PaperlayerView exact revision on arXiv

Source-linked AI summary

Smart and Reconfigurable Wireless Communications: From IRS Modeling to Algorithm Design

Xianghao Yu, Vahid Jamali, Dongfang Xu, Derrick Wing Kwan Ng, Robert Schober

arXiv:2103.07046v1cs.ITeess.SPmath.IT

TL;DR

IRS system design lacks a well-balanced modeling basis, even as IRSs offer a way to customize radio propagation. The article compares three IRS models, proposes a tiled codebook framework, and studies secure communication and SWIPT designs. Its case studies report improved secure-system sum-rate over simple-transmission and no-IRS baselines, and reduced SWIPT transmit power over random-phase and no-IRS baselines.

  • Problem

    IRS design and evaluation lack models that balance accuracy, tractability, and hardware complexity, despite model choice affecting applicable system-design methods.

    Source: §I · pp. 2–3

  • Method

    The article compares three IRS models, introduces a tile-and-transmission-mode-codebook framework for scalable optimization, and studies secure communication and SWIPT systems.

    Source: §I · pp. 2–3, §IV · p. 12

  • Results

    Secure-system sum-rate improves over simple-transmission and no-IRS baselines, while SWIPT transmit power falls versus random-phase and no-IRS baselines.

    Source: A. Secure Wireless Communications via… · p. 12, B. IRS-assisted SWIPT Systems · pp. 13–14

  • Takeaways & Limitations

    The tile framework lets designers adjust tile count and mode-set size to trade transmit-power performance against computational complexity in large-scale IRS optimization.

    Source: B. IRS-assisted SWIPT Systems · p. 14, §V · p. 15

    Design based on long-term statistical CSI and mathematical optimization of IRS positions remain open problems.

    Source: §V · pp. 15–16

Abstract

from arXiv · show

Intelligent reflecting surfaces (IRSs) have been introduced into wireless communications systems due to their great potential to smartly customize and reconfigure radio propagation environments in a cost-effective manner. Despite the promising advantages of IRSs, academic research on IRSs is still in its infancy. In particular, the design and analysis of IRS-assisted wireless communication systems critically depend on an accurate and tractable modeling of the IRS. In this article, we first present and compare three IRS models, namely the conventional independent diffusive scatterer-based model, physics-based model, and impedance network-based model, in terms of their accuracy, tractability, and hardware complexity. Besides, a new framework based on partitioning the IRS into tiles and employing codebooks of transmission modes is introduced to enable scalable IRS optimization. Then, we investigate the impact of the three considered IRS models on system design, where several crucial technical challenges for the efficient design of IRS-assisted wireless systems are identified and the corresponding solutions are unraveled. Furthermore, to illustrate the properties of the considered models and the efficiency of the proposed solution concepts, IRS-assisted secure wireless systems and simultaneous wireless information and power transfer (SWIPT) systems are studied in more detail. Finally, several promising future research directions for IRS-assisted wireless systems are highlighted.

I. INTRODUCTION

IRSs offer programmable, low-power radio environments, but their design requires models that balance accuracy, tractability, and hardware complexity. The article compares IRS models and scalable design approaches, including the conventional IDS baseline and its limits.

I. INTRODUCTION

  • IRSs can customize propagation at low power and cost, but their configuration and system impacts still require accurate characterization and are not yet satisfactorily addressed.IRS elements are passive and typically consume less than 1 mW each.Source: §I · p. 2
  • The article compares three IRS models and their design implications, introduces a scalable optimization framework, and examines secure communications and SWIPT applications.It also identifies design challenges, discusses potential solutions, and highlights future research directions.Source: §I · pp. 2–3, §II · p. 3

A. Independent Diffusive Scatterer-based (IDS) Model

  • The IDS baseline models each reflecting element as an independent phase-shifting scatterer with unit-modulus reflection coefficients, but omits practical physical properties and complicates large-IRS resource allocation.The omitted properties include element size, polarization, inter-element connectivity, and wave arrival and departure angles.Source: A. Independent Diffusive… · p. 3

B. Physics-based (PHY) Model

The PHY model uses physical propagation properties to describe IRS reflection, building on recent EM-compliant modeling from a communication-theoretical perspective.

B. Physics-based (PHY) Model

  • Communication-theoretical work on EM-compliant IRS models has only recently drawn attention, following a rich physics and electromagnetics modeling history,.Earlier approaches modeled IRS discontinuities with effective surface currents and Maxwell’s equations [8], or as electrically and magnetically polarizable reflecting-element arrays [10].Source: B. Physics-based (PHY) Model · p. 4
  • Because wireless channels typically have limited scatterers, especially when the direct link is blocked, directing reflected waves toward strong channel paths is crucial for IRS performance.Source: B. Physics-based (PHY) Model · p. 4
  • Under a far-field assumption, the generalized radar cross section (GRCS) models how an incident plane wave is reflected toward an intended direction for a given IRS phase configuration.The GRCS depends on the incident and reflected directions and accounts for element size and spacing, physical properties omitted by the IDS model.Source: B. Physics-based (PHY) Model · p. 4

C. Impedance Network-based (INW) Model

The impedance network-based (INW) model represents the IRS as a multi-port network with a general scattering matrix. Its additional degrees of freedom allow it to adjust both phase and magnitude, and it can be combined with the physics-based model’s GRCS.

C. Impedance Network-based (INW) Model

  • INW models the IRS as a multi-port network characterized by a general scattering matrix Θ, controlled through an effective impedance matrix Z.Reflecting elements may be connected fully or partially through the configurable impedance network.Source: C. Impedance Network-based (INW) Model · pp. 4–5
  • In a single-connected IRS, elements are unconnected, so INW reduces to the baseline IDS model, with Θ = Φ.Its impedance matrix is the same as the one presented in.Source: C. Impedance Network-based (INW) Model · p. 5
  • A fully-connected IRS connects every element to all others, while a partially-connected IRS fully connects elements within groups, yielding a block-diagonal Θ.The fully-connected scattering matrix is complex symmetric and unitary; each partially-connected block has the same properties.Source: C. Impedance Network-based (INW) Model · p. 5

COMPARISON OF DIFFERENT MODELS AND PROPERTIES OF TC FRAMEWORK

  • INW’s additional degrees of freedom let it adjust both the phases and magnitudes of impinging waves, and combining it with the PHY model’s GRCS yields a physically explainable model.The article says this integration can also improve performance compared with the previous two models that do not use impedance networks.Source: COMPARISON OF DIFFERENT MODELS AND… · p. 6

D. Tile and Codebook-based (TC) Framework

The tile and codebook-based (TC) framework scales IRS optimization by grouping reflecting elements into tiles and selecting from predesigned transmission modes. The tile count and codebook size balance scalability against achievable performance.

D. Tile and Codebook-based (TC) Framework

  • The TC framework groups reflecting elements into N tiles and uses a codebook of M predesigned phase-shift configurations, or transmission modes, for each tile.The modes are designed offline and stored in the codebook; a suitable mode is selected online for transmission or channel estimation.Source: D. Tile and Codebook-based (TC)… · p. 6
  • The framework can be applied to IDS, PHY, and INW models, with the cited example combining TC and PHY.The framework was proposed to address the practical infeasibility of optimizing every element and estimating every corresponding channel gain for large IRSs.Source: D. Tile and Codebook-based (TC)… · pp. 6–7
  • Choosing N too large approaches non-TC design, while N = 1 requires a large M for satisfactory performance, so both parameters should balance scalability and performance.When each tile contains one reflecting element, N equals the number of elements and the framework reduces to the conventional non-TC framework.Source: D. Tile and Codebook-based (TC)… · p. 7

III. DESIGN CHALLENGES AND SOLUTIONS

IRS-assisted system design faces coupled optimization variables and model-specific constraints, alongside challenges in resource allocation. The article outlines algorithmic approaches tailored to these constraints.

A. Joint Design of Active and Passive Beamforming

  • Joint active and passive beamforming is non-convex because the IRS beamforming matrix multiplies the transmitter’s active beamforming vectors.Alternating optimization solves disjoint blocks in turn; bilinear transformation treats their product as a new variable and enforces additional constraints.Source: A. Joint Design of Active and Passive… · p. 7

COMPARISON OF DIFFERENT TECHNICAL CHALLENGES AND POTENTIAL ALGORITHMS

  • For IDS, unit-modulus constraints can be handled with manifold optimization or reformulated for inner approximation and successive convex approximation.The reformulation uses a rank-constrained problem expressed as a difference of matrix norm functions.Source: COMPARISON OF DIFFERENT TECHNICAL… · p. 8
  • PHY optimization combines binary reflection-beam selection with unit-modulus phase optimization, while INW’s complex symmetric unitary constraint can be handled with manifold optimization.The PHY approaches include manifold optimization and enumeration-based algorithms such as branch-and-bound.Source: COMPARISON OF DIFFERENT TECHNICAL… · p. 8
  • TC mode selection creates a mixed-integer problem addressable by branch-and-bound, or approximately by quadratic penalty or ADMM methods.Table II summarizes constraints introduced by the models and TC framework alongside available resource-allocation algorithms.Source: COMPARISON OF DIFFERENT TECHNICAL… · pp. 8–9

B. Channel State Information (CSI) Acquisition

Passive IRSs cannot directly emit pilots for channel estimation, making accurate CSI acquisition challenging. The section describes estimation approaches and methods for designing systems robust to CSI errors.

B. Channel State Information (CSI) Acquisition

  • Because passive IRSs lack RF chains, they cannot estimate IRS-assisted channels directly by emitting pilot symbols, so novel CSI acquisition methods are needed.System designs must also account for inevitable CSI estimation error.Source: B. Channel State Information (CSI)… · p. 9
  • For single-antenna receivers, DFT-based passive beamforming is used to acquire cascaded channels in IDS and INW; with multiple antennas, the two channel segments can be estimated alternately.For PHY, sparsity in angular domains and propagation paths can be exploited using compressed-sensing methods.Source: B. Channel State Information (CSI)… · p. 9
  • Under TC, CSI acquisition overhead scales with the numbers of tiles N and transmission modes M, which can be chosen to trade performance against complexity or signaling overhead..Source: B. Channel State Information (CSI)… · p. 9
  • Deterministic CSI errors can be bounded using the S-procedure, while statistical errors lead to chance constraints handled with distribution-based reformulations or Bernstein-type inequalities.Because IRS design variables appear in product form, these techniques may not apply directly; inequalities such as the triangle inequality can decouple products to facilitate convex reformulation.Source: B. Channel State Information (CSI)… · pp. 9–10

C. Hardware Impairments

Hardware impairments arise in both transceiver RF chains and IRS elements, complicating impairment-aware design; the INW model presents an additional unresolved challenge for characterizing IRS phase impairments.

C. Hardware Impairments

  • Transceiver RF-chain impairments are modeled by adding Gaussian distortion noise whose variance is proportional to transmit or received signal power.The article identifies power amplifiers, mixers, analog-to-digital converters, and oscillators as examples of components with non-negligible impairments.Source: C. Hardware Impairments · p. 10
  • IRS impairments can be modeled as finite-resolution phase shifts or as random phase errors, but their effects are not straightforward to determine for the connected elements in the INW model.The article calls for more research to characterize impairments under the INW model.Source: C. Hardware Impairments · p. 10
  • Impairment-aware design is difficult because the beamformer and IRS reflection matrix appear in both the numerator and denominator of the SNR expression.Majorization minimization can use a surrogate for the quotient-form SNR, and the SNR is expected to saturate at exceedingly high transmit power even with optimal design.Source: C. Hardware Impairments · pp. 10–11

D. Multi-IRS Systems and IRS Deployment

Multiple IRSs can fill coverage holes, but deployment must balance path loss against the rank needed for MIMO spatial multiplexing. For system design, distributed IRSs can be treated jointly as a virtual mega IRS.

D. Multi-IRS Systems and IRS Deployment

  • Multiple IRSs are proposed to fill coverage holes, with locations selected once using channel statistics, building distribution, and population density.LoS links can reduce path loss, but a pure LoS channel matrix is generally rank-deficient; physically separated IRSs can ideally provide full-rank MIMO channels with low path loss.Source: D. Multi-IRS Systems and IRS Deployment · p. 11
  • Negligible inter-IRS reflections allow distributed IRSs to be modeled as one virtual mega IRS, with their reflection matrices stacked as one optimization variable.The direct and reflecting channel matrices can also be treated jointly as one effective channel matrix, and the tile-and-codebook framework can reduce design complexity.Source: D. Multi-IRS Systems and IRS Deployment · p. 11

IV. CASE STUDIES

The case studies examine secure multi-IRS transmission under the IDS and INW models and SWIPT design using the PHY model and tile-and-codebook framework. In the secure case, element distribution and IRS model affect the reported sum-rate.

A. Secure Wireless Communications via IRSs

  • The secure-system design uses alternating optimization for IRS phases, transmit beamforming, and artificial-noise covariance, with generalized S-procedure treatment of CSI uncertainty.The IDS unit-modulus constraint is handled by IA, the INW complex symmetric unitary constraint by MO, and objective nonconvexity by SCA.Source: A. Secure Wireless Communications via… · p. 12
  • With ten reflecting elements across two IRSs, uniform allocation (M1 = 5) is preferable for physical-layer security to placing all elements at one IRS (M1 = 0 or M1 = 10).The passage attributes the benefit to multiple independent propagation paths and macro-diversity; Fig. 2 plots average system sum-rate against the number of elements at one IRS.Source: A. Secure Wireless Communications via… · pp. 12–13, Figure 2 · p. 13
  • The INW model achieves higher average system sum-rate than the IDS model in the secure-system case study.Source: A. Secure Wireless Communications via… · p. 13

B. IRS-assisted SWIPT Systems

For SWIPT with a large IRS, the design jointly selects transmit beamforming and tile transmission modes under information-decoding and energy-harvesting quality-of-service requirements. The reported schemes reduce transmit power against the stated baselines, while tile and mode-set sizes trade computational complexity against performance.

B. IRS-assisted SWIPT Systems

  • The design uses the PHY model and tile-and-codebook framework for a 200-phase-element IRS, jointly optimizing transmit beamforming and transmission-mode selection under receiver QoS requirements.Branch-and-bound and successive convex approximation provide optimal and suboptimal solutions, respectively.Source: B. IRS-assisted SWIPT Systems · p. 13
  • The proposed optimal and suboptimal SWIPT schemes significantly reduce average total transmit power versus random-phase and no-IRS baselines.Fig. 3 plots average total transmit power against the minimum required SINR of information-decoding receivers.Source: B. IRS-assisted SWIPT Systems · pp. 13–14, Figure 3 · p. 14
  • Increasing the number of tiles N and transmission modes M can reduce required transmit power, but raises computational complexity.The optimization complexity scales with N and M, and the optimal and suboptimal schemes have a small performance gap.Source: B. IRS-assisted SWIPT Systems · p. 14

V. CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS

The conclusion distinguishes the three IRS models and presents the TC framework as a way to facilitate large-system design. It also identifies open directions involving high-frequency systems, statistical CSI, and AI-enabled design.

V. CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS

  • The IDS model offers simplicity, the PHY model characterizes IRS responses to waves from different impinging directions, and the INW model allows connected reflecting elements at higher hardware complexity.The article also presents a qualitative comparison of the models and the TC framework.Source: §V · pp. 14–15
  • The TC framework is advocated to facilitate the design of large IRS-empowered systems.Fig. 4 presents a qualitative comparison of the IRS models and the framework in terms of optimization, tractability for analysis, and performance.Source: §V · p. 15
  • For high-frequency systems, the PHY model is identified as a candidate for capturing the properties of limited-scattering propagation environments.The passage highlights millimeter-wave and Terahertz systems, where signals are vulnerable to blockages because of poor scattering.Source: §V · p. 15
  • Long-term statistical CSI could reduce signaling and hardware implementation complexity, while IRS position optimization based on long-term CSI remains open.The passage contrasts this direction with current methodologies that rely mostly on instantaneous CSI and rapid switching of reflecting elements.Source: §V · pp. 15–16
  • AI-based techniques are proposed as a direction for reducing design complexity, with data-driven DL supporting real-time allocation and model-driven DL reducing demand for training data.The passage notes that optimization techniques currently used still have relatively high computational complexity.Source: §V · p. 16

Related papers

Loading 2103.07046v1…