 Jan 2025, Song et al., 2024).

Double-IRS Channel:

For two chained IRSs (sizes $a_n > 1$ 2, $a_n > 1$ 3), channel is

$a_n > 1$ 4

where $a_n > 1$ 5 models the inter-IRS channel (rank-1 under LoS), yielding received power $a_n > 1$ 6; with $a_n > 1$ 7, total gain is $a_n > 1$ 8, compared to $a_n > 1$ 9 for a single IRS (Han et al., 2020, Wu et al., 15 Jan 2025).

Hybrid Multi-IRS ISAC Model:

Incorporates $a_n = 1$ 0 IRSs (one semi-passive, $a_n = 1$ 1 passive). Communication:

$a_n = 1$ 2

Sensing (echo at IRS 1 sensors, after clutter cancellation): $a_n = 1$ 3 Degrees of spatial multiplexing $a_n = 1$ 4 under angular orthogonality. Sensing accuracy is quantified by the Cramér–Rao bound (CRB) for angle estimation, which increases with $a_n = 1$ 5 for fixed element budget $a_n = 1$ 6 (Chen et al., 27 Jun 2025).

Active IRS Model (per-element gain):

$a_n = 1$ 7

with SNR degradation due to amplifier noise: $a_n = 1$ 8 (Song et al., 2024, Wu et al., 15 Jan 2025).

4. Performance Metrics, Scaling Laws, and Tradeoffs

IRS system performance is governed by tradeoffs between array size, multiplicity, element allocation, power constraints, and functional requirements (e.g., communication vs. sensing).

Beamforming Gain: For a passive $a_n = 1$ 9-element IRS, coherent beamforming achieves $a_n = 1$ 0 power gain per surface. With cascaded IRSs, gains multiply: total $a_n = 1$ 1 for $a_n = 1$ 2 IRSs in a chain (Han et al., 2020, Wu et al., 15 Jan 2025).
Spatial Multiplexing DoF: In multi-IRS deployments, spatial DoF scale as the number of orthogonal IRS paths, up to $a_n = 1$ 3 (Chen et al., 27 Jun 2025).
Sensing-Communication Tradeoff: In hybrid multi-IRS ISAC, increasing number of IRSs ( $a_n = 1$ 4) increases communication DoF linearly but raises the angle estimation CRB cubically, under fixed total element constraint $a_n = 1$ 5 (Chen et al., 27 Jun 2025). Communication-optimal designs become sufficient for sensing only above a calculable threshold in total elements.

Design Parameter	Communication	Sensing (CRB)	Scaling Law (Chen et al., 27 Jun 2025)
Number of IRSs $a_n = 1$ 6	$a_n = 1$ 7	$a_n = 1$ 8	$a_n = 1$ 9; $a_n \geq 1$ 0
Element Split $a_n \geq 1$ 1	Even split: max DoF	Concentration boosts
		beamforming gain

For double-IRS (LoS, balanced element split), the end-to-end gain scales as $a_n \geq 1$ 2, while for single IRS it scales as $a_n \geq 1$ 3 (Han et al., 2020). This multiplicative array gain is only realized when inter-IRS channels remain high-rank or ideally rank-1; otherwise, performance degrades.

Active vs Passive: Active IRS can compensate for large path loss via per-element gain $a_n \geq 1$ 4, but adds amplifier noise ( $a_n \geq 1$ 5), requiring higher power and advanced thermal management (Song et al., 2024, Wu et al., 15 Jan 2025).

5. Physical Deployment Guidelines and Implementation Considerations

Optimal IRS physical architecture design requires joint consideration of electromagnetic layout, system topology, hardware capabilities, and environmental compatibility (Wu et al., 15 Jan 2025, Wu et al., 17 Jun 2025).

Element Spacing: $a_n \geq 1$ 6; larger spacings risk grating lobes.
Aperture Size: $a_n \geq 1$ 7; at mmWave, $a_n \geq 1$ 8 fits on $a_n \geq 1$ 90.8 m $d \sim \lambda/2$ 0 0.8 m panel.
Phase Resolution: Typical $d \sim \lambda/2$ 1 bits; higher $d \sim \lambda/2$ 2 reduces quantization loss at increased complexity.
Control: Microcontroller, FPGA, or ASIC; communicates with BS via low-latency backhaul.
Power: Passive IRSs require $d \sim \lambda/2$ 3 for control; active IRSs, 1–10 W/100 elements for amplifiers.
Deployment Strategies:
- Centralized (single large panel): Maximal gain for sparse multipath, optimal for point-to-point.
- Distributed (multiple panels): Enhances multiplexing; equal element split $d \sim \lambda/2$ 4 provides max DoF.
- Hybrid/movable: Mechanically steerable arrays (actuator, robotic, drone-mounted) provide dynamic coverage (Wu et al., 17 Jun 2025).
- Height/orientation: 3–15 m above ground, panel normal should bisect TX–IRS–RX angles for maximal gain (Wu et al., 15 Jan 2025).

Field trials confirm up to $d \sim \lambda/2$ 5 dB RSRP gain and severalfold increases in spectral and energy efficiency with both single and double IRS architectures (Wu et al., 15 Jan 2025).

6. Fabrication, Materials, and Cost Factors

Physical realization ranges from low-cost, mass-fabricated panels to advanced, actively-cooled, sensor-integrated modules.

Inkjet-Printed IRS: Paper-based metasurfaces using silver nanoparticle ink on ordinary office paper with MDF spacer and copper ground plane. Efficient phase gradients achieved by variable patch length supercells; provides practical demonstration of mass-production and deployment in cost-sensitive indoor scenarios (Takimoto et al., 2024).
PCB–Based IRS: Rigid or flexible PCB, etched meta-atom patterns, with integrated control circuitry and biasing.
Thermal Management: Critical for active IRS; passive designs require little beyond environmental sealing (Wu et al., 17 Jun 2025).
Calibration: Built-in pilot transmitters/receivers, over-the-air phase response measurement circuits, and control software for real-time correction (Wu et al., 17 Jun 2025).
Scalability: Design is fully scalable by tiling; hybrid, modular panels allow coverage of arbitrarily large domains.

Material cost for paper-based IRS is %%%%49 $a_n > 1$ 50%%%%240 \times 360 $mm), with full assembly in under 30 minutes without clean-room processing (<a href="/papers/2401.07276" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Takimoto et al., 2024</a>).</p> <h2 class='paper-heading' id='challenges-tradeoffs-and-emerging-directions'>7. Challenges, Tradeoffs, and Emerging Directions</h2> <p>Key challenges in IRS physical architectures include:</p> <ul> <li><strong>Robustness to Environmental Stressors:</strong> Weather, wind, and obstruction dynamics; passive IRSs generally more robust than active/amplified versions (<a href="/papers/2501.08576" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Wu et al., 15 Jan 2025</a>).</li> <li><strong><a href="https://www.emergentmind.com/topics/critical-success-index-csi" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">CSI</a> Acquisition and Control Latency:</strong> Overhead and latency increase with$ d \sim \lambda/2$8; low-latency digital interface and real-time calibration required for large arrays (Wu et al., 15 Jan 2025, Wu et al., 17 Jun 2025).

Element Coupling and Mutual Impedance: Particularly acute in tightly-packed or large-aperture panels; requires calibration and careful meta-atom design (Wu et al., 17 Jun 2025).

Hybrid and Movable Platforms: Integration of movable antennas expands spatial DoF, but requires joint mechanical-electronic optimization and may introduce maintenance complexity (Wu et al., 17 Jun 2025).

Scalable Networking: Networked IRSs (multi-view sensing, region-wise ISAC, beam routing) require graph-theoretic path selection, distributed synchronization, and possibly in-band cooperation (Chen et al., 27 Jun 2025, Song et al., 2024).

Active areas of research include near-field large-aperture design, AI-driven deployment optimization, robust architectures for hardware and environmental impairments, and hybrid passive-active panel arrangements (Wu et al., 15 Jan 2025, Wu et al., 17 Jun 2025).

References:

(Song et al., 2024) An Overview on IRS-Enabled Sensing and Communications for 6G (Chen et al., 27 Jun 2025) Multi-IRS Aided ISAC System: Multi-Path Exploitation Versus Reduction (Han et al., 2020) Cooperative Double-IRS Aided Communication: Beamforming Design and Power Scaling (Wu et al., 15 Jan 2025) Intelligent Reflecting Surfaces for Wireless Networks: Deployment Architectures, Key Solutions, and Field Trials (Wu et al., 17 Jun 2025) Integrating Movable Antennas and Intelligent Reflecting Surfaces (MA-IRS): Fundamentals, Practical Solutions, and Opportunities (Takimoto et al., 2024) Inkjet printed intelligent reflecting surface (IRS) for indoor applications