PaperlayerView exact revision on arXiv

Source-linked AI summary

Intelligent Reflecting Surface Aided Wireless Communications: A Tutorial

Qingqing Wu, Shuowen Zhang, Beixiong Zheng, Changsheng You, Rui Zhang

arXiv:2007.02759v2cs.ITcs.LGcs.NI

TL;DR

Integrating IRS into wireless networks raises challenges in reflection optimization, channel estimation, and deployment. This tutorial surveys IRS-aided communication and concludes that it marks a shift from active-only wireless design toward systems with active and passive components.

  • Problem

    Accurately acquiring channel state information for IRS-reflected links is a central challenge in integrating IRS into wireless networks.

    Source: F. Other Related Work and Future… · p. 43

  • Method

    The paper provides a comprehensive tutorial on IRS technology, covering fundamentals, communication challenges, and directions for further investigation.

    Source: §VII · p. 67

  • Results

    The authors characterize IRS-aided wireless communication as a paradigm shift from systems with active components alone to designs combining active and passive components.

    Source: §VII · p. 67

  • Takeaways & Limitations

    The paper identifies promising directions for further investigation in IRS-aided wireless communication.

    Source: §VII · p. 67

Abstract

from arXiv · show

Intelligent reflecting surface (IRS) is an enabling technology to engineer the radio signal prorogation in wireless networks. By smartly tuning the signal reflection via a large number of low-cost passive reflecting elements, IRS is capable of dynamically altering wireless channels to enhance the communication performance. It is thus expected that the new IRS-aided hybrid wireless network comprising both active and passive components will be highly promising to achieve a sustainable capacity growth cost-effectively in the future. Despite its great potential, IRS faces new challenges to be efficiently integrated into wireless networks, such as reflection optimization, channel estimation, and deployment from communication design perspectives. In this paper, we provide a tutorial overview of IRS-aided wireless communication to address the above issues, and elaborate its reflection and channel models, hardware architecture and practical constraints, as well as various appealing applications in wireless networks. Moreover, we highlight important directions worthy of further investigation in future work.

I. INTRODUCTION · A. Motivation · B. What is IRS?

The section motivates IRS as a passive, low-cost way to make wireless propagation reconfigurable and complement active-network expansion toward B5G/6G goals. It describes tunable reflecting or refracting surfaces and their potential applications, alongside growing commercial and research interest.

A. Motivation

  • 6G targets higher data rates, energy efficiency, coverage, connectivity, reliability, and lower latency, while existing technology trends may not meet these goals or deliver affordable capacity growth.The authors call for future capacity growth with low cost, complexity, and energy consumption.Source: A. Motivation · p. 2
  • Mobility-driven channel variation challenges ultra-reliable communication, while conventional approaches incur overhead and have limited control over largely random channels.The section frames this limited channel control as a remaining barrier to high-capacity, ultra-reliable communication.Source: A. Motivation · pp. 2–3

B. What is IRS?

  • An IRS uses many passive elements to alter incident signals and reconfigure propagation toward desired channel realizations or distributions, addressing fading and interference.Each element can independently impose a controllable amplitude and/or phase change.Source: B. What is IRS? · p. 3
  • IRS-aided MIMO can use substantially fewer base-station antennas without compromising user QoS, reducing hardware cost and energy consumption, especially at higher frequencies.The approach uses the IRS's large aperture to create fine-grained reflected beams via smart passive reflection.Source: B. What is IRS? · p. 4
  • IRSs shift active-only heterogeneous networks toward a hybrid active-passive architecture; their lower cost permits denser deployment and cost-sensitive choices of active-base-station-to-IRS ratios for capacity scaling [11].The passage presents this as a way to sustain network capacity growth with cost.Source: B. What is IRS? · p. 5
  • For dead zones, an IRS can create a virtual line-of-sight link around obstacles, extending coverage in blockage-vulnerable mmWave and THz communications; at cell edges, it can improve desired signal power and suppress neighboring-cell interference.These examples illustrate IRS use for coverage extension and co-channel interference reduction.Source: B. What is IRS? · pp. 5–6
  • For SWIPT, IRS reflect beamforming can offset long-distance power loss, while indoor installations can improve coverage and create high-capacity hotspots for eMBB and mMTC.Suggested indoor placements include ceilings, walls, and furniture in settings such as factories, stadiums, shopping centers, and airports.Source: B. What is IRS? · p. 6
  • IRS can support URLLC for remote control and smart transportation by compensating Doppler effects [12].The passages identify these as examples of applications supported by IRS.Source: B. What is IRS? · p. 6
  • IRS can make the “dumb” environment intelligent, potentially benefiting vertical industries such as transportation, manufacturing, and smart cities.The passages place these potential benefits in the context of 5G/6G.Source: B. What is IRS? · p. 6
  • IRS has been recognized as a promising future 6G technology, [14], amid industry interest in commercialization and pilot projects.The passages direct readers to Table I for more information about the pilot projects.Source: B. What is IRS? · p. 6
  • Related terms include intelligent wall, smart reflectarray, RIS,, LISA, and RFocus [20], which share the principle of passive, tunable reflecting or refracting surfaces.The passages note that these terms differ in terminology but are based on the same principle.Source: B. What is IRS? · p. 6

C. What’s New? … A. IRS Signal and Channel Model

The tutorial examines IRS as a means of reconfiguring wireless propagation and addresses related communication-design challenges. It introduces IRS signal and channel models, including element-level cascaded reflection paths and their double path loss, while noting practical constraints and cautioning against ideal-PEC sum-distance scaling for finite tunable IRSs.

C. What’s New?

  • IRS extends reconfigurable metasurfaces from controlling electromagnetic waves to enabling smart, reconfigurable propagation environments for wireless communication.The paper contrasts this communication use with metasurface applications such as cloaking, imaging, radar sensing, and holograms.Source: C. What’s New? · p. 6

D. Objective, Contribution, and Organization

  • This paper presents the first tutorial on IRS-aided wireless communications, systematically treating passive reflection optimization, channel estimation, and deployment from communication perspectives.It aims to review state-of-the-art results and provide technical discussion to facilitate future research in IRS-aided networks.Source: D. Objective, Contribution, and… · pp. 8–9

II. IRS FUNDAMENTALS

  • IRS fundamentals cover signal and channel models, hardware architecture, practical constraints, and the resulting issues for future investigation.The paper also outlines later sections on reflection optimization, channel estimation, deployment, and other IRS topics.Source: §II · p. 9, D. Objective, Contribution, and… · p. 9

A. IRS Signal and Channel Model

  • The illustrative model is a narrow-band, single-antenna point-to-point link assisted by an IRS with N passive reflecting elements.Each element attenuates and phase-shifts its incident signal; the model assumes negligible delay relative to the signal bandwidth and considers phase shifts in [0, 2π).Source: A. IRS Signal and Channel Model · p. 10
  • The IRS-reflected channel combines the transmitter-to-element channel, element reflection, and element-to-receiver channel, with the N element signals superposed.The IRS reflection is represented by a diagonal complex matrix, corresponding to independent element reflections without joint processing.Source: A. IRS Signal and Channel Model · p. 11
  • The model assumes no coupling between neighboring elements and ignores signals reflected by the IRS more than once because of substantial path loss.Under these assumptions, all IRS elements reflect incident signals independently.Source: A. IRS Signal and Channel Model · p. 11
  • Under the far-field assumption, each IRS-reflected path suffers double attenuation described by a product-distance path-loss model.The average received power via an element is inversely proportional to the product of the two hop-distance path-loss terms.Source: A. IRS Signal and Channel Model · p. 12
  • The sum-distance model for an infinitely large PEC under free-space propagation is generally inapplicable to element-level IRS channels and finite-size tunable PECs.It is inappropriate to use that model to conclude that received power scales with the number of PECs through multiplicative passive beamforming gains, even under free-space propagation.Source: A. IRS Signal and Channel Model · pp. 12–13
  • More theoretical and experimental study of IRS-reflected-channel path loss is needed [40]–[44].Source: A. IRS Signal and Channel Model · p. 13

B. IRS Architecture, Hardware, and Practical Constraints

IRS uses reconfigurable reflecting elements and a controller to adapt reflections to changing channels, while its hardware choices determine control flexibility and practical performance limits.

B. IRS Architecture, Hardware, and Practical Constraints

  • Because wireless channels vary with transmitter, receiver, and environmental mobility, IRS elements need dynamically adjustable reflection coefficients for real-time adaptation.Source: B. IRS Architecture, Hardware, and… · p. 13
  • The typical IRS has tunable metallic patches, a copper layer to reduce signal leakage, and a control-circuit layer that tunes reflection amplitude and/or phase in real time.A smart controller triggers and determines reflection adaptation.Source: B. IRS Architecture, Hardware, and… · p. 14
  • IRS elements can be reconfigured mechanically, with functional materials, or with electronic devices; electronic devices are widely adopted for fast response and low reflection loss.The passage also identifies relatively low energy consumption and hardware cost as advantages of electronic devices.Source: B. IRS Architecture, Hardware, and… · pp. 14–15
  • 5 MHz switching frequency (0.2 µs switching time) is much faster than millisecond-scale channel coherence, supporting PIN-diode IRS use with time-varying channels.Biasing a PIN diode switches an element between ON and OFF states, producing a π phase-shift difference.Source: B. IRS Architecture, Hardware, and… · p. 15
  • Amplitude control adds flexibility for communication objectives and can trade reflection performance against hardware cost, while independent amplitude and phase control requires more sophisticated hardware.The passage notes amplitude control is generally less costly to implement than phase control.Source: B. IRS Architecture, Hardware, and… · p. 15
  • Ideal continuous, independent amplitude and phase tuning yields feasible reflection sets that typically imply theoretical performance limits under practical reflection models.Source: B. IRS Architecture, Hardware, and… · pp. 15–16

1) Discrete reflection amplitude and phase shift: · 2) Coupled reflection amplitude and phase shift:

Finite reflection levels reduce IRS hardware and control complexity but constrain performance and optimization. Practical elements also couple reflection amplitude with phase, while IRS operation still consumes substantially less energy than active relays.

1) Discrete reflection amplitude and phase shift:

  • Finite discrete amplitude and phase levels need few control bits per element, but implementing eight phase levels requires at least three PIN diodes.Two-level examples include reflecting-or-absorbing amplitude control and 0-or-π phase control,.Source: 1) Discrete reflection amplitude and… · p. 16
  • For bβ and bθ control bits, the corresponding amplitude and phase-shift level counts are Kβ = 2^bβ and Kθ = 2^bθ.The modeled levels lie in [0, 1] for amplitude and [0, 2π) for phase shift.Source: 1) Discrete reflection amplitude and… · pp. 16–17
  • Quantized amplitude and phase control coarsens reflected-signal control, degrades communication performance, and creates harder discrete optimization variables.The discrete variables are generally more difficult to handle than their continuous counterparts.Source: 1) Discrete reflection amplitude and… · p. 17
  • Discrete phase-shift control costs more than amplitude control but achieves better passive beamforming at the same number of control bits or levels per element.Practical designs may implement only discrete phase-shift control or only discrete amplitude control to reduce hardware cost and complexity.Source: 1) Discrete reflection amplitude and… · p. 17

2) Coupled reflection amplitude and phase shift:

  • A practical IRS element’s reflection amplitude is nonlinearly coupled to its phase shift, so the two cannot be adjusted independently.The amplitude is typically lowest at zero phase shift and rises toward one as phase approaches π or −π, consistent with experiments.Source: 2) Coupled reflection amplitude and… · p. 18
  • Tuning effective capacitance or resistance changes reflection amplitude; phase near zero enhances currents and energy dissipation, while phase near π or −π diminishes them and reduces energy loss.This mechanism accounts for the amplitude minimum near zero phase and maximum near π or −π.Source: 2) Coupled reflection amplitude and… · p. 18
  • Optimal reflection design must balance each element’s reflected signal amplitude and phase so all reflected signals combine at the receiver for maximum power or SNR.The practical element model therefore constrains reflection-coefficient design.Source: 2) Coupled reflection amplitude and… · p. 19
  • An IRS controller with 256 reflecting elements consumes only about 0.72 W [41], significantly less than an active BS/relay in practice; IRS reflects without power amplification but still needs power for reconfiguration.Source: 2) Coupled reflection amplitude and… · p. 19

C. Other Related Work and Future Direction · III. IRS REFLECTION OPTIMIZATION · A. IRS-aided SISO System: Passive Beamforming Basics and Power Scaling Order

The paper identifies unresolved IRS hardware and channel-modeling challenges, then develops passive-reflection optimization from a perfect-channel-knowledge setup to a SISO beamforming solution. In the SISO case, phase alignment enables coherent combining, while the optimal design depends on the cascaded IRS channels.

C. Other Related Work and Future Direction

  • Mutual coupling can invalidate the linear IRS channel model as denser elements interact, motivating nonlinear modeling and decoupling techniques.Increasing element count within a fixed IRS size reduces spacing and can make coupling non-negligible; suggested isolation approaches include defected ground structures, parasitic scatterers, and neutralization.Source: C. Other Related Work and Future… · p. 20
  • Measurements in [41] show that IRS phase shifts can depend strongly on incident angle, complicating reflection optimization across multipath AoAs and potentially invalidating TDD reciprocity for downlink training.If reciprocity fails, uplink-based channel estimation is not applicable to downlink communication.Source: C. Other Related Work and Future… · p. 20
  • The constant-phase-over-bandwidth assumption may fail when B is comparable to fc, causing frequency-dependent phase errors in multicarrier signals and motivating compensation and further study of hardware imperfections.The passage also identifies frequency/time offset and phase noise as hardware effects whose impact on reflection design and achievable performance remains largely unknown.Source: C. Other Related Work and Future… · p. 20

III. IRS REFLECTION OPTIMIZATION

  • Reflection optimization is studied across single- and multi-user, single- and multi-antenna, narrowband and broadband, and single- and multi-cell setups, assuming perfect channel knowledge.Channel estimation is addressed separately in Section IV.Source: §III · p. 21

A. IRS-aided SISO System: Passive Beamforming Basics and Power Scaling Order

  • For SISO narrowband flat fading, optimal IRS phases align reflected signals with the direct AP-user signal for coherent combining, independent of their reflection amplitudes,.This alignment maximizes received signal power; when the direct link is negligible, a common phase shift does not change the optimum.Source: A. IRS-aided SISO System: Passive… · pp. 21–22
  • Maximizing the IRS reflection amplitudes helps achieve maximum user receive power, with coherent combining given as the reason for this benefit.Source: A. IRS-aided SISO System: Passive… · p. 23
  • The SISO optimal IRS design depends on the cascaded channels through the IRS, not on knowing the individual AP-to-IRS and IRS-to-user channels.The passage notes that this property can simplify IRS channel-estimation design.Source: A. IRS-aided SISO System: Passive… · p. 23

2. Assume i.i.d. Rayleigh fading · 2) Comparison with M-MIMO and MIMO Relay: · B. IRS-aided MISO System: Joint Active and Passive Beamforming

This section examines IRS scaling effects on received power and coverage, compares IRS with active-array and relay systems, and discusses joint active and passive beamforming methods and their optimization challenges. It also describes how fixed MRT choices perform differently depending on user location relative to the AP and IRS, while joint beamforming balances direct and reflected transmission.

2. Assume i.i.d. Rayleigh fading

  • IRS-aided receive power grows as O(N^2), combining O(N) reflection gain and O(N) AP–IRS aperture gain, and AP transmit power can fall by 1/N^2 without reducing receive SNR.The transmit-power scaling is stated for sufficiently large N.Source: §2 · p. 23
  • About 2 bps/Hz of achievable rate is gained when N doubles from 200 to 400 across the considered channel models, validating the asymptotic scaling.The observation applies when N is sufficiently large.Source: §2 · pp. 23–24, Figure 9 · p. 24
  • To maintain the same average receive power at fixed AP transmit power, the IRS element count must increase linearly with the squared coverage distance.The passage identifies d2 as the relevant distance in the coverage expression.Source: §2 · pp. 24–25

2) Comparison with M-MIMO and MIMO Relay:

  • FD AF MIMO relays achieve only O(N) receive-SNR scaling because forwarded relay noise grows linearly with N, and HD AF relays have the same scaling order.The relay signal power itself scales as O(N^2), but processing noise limits the SNR gain.Source: 2) Comparison with M-MIMO and MIMO Relay · p. 25
  • Because passive IRS elements need no transmit RF chains, a given hardware cost can support more reflecting elements and potentially comparable or superior performance to active M-MIMO or relay systems.The passage cautions that equal active and passive element counts may be an unfair cost comparison.Source: 2) Comparison with M-MIMO and MIMO Relay · p. 25
  • At small N, IRS performs worst due to product-distance path loss and limited passive gain, but its rate gaps narrow as N grows and it can outperform HD MIMO relays [8].The comparison also attributes IRS's advantage at large N to O(N^2) versus O(N) beamforming gain and more spectral-efficient FD operation.Source: 2) Comparison with M-MIMO and MIMO Relay · p. 25, Figure 10 · p. 26

B. IRS-aided MISO System: Joint Active and Passive Beamforming

  • In the MISO case, AP transmit beamforming and IRS phase shifts must be jointly optimized; alternating optimization fixes each in turn and reaches at least a locally optimal solution.With fixed IRS phases, the AP uses maximum-ratio transmission; the method applies to the reverse-link SIMO case as well.Source: B. IRS-aided MISO System: Joint Active… · pp. 26–27
  • The phase-only MISO formulation must balance channel powers across transmit antennas and is NP-hard, as its equivalent unit-modulus QCQP is generally NP-hard.Thus, an optimal solution cannot generally be obtained with complexity polynomial in N.Source: B. IRS-aided MISO System: Joint Active… · pp. 27–28
  • AP-user MRT performs close to AO when the user is nearer the AP but loses considerable rate when the user is nearer the IRS.This follows the shift from a dominant AP-user direct link to a dominant IRS-user link.Source: B. IRS-aided MISO System: Joint Active… · p. 28
  • AP-IRS MRT behaves oppositely to AP-user MRT, and for d ≤40 m it can achieve a lower rate than conventional MRT without the IRS.The comparison shows that using the IRS does not guarantee a rate gain when transmit beamforming is not designed based on all channels.Source: B. IRS-aided MISO System: Joint Active… · p. 28
  • Joint active and passive beamforming is essential to balance transmission directly toward the user and via IRS reflection to maximize received signal power.The comparison considers AO joint design alongside AP-user MRT and AP-IRS MRT schemes.Source: B. IRS-aided MISO System: Joint Active… · p. 28

C. IRS-aided MIMO and OFDM Systems · 1) IRS-aided MIMO System:

IRS-aided MIMO and OFDM systems require reflection optimization for multi-antenna and frequency-selective channels. For MIMO, the tutorial models the effective channel and describes joint reflection and transmit-covariance optimization, an alternating algorithm, and resulting channel-capacity and spatial-multiplexing gains.

C. IRS-aided MIMO and OFDM Systems

  • IRS reflection optimization becomes more challenging for multi-antenna MIMO and broadband OFDM because it must handle multi-antenna and/or multipath channels with differing delays.Source: C. IRS-aided MIMO and OFDM Systems · p. 29

1) IRS-aided MIMO System:

  • The effective MIMO channel is H̃ = Hd + HrΘG, and achieving capacity requires jointly optimizing the IRS reflection and AP transmit covariance matrix [59].The channels Hd, G, and Hr represent the AP-user, AP-IRS, and IRS-user links, respectively.Source: 1) IRS-aided MIMO System · p. 29
  • With continuous phase shifts and unit reflection amplitudes, the capacity problem is non-convex because MIMO capacity is a log-determinant function of the IRS matrix and covariance jointly.Source: 1) IRS-aided MIMO System · p. 30
  • The AO algorithm alternates optimization of one IRS phase or the transmit covariance, converges to at least a locally optimal solution, and has polynomial complexity in N, Mt, or Mr.For fixed covariance and other phases, the remaining phase has a closed-form optimum; for fixed IRS reflection, covariance follows eigenmode transmission with water-filling spatial power allocation,.Source: 1) IRS-aided MIMO System · p. 30
  • IRS reflection design can improve channel total power, condition number, and rank; for low-rank or rank-one direct channels, it can yield higher rank and larger spatial multiplexing gain than without IRS.Source: 1) IRS-aided MIMO System · p. 30
  • The unit-amplitude assumption reflects the practical difficulty of jointly tuning each element’s amplitude and phase, while flexible amplitude design within [0, 1] may further improve capacity.Source: 1) IRS-aided MIMO System · p. 30
  • In Fig. 12’s setup, the AO-based IRS-aided MIMO system achieves significantly higher rate than both the no-IRS MIMO capacity and a random-phase IRS benchmark.The benchmark independently and uniformly generates each IRS phase shift in [0, 2π).Source: Figure 12 · p. 31, 1) IRS-aided MIMO System · p. 31

2) IRS-aided OFDM System: · D. IRS-aided Multi-user System

The OFDM section models frequency-selective channels under shared, frequency-flat IRS phases and compares optimization schemes; the following section frames narrow-band multiuser SISO and MISO access schemes for comparison with conventional systems.

2) IRS-aided OFDM System:

  • For frequency-selective OFDM with B ≪ fc, a CP of at least L=max(L0,L1+L2−1) removes inter-symbol interference, while each IRS phase affects all subcarriers identically.The IRS reflection is frequency-flat in this model.Source: 2) IRS-aided OFDM System · pp. 31, 33
  • In Fig. 13, SCA achieves significantly higher rate than OFDM without IRS and with random IRS phases plus optimal power allocation.The illustrated system uses random IRS phases with the corresponding optimal transmit power allocation.Source: Figure 13 · p. 33, 2) IRS-aided OFDM System · p. 34
  • Rate maximization jointly tunes shared IRS phases for frequency-varying channels and per-subcarrier powers; SCA uses a Taylor-based concave lower bound and converges to stationary points with polynomial complexity.The optimization is more difficult than in the narrow-band case.Source: 2) IRS-aided OFDM System · pp. 33–34
  • Strongest-CIR maximization, which aligns IRS phases with the strongest-power time-domain path, performs close to SCA at lower complexity.It is presented as an efficient alternative to the SCA-based algorithm.Source: 2) IRS-aided OFDM System · p. 34
  • An ideal frequency-selective IRS rate upper bound substantially exceeds practical frequency-flat SCA performance, with the gap increasing as the number of subcarriers grows.This comparison identifies the lack of frequency-selective reflection due to passive operation as a fundamental limitation of IRS-aided OFDM.Source: 2) IRS-aided OFDM System · p. 34
  • For MIMO-OFDM, IRS reflection design must address channels across space and frequency and jointly optimize transmit covariance matrices; AO with convex relaxation improves rate over systems without IRS.The cited results report that IRS remains effective despite the lack of frequency selectivity.Source: 2) IRS-aided OFDM System · p. 34

D. IRS-aided Multi-user System

  • The multi-user section sets up narrow-band SISO comparisons of OMA (TDMA or FDMA) and NOMA, plus MISO SDMA serving multiple single-antenna users simultaneously, to compare IRS and conventional systems.The stated objective is to highlight differences in system design and performance optimization with versus without IRS.Source: D. IRS-aided Multi-user System · p. 35

1) OMA versus NOMA: · 2) SDMA: · E. Passive Beamforming with Discrete Reflection Amplitude and Phase Shift

The section examines IRS reflection strategies across multiple-access schemes, SDMA, and practical discrete-control constraints, including their effects on multi-user performance and beamforming optimization. It compares discrete phase-shift and binary amplitude control, noting their differing effectiveness.

1) OMA versus NOMA:

  • In the illustrated two-user case, TDMA and NOMA require less transmit power than FDMA, but symmetric rate targets favor TDMA over NOMA, whose performance is less sensitive to rate disparity.The broader comparison attributes TDMA’s advantage over FDMA to the latter’s lack of frequency-selective IRS reflection; unlike traditional multi-user systems, IRS adaptation can also let TDMA outperform NOMA for near-IRS users with symmetric targets.Source: 1) OMA versus NOMA · pp. 35–36
  • For K > 2, FDMA and NOMA reflection design becomes harder as more user channels must be served simultaneously, limiting each user’s passive beamforming gain.A dynamic OFDMA scheme serves selected user subsets in separate time slots and adjusts IRS beamforming across slots to exploit time selectivity and multi-user channel diversity, outperforming optimized fixed reflections.Source: 1) OMA versus NOMA · p. 36

2) SDMA:

  • IRS can reduce channel correlation among nearby users through controllable signal paths, enabling interference cancellation and improved SINR for users both aided and unaided by IRS.This addresses conventional SDMA’s difficulty serving nearby or cell-edge users whose similar channel directions increase co-channel interference and undermine zero-forcing.Source: 2) SDMA · p. 37
  • Multi-user active and passive beamforming are coupled; alternating optimization yields high-quality suboptimal solutions, while a penalty-based BCD method decouples QoS constraints and guarantees at least a locally optimal solution, [64], [65],.AO can become inefficient as QoS constraints increase because of the risk of undesired suboptimal solutions.Source: 2) SDMA · p. 37

E. Passive Beamforming with Discrete Reflection Amplitude and Phase Shift

  • Discrete phase shifts and/or discrete reflection amplitudes may misalign signals at designated receivers, degrading communication performance versus the ideal; these constraints also make IRS reflection-design optimization harder than continuous counterparts.Source: E. Passive Beamforming with Discrete… · p. 38
  • Relaxing discrete constraints, solving with continuous values, then quantizing to nearest allowed levels reduces complexity to polynomial orders of N but can incur quantization loss and remain suboptimal.The loss depends on the number of discrete levels and N; alternating optimization can further refine each element’s discrete amplitude or phase while holding the others fixed.Source: E. Passive Beamforming with Discrete… · p. 38
  • The discrete-phase solution first optimizes continuous shifts and quantizes each to its nearest allowed value; its objective is a lower bound that converges to the continuous-phase maximum as bθ → ∞.Source: E. Passive Beamforming with Discrete… · p. 39
  • For sufficiently large N, discrete-phase IRS retains O(N^2) receive-power scaling, and its power ratio η(bθ) depends on the phase-level count rather than N.This means the asymptotic scaling order matches the continuous-phase case.Source: E. Passive Beamforming with Discrete… · p. 39
  • η(1) = −3.9 dB and η(2) = −0.9 dB; for sufficiently large N, simulated 1-bit and 2-bit phase-shift SNR losses approach 3.9 dB and 0.9 dB, respectively.The 2-bit design is described as close to optimal within 1 dB.Source: E. Passive Beamforming with Discrete… · pp. 39, 41
  • The binary amplitude scheme in [20] incurs about 10 dB receive-power loss relative to continuous phase shifts.It fixes all phases to zero and optimizes each reflection amplitude as either 0 or 1.Source: E. Passive Beamforming with Discrete… · p. 39
  • 1-bit phase control uses both reflected signals, whereas 1-bit amplitude control uses only one, so amplitude control generally performs worse or requires a larger IRS to match it.The simulations likewise show substantial power/SNR loss for 1-bit amplitude control because ON/OFF control cannot fully utilize the reflected signals.Source: E. Passive Beamforming with Discrete… · p. 41

F. Other Related Work and Future Direction · IV. IRS CHANNEL ESTIMATION · A. Problem Description and Challenges

The paper surveys future directions for IRS reflection design and frames accurate channel estimation as a key practical challenge. It examines estimation in semi-passive and fully passive IRS configurations, where passive elements cannot transmit training signals.

F. Other Related Work and Future Direction

  • Multiple-AP and multiple-IRS networks require joint design of IRS reflections and AP transmissions to improve desired-signal power and mitigate intra-cell and inter-cell interference [77], [78].Source: F. Other Related Work and Future… · p. 41
  • The ideal-model O(N^2) asymptotic power-scaling order still holds with phase-shift-dependent non-uniform reflection amplitudes; practical hardware constraints and efficient algorithms, including machine learning for large IRSs, merit study.Source: F. Other Related Work and Future… · pp. 41–42
  • Research directions include outage probability and average BER [81], asymptotic max-min SINR under large-scale fading, and reflection design and spatial-throughput analysis for large hybrid networks.Source: F. Other Related Work and Future… · pp. 42–43
  • Reported IRS gains assume perfect CSI, but accurately acquiring IRS-reflected-link CSI is difficult because IRSs are passive and have many elements, making estimation and optimization with imperfect or partial CSI important.Source: F. Other Related Work and Future… · pp. 43–44

A. Problem Description and Challenges

B. Semi-Passive IRS Channel Estimation · C. Passive IRS Channel Estimation

The section surveys semi-passive and fully passive IRS channel-estimation approaches, including their protocols, constraints, and challenges. It also discusses passive-estimation techniques and training/reflection design for managing overhead and supporting beamforming and achievable rate.

B. Semi-Passive IRS Channel Estimation

  • Semi-passive IRS alternates sensor-based channel sensing with reflection, and each coherence interval allocates phases to channel estimation, CSI exchange and joint beamforming, then data transmission.Sensors receive pilots while reflecting elements are off; they are deactivated when the IRS reflects data signals.Source: B. Semi-Passive IRS Channel Estimation · p. 46
  • Semi-passive sensors estimate BS/user-to-IRS CSI, but reverse-link CSI is available only through TDD reciprocity, making this estimation infeasible in FDD.Sensor channels are correlated with, but not identical to, channels at the reflecting elements; high-dimensional IRS channels must be reconstructed from these lower-dimensional estimates.Source: B. Semi-Passive IRS Channel Estimation · pp. 46–47
  • Semi-passive estimation accuracy depends on sensor count, ADC resolution and sensing time; more sensors, finer quantization and longer sensing can improve reconstruction, but systematic trade-off studies remain lacking.Compressed sensing, interpolation and machine learning can exploit spatial correlation to reconstruct the higher-dimensional channels.Source: B. Semi-Passive IRS Channel Estimation · p. 47

C. Passive IRS Channel Estimation

  • Fully passive IRS estimation directly measures cascaded user-IRS-BS channels, whose coefficient count is generally larger than that of separate links, despite redundancy from their shared IRS-BS channel.Cascaded-link CSI is generally sufficient for joint passive/active beamforming without loss of optimality, and requires no sophisticated channel reconstruction.Source: C. Passive IRS Channel Estimation · pp. 47–48
  • Passive estimation uses pilots and a pre-designed IRS reflection pattern to estimate direct and cascaded channels, then designs IRS reflection and receiver beamforming for data transmission.In TDD, uplink CSI can support downlink reflection design; in FDD, the BS and users swap roles for downlink estimation, with users feeding CSI back to the BS.Source: C. Passive IRS Channel Estimation · p. 48
  • The new, non-trivial passive-estimation problem is to jointly design pilots, IRS patterns and receiver processing to estimate direct and cascaded channels with minimum pilot-symbol overhead.The exposition first considers a single-user, flat-fading SISO system with one antenna at the BS and user.Source: C. Passive IRS Channel Estimation · p. 49
  • Channel coefficients can be estimated over N + 1 pilot-symbol durations; estimating each sub-surface’s IRS-BS channel greatly reduces overhead, and grouping flexibly trades training/design overhead/complexity against IRS passive beamforming gain.Source: C. Passive IRS Channel Estimation · p. 49
  • Using orthogonal pilots over different BS/user transmit antennas for downlink/uplink training increases total overhead by MB or Mu times versus the single-user SISO case; low-rank, sparsity, and spatial correlation can facilitate cascaded-channel training.Deep learning and hierarchical searching algorithms have also been developed.Source: C. Passive IRS Channel Estimation · p. 50
  • For multiple users, exploiting their common IRS-BS channel allows a reference user’s cascaded channel to support estimation of others through low-dimensional scaling factors.The other users’ cascaded channels are scaled versions of the reference channel, so only scaling factors rather than full high-dimensional channels need estimation; considers the multi-user MISO case.Source: C. Passive IRS Channel Estimation · p. 50
  • Broadband frequency-selective estimation is harder because it requires more coefficients while IRS reflections are frequency-flat, but OFDM redundancy enables concurrent multi-user estimation with customized pilots [83].The cited approach requires only N + 1 OFDM symbols; OFDM-subcarrier redundancy supports more concurrent users and yields overhead independent of the number of BS receive antennas [83].Source: C. Passive IRS Channel Estimation · pp. 50–51
  • Finite phase-shift levels make orthogonal training reflections difficult to design; quantization can construct near-orthogonal coefficients, but estimation errors are higher and correlated,.The resulting correlated errors can induce correlated interference during data transmission, which must be considered in passive beamforming design,.Source: C. Passive IRS Channel Estimation · pp. 51–52
  • Existing all-at-once estimation of every IRS-element cascaded channel requires long training time that may be unavailable in short-packet scenarios.The cited discussion contrasts all-at-once estimation with scenarios where long training time is practically unavailable.Source: C. Passive IRS Channel Estimation · p. 52
  • Flexible training protocols with adjustable pilot durations can adapt to different transmission requirements and channel coherence times.Source: C. Passive IRS Channel Estimation · p. 52
  • Hierarchical training patterns progressively resolve cascaded IRS CSI through element grouping, enabling successively refined passive beamforming and improved rate block by block.Source: C. Passive IRS Channel Estimation · p. 52
  • With fixed coherence time, achievable rate first increases and then decreases as the element grouping ratio rises, trading improved CSI and beamforming gain against data time.Too little training can yield insufficient CSI and inefficient reflection design; too much leaves less time for data transmission.Source: C. Passive IRS Channel Estimation · pp. 52–53
  • DFT-based reflection patterns achieve much larger rates than ON/OFF and heuristic random-phase patterns, while discrete phase shifts incur loss versus continuous adjustment.The discrete-phase comparison uses Kθ uniformly quantized levels in [0, 2π).Source: C. Passive IRS Channel Estimation · p. 53

D. Other Related Work and Future Direction

The section identifies unresolved channel-estimation questions involving semi-passive IRS, broadband channels, and practical system assumptions.

D. Other Related Work and Future Direction

V. IRS DEPLOYMENT … 1) Optimal Deployment of Single IRS:

IRS deployment choices shape reflected channels and system performance, while practical constraints and passive double path loss make placement distinct from active-node deployment. For a single IRS in the stated link setup, placing it near the AP or user maximizes receive SNR, whereas a middle placement gives the smallest SNR.

V. IRS DEPLOYMENT

  • IRS can act as a local hub that improves nearby users’ communication performance through optimized signal reflection.Table IV is identified as a summary of representative works, but its contents are not provided here.Source: §V · p. 53

A SUMMARY OF REPRESENTATIVE WORKS ON IRS CHANNEL ESTIMATION

  • IRSs can be placed near the BS/AP or users, or divided into smaller distributed IRSs; these choices change reflected-channel realizations and distributions, and thus fundamental performance limits.Deployment must also account for cost, user demand and distribution, space, and propagation environment.Source: A SUMMARY OF REPRESENTATIVE WORKS ON… · p. 54
  • Because passive IRS reflection causes severe product-distance/double path-loss without signal amplification/regeneration, an IRS should be close to the transmitter or receiver, unlike active BSs/APs, which need geographic separation for coverage maximization,.Source: A SUMMARY OF REPRESENTATIVE WORKS ON… · pp. 54–55

A. IRS Deployment Optimization at the Link Level

  • The deployment discussion studies maximum achievable rates or capacity by optimizing IRS placement, separating link-level single-user design from network-level multi-user design.It also notes other, more general deployment setups and future directions.Source: A SUMMARY OF REPRESENTATIVE WORKS ON… · p. 55

1) Optimal Deployment of Single IRS:

  • For β0 = −30 dB, placing the IRS near the user or AP yields the largest SNR, whereas placing it around the middle between them leads to the smallest SNR.Source: 1) Optimal Deployment of Single IRS · p. 57
  • Under blocked direct-link and free-space LoS assumptions, double path loss makes the optimal IRS position one that minimizes either the user–IRS or IRS–AP distance.The link setup assumes single-antenna AP and user and has N reflecting elements forming one IRS.Source: 1) Optimal Deployment of Single IRS · pp. 56–57

2) Single IRS versus Multiple Cooperative IRSs: · 3) LoS versus Non-LoS (NLoS): · B. IRS Deployment at the Network Level: Distributed or Centralized?

This section examines IRS cooperation and deployment choices, weighing cooperative gains against added path loss and the LoS–NLoS trade-off between low path loss and spatial multiplexing. It compares centralized and distributed deployment in multi-user networks, including achievable-rate regions across TDMA, NOMA, and FDMA, alongside practical deployment constraints.

2) Single IRS versus Multiple Cooperative IRSs:

  • Multiple cooperative IRSs trade added inter-IRS path loss against multiplicative beamforming gains; under far-field rank-one coupling, aligned two-IRS gain scales as O((N/2)^4), versus O(N^2) for one IRS.The additional path loss in the two-IRS case is fixed for a given setup, while the cooperative beamforming gain is expected to become more pronounced as N increases.Source: 2) Single IRS versus Multiple… · pp. 57–58

3) LoS versus Non-LoS (NLoS):

  • LoS placement reduces path loss, but low-rank LoS channels constrain high-SNR spatial multiplexing,; mixed LoS/NLoS components can improve channel rank, conditioning, and capacity.This is especially relevant at mmWave/THz, where NLoS paths are weakened by blockage and penetration losses.Source: 3) LoS versus Non-LoS (NLoS) · pp. 58–59

B. IRS Deployment at the Network Level: Distributed or Centralized?

  • For TDMA, centralized deployment provides O(N^2) beamforming gain per user versus O((N/K)^2) when distributed, and its rate region contains the distributed region under twin channels.The centralized deployment’s rate advantage over distributed deployment becomes more pronounced as K increases, as each distributed user receives fewer reflecting elements.Source: B. IRS Deployment at the Network… · p. 60
  • Under the twin channel condition, centralized deployment outperforms distributed deployment in multi-user capacity region for NOMA and achievable-rate region for FDMA; its larger number of reflecting elements provides more flexibility to trade off users’ individual achievable rates.Source: B. IRS Deployment at the Network… · pp. 60–61
  • Centralized deployment yields higher achievable-rate regions than distributed deployment for TDMA, NOMA, and FDMA, with larger gains when user rates are asymmetric.The comparison assumes channels follow a free-space LoS model.Source: B. IRS Deployment at the Network… · p. 61
  • Centralized deployment substantially improves the farther user’s rate, helping mitigate the near-far issue and balance the two users’ rates.The comparison assumes twin channel conditions,; different LoS probabilities, NLoS fading, or channel correlations may change the results.Source: B. IRS Deployment at the Network… · pp. 61–62
  • Distributed deployment requires more IRSs and backhaul links, increasing network overhead, while a large centralized IRS near the AP may be infeasible due to site or space constraints.Multiple distributed IRSs at the user side are generally more flexible; comparisons may also differ in scenarios such as MIMO or multiple users per distributed IRS.Source: B. IRS Deployment at the Network… · pp. 61–62

C. Other Related Work and Future Direction … B. Wireless Power Transfer

The extensions discuss IRS deployment and physical-layer security challenges, while showing how IRSs can support secrecy and wireless power transfer. They also identify channel-state-information costs and scalable deployment as open design concerns.

C. Other Related Work and Future Direction

  • IRS deployment research should extend prior results to cooperative IRSs, other multi-user/MIMO models, hybrid deployment, and general multi-cell networks with inter-cell interference.The broader deployment problem includes heterogeneous user QoS requirements.Source: C. Other Related Work and Future… · p. 62
  • In [77]’s simplified setup, one IRS serving two adjacent cells should be deployed at the intersection of their edges to maximally benefit both cells; with multiple IRSs serving multiple cells, the optimal deployment/association strategy remains unknown.Source: C. Other Related Work and Future… · p. 62

VI. EXTENSIONS

  • Section VI overviews other relevant IRS topics that are important and merit attention.The extensions include physical-layer security and wireless power transfer.Source: §VI · p. 62

A. Physical-Layer Security

  • IRS passive beamforming can increase legitimate-user rates while reducing eavesdropper rates, and joint artificial-noise design can further suppress eavesdropping without compromising legitimate-user performance,.These approaches address cases where conventional physical-layer security techniques are ineffective.Source: A. Physical-Layer Security · p. 63
  • IRS-aided secrecy and WPT depend on CSI that is difficult or costly to obtain, and large-scale secrecy networks need further IRS deployment study.For WPT, channel training must balance estimation accuracy against training energy and the time left for energy harvesting; deployment choices also affect performance.Source: A. Physical-Layer Security · p. 63, B. Wireless Power Transfer · p. 64

B. Wireless Power Transfer

  • IRS-assisted WPT can enhance received power for nearby IoT devices, reduce transmit power, and improve rate and energy performance in SWIPT systems, –[130].The proposed approach places IRSs near IoT devices, in line of sight with energy transmitters and receivers, and uses their large aperture and passive beamforming gains.Source: B. Wireless Power Transfer · p. 64
  • IRS-assisted WPT/SWIPT requires balancing channel estimation with power transfer: too little training reduces beamforming gain, while too much consumes receiver energy and harvesting time.The choice between single or multiple cooperative IRSs and centralized or distributed deployment also affects system performance.Source: B. Wireless Power Transfer · p. 64

C. UAV Communications … VII. CONCLUSION

The tutorial presents IRS as a basis for reconfigurable wireless networks combining active and passive components, surveys applications and communication challenges, and identifies directions for further research.

C. UAV Communications

  • AIRS mounts an IRS on a UAV [131], enabling stronger ground-node LoS links, 360° reflection, service to more users, and mobile coverage expansion.UAV mobility can bring AIRS closer to geographically separated users sequentially, improving communication performance.Source: C. UAV Communications · pp. 64–65
  • AIRS deployment requires joint 3D placement or trajectory, passive beamforming, and user association design, plus practical channel estimation and tracking over 3D trajectories.The designs should maximize average user rates under practical UAV-ground channel models; channel estimation and tracking are more challenging than for fixed terrestrial IRSs.Source: C. UAV Communications · p. 65

D. MmWave Communications

  • In mmWave systems, properly deployed IRSs create virtual LoS channels between BS/APs and users through smart reflections, bypassing main obstructors and efficiently resolving critical issues including blockage and higher propagation loss.Source: D. MmWave Communications · p. 65
  • IRS-aided mmWave systems need initial BS/AP and IRS beam alignment before estimation and transmission, but narrow IRS beams and multi-user scanning create high training overhead.IRS passive training must also be jointly designed with active BS/AP beam training to adapt to mobility-driven channel variation; user-by-user overhead scales linearly with user count.Source: D. MmWave Communications · pp. 65–66

E. Mobile Edge Computing

  • IRS passive beamforming can raise single-user offloading capacity while reducing latency without high transmission energy, and can redistribute multi-user loads for better resource use and lower latency [141]–[143].Improved links can allow users to offload to less-loaded, farther edge servers instead of congesting a nearby server.Source: E. Mobile Edge Computing · p. 66
  • MEC requires joint design of time-varying IRS beamforming, offloading order, and user time allocation using task arrivals, deadlines, and channel conditions.The tutorial identifies this as a direction for future investigation.Source: E. Mobile Edge Computing · pp. 66–67

F. Other Miscellaneous Topics

VII. CONCLUSION

  • The tutorial frames IRS-aided communication as a shift from active-only systems to hybrid active-passive networks and surveys fundamentals, current approaches to key challenges, and future research directions.IRS research is described as still in its infancy, with further work needed to unlock its potential in B5G/6G.Source: §VII · p. 67

Related papers

Loading 2007.02759v2…