Received: 20 June 2022 Revised: 1 November 2022 Accepted: 6 January 2023 IET Communications

DOI: 10.1049/cmu2.12571

REVIEW

A survey on reconfigurable intelligent surfaces: Wireless

communication perspective

Saber Hassouna Muhammad Ali Jamshed James Rains Jalil ur Rehman Kazim
Masood Ur Rehman Mohammad Abualhayja Lina Mohjazi Tei Jun Cui
Muhammad Ali Imran Qammer H. Abbasi

James Watt School of Engineering, University of Abstract

Glasgow, Glasgow, UK
Using reconfigurable intelligent surfaces (RISs) to improve the coverage and the data rate
of future wireless networks is a viable option. These surfaces are constituted of a significant
Correspondence
Saber Hassouna, James Watt School of Engineering, number of passive and nearly passive components that interact with incident signals in a
University of Glasgow, Glasgow, UK. smart way, such as by reflecting them, to increase the wireless system’s performance as a
Email: [email protected]
result of which the notion of a smart radio environment comes to fruition. In this survey, a
study review of RIS-assisted wireless communication is supplied starting with the principles
Funding information
Engineering and Physical Sciences Research Council, of RIS which include the hardware architecture, the control mechanisms, and the discus-
Grant/Award Number: EP/T021063/1 sions of previously held views about the channel model and pathloss; then the performance
analysis considering different performance parameters, analytical approaches and metrics
are presented to describe the RIS-assisted wireless network performance improvements.
Despite its enormous promise, RIS confronts new hurdles in integrating into wireless net-
works efficiently due to its passive nature. Consequently, the channel estimation for, both
full and nearly passive RIS and the RIS deployments are compared under various wireless
communication models and for single and multi-users. Lastly, the challenges and potential
future study areas for the RIS aided wireless communication systems are proposed.

1 INTRODUCTION 1000 Gbps and a user experience data rate of 1 Gbps. The entire
network performance must be enhanced in order to deliver
Although the evolutionary aspect of fifth-generation (5G) has advanced multimedia services to a large number of users, for
acquired substantial traction, the promised revolutionary view example, by aiming to achieve spectral efficiency that is twice
of 5G, a system running nearly entirely at millimeter wave as high as 5G. As a result, it is critical to developing sustain-
(mmWave) frequencies and enabling diverse internet of things ably new and inventive technologies to enable future wireless
(IoT) services, has remained a mirage so far [1, 2]. Although the network capacity increase at a moderate and manageable bud-
5G wireless network is still deployed around the world, both get, complexities and power consumption with the widespread
academia and industry are excited about the future beyond 5G adoption of user devices that will form the future of IoT.
(B5G) which seeks to satisfy more demanding requirements On the other hand, because of user mobility, time-varying
than 5G, such as ultra-high data rates, for example, Gigabit wireless channels are a major challenge in building ultra-
Per Second (Gbps), energy efficiency (EE), global coverage and reliable wireless communications. Traditional ways to address
connectivity, spectral efficiency (SE) as well as high reliability this problem are either by using different modulation, cod-
and low air latency [2, 3]. Figure 1 shows the vision and the ing and diversity plans to compensate for channel fading, or
expectation for the 6G [4] key performance requirements in adjusting to it using modified power, rate management, and
comparison with 5G. Compared to 5G, 6G must offer a much beamforming methods [5]. However, they require extra costs
higher data rate. While the peak data rate for 5G was intended and have a restricted amount of influence over the essen-
to be 20 Gbps, the goal for 6G is to deliver a peak data rate of tially random nature of wireless channels, making the basic

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the
original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2023 The Authors. IET Communications published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology.

IET Commun. 2023;17:497–537. wileyonlinelibrary.com/iet-com 497

498 HASSOUNA ET AL.

RIS’s aided future wireless networks have recently prompted

substantial research. In the literature, a few publications have
provided an overview of research on RIS and its variants from
various perspectives. Realistic simulations and tests have shown
the smart radio environment’s potential to increase transmis-
sion performance in a variety of wireless networks by using
the RIS’s tunability. The author in [7] described the active wall,
and how it works by manipulating the wireless environment
via an active frequency-selective surface (FSS). The intelligent
wall’s major objective is to switch the active FSS on and off
dynamically to modify its EM properties, which affects the
propagation environment and hence, the system performance.
More interestingly, in [8] the authors proved the advantages of
deploying the low-cost devices into a building’s walls to trans-
mit and reflect wireless waves actively and passively, respectively.
The paper [9] provides a novel technique for controlling the
FIGURE 1 Comparison between 5G and 6G [4] behavior of wireless environments that is predictable and pro-
grammable. The major enabler is the hypersurface tile, a novel
class of planar meta-materials that can engage with encroach-
obstacle to establishing wireless communications with high ing EM radiation in a manageable way. EM waves may be
capacity and reliability insurmountable. Moreover, signal trans- re-engineered in many ways using hypersurface tiles, including
mission is subject to reflections, diffractions, and scattering total absorption, polarisation manipulation, and more. Multiple
before arriving at the destination, resulting in a plethora of tiles are used to cover items in both indoor and outdoor con-
arbitrarily degraded and deferred extra versions of the source ditions, such as facades of the buildings and other items. To
waves along various paths owing to the unpredictability of the better meet the demands of communication devices, an external
radio environment. This channel fading change becomes a fun- software service determines and delivers the ideal interaction
damental restricting element in wireless network EE and SE types per tile. Simulations are used to evaluate the new con-
performance optimization. cept’s potential. The researchers in [10] implement and evaluate
The existing modern physical layer solutions are insufficient, many physical layer building-block solutions for a configurable
and overall progress is still modest, necessitating new and rad- wireless network and proposed the KPCONFIG method which
ical physical layer solutions. There is attracting attention in is a new way to setting the programmable wireless environ-
new communication patterns that exploit the propagation envi- ments (PWE), allowing for multi-user support and flexibility
ronment’s extreme randomness to achieve the target of the in expressing resource sharing policies and user communica-
simplicity of the transceiver components and the quality of ser- tion goals. Furthermore, it uses an accessible model based on
vice (QoS). The RIS has recently been added in the wireless well-known ray-tracing principles to illustrate the underlying
communications by the academic researchers [6]. The RIS is a physics of hypersurfaces’ unique ray-manipulation capabilities.
fundamental facilitator for achieving the concept of smart radio The incident radio frequency (RF) waves may be configured
environments (SREs) by rendering the wireless environment with the desired response by regulating the distribution of the
configurable and adjustable. current over the hypersurfaces, resulting in a reconfigurable
RIS is an inexpensive adaptable surface, such as a smart wireless environment. The programmable wireless environment
lightweight composite structure layer, that can modulate wire- can improve the performance of transmission relating to signal
less signals encroaching on it in ways that can be configured strength, communication coverage, EE, and SE performance
and modified using external inputs. As a result, one of the most by reducing signal pathloss, multi-path fading, and co-channel
important characteristics of RISs is their flexibility to be recon- interference. Rather than using hypersurfaces, the authors in
figured after being deployed in a wireless environment. RIS is a [11] proposed Scatter-MIMO, which employs a smart surface
metasurface, which is a two-dimensional (2D) electromagnetic to maximize the scattering in the surroundings to give MIMO
(EM) material surface constitutes of multiple passive scattering spatial multiplexing gain.
units. Each unit in the surface can be modified in a method Wireless networks aided by the RIS are expected to change
specified by software to change the EM characteristics of the the existing network optimization patterns by incorporating the
incident signal’s reflection on the scattering units. By heavily smart wireless environment into network optimization issues
placing RISs in wireless networks and intelligently organizing and are predicted to take a proactive role in the future wire-
their reflections, the propagating signals in wireless channels less networks [16]. There have been several recent works
between sources and destinations can be freely reconfigured reviewing the RIS based smart radio environment [3, 13] and
to obtain targeted realizations and distributions. Consequently, repeating these reviews would not do a fair study however in
they introduce a better way to essentially address the fading comparison to prior publications, our work provides a thor-
channel deficiency, interference problem and potentially gives ough investigation of the RIS’s theoretical foundations as well
a practical solution. as a current evaluation of its most recent uses in wireless
HASSOUNA ET AL. 499

from the RIS architecture and control mechanism covering,

the design and the types of tuning mechanism, circuit, geo-
metric and material. We classify them from the perspective
of research volume as per Figure 5 and by referring to dif-
ferent papers in Table 2 where the tuning method is shown
for different publications in the literature. In addition to
the channel correlation fading and practical pathloss models
which characterize the signal and channel model in RIS-aided
communication. Then, we concentrate on the performance
analysis and optimization techniques by categorizing differ-
ent papers according to their design goal, RIS functionality
and the applied algorithms with their degree of conversion
and optimality. For instance, the majority of the authors are
relying on the alternating optimization (AO) method with
sophisticated techniques and algorithms to maximize the data
rate. Consequently, the need for low complex techniques
and less elapsed time algorithms is becoming vital to jointly
optimize the passive and active transmissions.
∙ We provide channel estimation methodologies that describe
the estimation protocols and pilot transmission, which are
not clarified in the other survey papers. Consequently, more
advanced pilot transmission dedicated to aiding the RIS sys-
tems is still necessitating more exploration and investigation
in the literature. Assuming the channel state information
(CSI) availability at the transmitter and the receiver is imprac-
tical and a preceding channel estimation phase is required at
the receiver. Furthermore, different RIS deployments meth-
ods were discussed to ensure that such passive RIS surfaces
are deployed in a different way from the active networks that
contain active nodes and relays.
∙ In the literature, the assumption of lossless amplitude and
continuous phase yields encouraging analytical and simu-
lation results. However, in actual hardware, energy loss is
inescapable, and a practical phase shift model should take
these losses into account.
∙ The shortcomings and the overoptimistic conclusions are
discussed thoroughly in this review paper. For example, much
research is considering certain system assumptions and drop-
ping or ignoring other parameters that are vital for practical
wireless systems, for example, mutual coupling, time-variant
channels, Doppler shifts & mobility, discrete phase shifts
models and electromagnetic interference are still needing
more extended efforts and future exploration to enhance the
research quality.

Table 1 illustrates the comparison of this work with the

existing magazines/surveys/tutorials in the context of RIS.
FIGURE 2 Structure of the paper Compared with other surveys, we strive to reveal some impor-
tant gaps which are not fully covered in the literature. The
tuning mechanism are further explored to reflect the key feature
networks. The following are the highlights of our important in the RIS from communication perspective. The integration
contributions: between the backscatter communication and the RIS is impor-
tant to be more investigated to reap more rewards of power
∙ An extensive study of the recent RIS works (as shown in consumption systems and reduced power IOT technologies.
Figure 2) including some areas of research that are not fully Furthermore, new additions, such as pilot transmission and
covered in the prior review publications in order to help the channel estimation for wideband communication are essen-
readers to grasp the RIS technology more deeply. We start tial to look for refined communication models that leverage
500 HASSOUNA ET AL.

TABLE 1 Comparison with other surveys/tutorials researches

Ref. Description Comparison to Our Work

[12] particularly interested in RIS operating principles, performance Different from other surveys and overviews, our paper presents crucial
[3] assessment, beamforming design, and resource management, as well as and critical technical aspects of reconfigurable intelligent surfaces
[13] the integration of RISs with other developing technologies including the physical operational principle and channel model. It is
[14] seeks to give an in-depth technical explanation to help and inspire future then evaluates the performance metrics and optimization techniques
[15] research in RIS-assisted wireless networks modelling, analysis, design, in the literature taking into account the limitations, shortcomings and
optimization, and implementation. impractical assumptions that lead to overoptimistic results. The pilot
Concentrates on the applications of the RIS in wireless communications transmission, channel estimation and different deployments methods
and look at several performance metrics and analytical methodologies. for RIS are all topics that still need more investigations to the best of
provides an overview of the technical and crucial features of mathematical our knowledge. An outlook for the lessons learnt and summary is
optimization and performance analysis of LIS systems, as well as a few given at the end of each section. Finally, a proposal for future
potential research paths for the formulation of real challenges beyond directions is highlighted to include the RIS applications not only
5G systems in the future. Outlines Capacity/data rate assessments, from the channel and performance perspective but also when
power/spectral optimizations, channel prediction, deep learning-based involving the RIS in digital information world.
design, and reliability analysis.

the electromagnetic properties to comply with realistic wireless frequency. The RIS is made up of a programmable metasurface
communication applications. Moreover, lower computational that can completely regulate the phase changes that individual
complexity RIS-based algorithms that consider reduced num- scattering components experience. This can be accomplished by
ber of operations and less elapsed run time are still requiring applying outside stimulus to the scattering components, causing
more investigations and thorough scrutiny their physical characteristics to alter, resulting in a change in the
The remainder of this paper is structured as set out in metasurface’s EM properties without refabrication [18].
Figure 2. The RIS hardware and functionality, including signal Figure 3a depicts a typical RIS design, which includes three
and channel models, hardware design, and control mechanisms layers and a smart controller. The first layer (RIS layer) is made
are covered in Section 2. Section 3 considers the performance up of a dielectric substrate with several tunable and reconfig-
analysis and optimization of RIS-assited wireless systems. We urable metallic patches put on it to directly regulate incoming
overview promising approaches for RIS channel estimates in waves. A copper substance is typically used in the second layer
Section 4, which apply to a variety of RIS topologies and com- to avoid transmission power losses due to RIS reflection. The
munication settings. In Section 5, we look at how to deploy RISs third layer is a control integrated board that is in charge of
at both the link and network level. We propose challenges and both excitation and real-time control of the reflecting elements’
research direction for the future in Section 6. Finally, Section 7 reflection amplitudes and phase shifts. Moreover, a smart con-
brings this survey paper to a conclusion. troller linked to each RIS also activates and decides reflection
adaptation, which may be done with a field programmable gate
array (FPGA). Other network components such as base stations
2 RIS HARDWARE AND (BSs), and user terminals can link to each other over wired or
FUNCTIONALITY wireless backhaul and control lines thanks to the RIS controller’s
role as a gateway. In practice, dedicated sensors can be deployed
In this section, we cover the fundamentals of RIS-assisted in the first layer, for example, interlaced with the RIS’s reflect-
wireless communication, including the major RIS architecture, ing elements, to detect the surrounding radio signals of interest
hardware, and control mechanism, as well as the signal and and assist the smart controller in designing the reflection coef-
channel models presented in the existing works of literature. ficients, to enhance RIS’s environmental learning capability
[3]. There are three basic categories for the different tuning
processes that have been proven in the literature namely:
2.1 RIS architecture and control mechanism
1) Circuit tuning comprises the integration or modification of
Snell’s law and the Fresnel equations control the intensities and individual impedance into the unit cell circuit model using
directions of reflected and diffracted waves [17]. When the wave changeable capacitors and switches inside and between unit
collides with a metasurface, the situation changes. A shifting cells.
in the resonance frequency and, as a result, changes in the 2) Geometric tuning refers to techniques that change the form
boundary conditions might emerge from the periodic arrange- of the unit cell physically, causing the accompanying circuit
ment of the scattering components. Hence, extra phase shifts model to change dramatically.
will be carried by the reflected and diffracted waves. The EM 3) Material tuning is the process of modifying the material
characteristics of the metasurface will be fixed once it is pro- properties of a substrate or small section of a unit cell
duced with a certain physical structure, allowing it to be utilized to change the responsiveness and characteristics of the
for a given aim, such as a ideal absorber working at a specific substrate layer or small component of the unit cell.
HASSOUNA ET AL. 501

TABLE 2 List of publications related to RIS from the smart radio environment (SRE) perspective

Reference Surface architecture Control Mechanism System Setup Achievement

[7] Active frequency selective surfaces ON-OFF PIN Diodes Multi-user wideband indoor Surfaces that are fully reflected with
(FSS) with PIN diodes connecting downlink OFDMA system proper coverage and can boost
metal parts of the FSS system performance by up to 80%
[8] Programmable Radio Environment prototype PRESS elements Multi-client’s wideband system Passively reflect or actively transmit
for Smart Spaces (PRESS) equipped with (SP4T) RF radio waves, and so attenuate or
Low-Cost antenna elements switches change phase of enhance signal strength by up to 26
connected to passive loads and each antenna by 𝜋∕2 dB, to reconfigure multipath
embedded in the walls of a building propagation
[9] Hypersurface tile equipped with Switch element Controllable 12 receivers, in both microwave Re-engineering EM waves, including
physical switch elements state (ON/OFF) and mmWave frequency steering in any direction, complete
bands, are uniformly absorption, polarisation
distributed in indoor space modification, and other techniques.
and are evaluated using a With maximum and minimum
map-based ray-tracer received power of 32.5 dBm and
12.4 dBm, respectively, and an
average received power of 20.6
dBm, the results demonstrate Good
Coverage.
[25] Spatial microwave modulators (SMM) Two states of resonant elements Two antennas source and Increasing or cancelling the wireless
equipped with102 controllable EM (the reflector and the parasitic receiver connected to transmission amplitude between
reflectors strip), 𝜋 state and 0 state network analyzer are located two antennas (Shaping complex
in a room that the spatial microwave field). SMM can perform
microwave modulator can be wave front shaping and concealing
placed on the walls of the the field around one single antenna
room on a correlation length wide area ( 6
cm at 2.4 GHz)
[26] Reflect-array panel with totally 48 each reflector is controlled by a Two pairs of wireless users in a Controlling the phase shift of each
reflector units and its peripheral bias voltage to tune the conference room where reflect-array element. The
circuits and varactors varactors (0.6 − 8pF) for smart reflect array hung on interference has been eliminated,
changing the capacitance and the walls and the interference-plus-noise ratio
hence the phase of each unit (SINR) has been enhanced to
around 30 dB, according to the
achieved results
[27] Intelligent receiving antenna array the information transfer Multi-user narrow band system Active surface for transmission and
capabilities of an intelligent with ideal free space reception. Consequently, the limit of
surface for every m2 deployed propagation the normalized capacity is enhanced
surface area when the wavelength approaches
zero
[28] Hypersurface tile with controllers that Dynamic meta-atoms include mm-wave setups that include a New physical layer security features
regulate the metasurface’s switch phase switching components Rx-Tx pair situated in can help avoid eavesdropping.
components like MEMS, CMOS non-line of sight (NLOS) Pathloss and multipath fading
transistors, or microfluidic over a defined floorplan and mitigation, as well as eavesdropping
switches that can change the walls covered with security, were proven in the 2.4 and
structure of the meta-atom hypersurface 60 GHz configurations
[29] Plasmonic antenna elements at each New intelligent plasmonic Ultra-Massive MIMO (UM In the mm-wave and THz-bands, new
transceiver side antenna arrays that can MIMO) intelligent plasmonic antenna arrays
function in transmission, capable of communications and
reception, reflection, and waveguiding have been developed.
waveguiding, the mm-wave The results demonstrate a
and THz-bands significant increase in transmission
distance and data rate
[30] RIS-assisted free-space optical (FSO) Comparable mirror-assisted A FSO communication system FSO systems with IRS assistance can
systems technology, which can be consists of a Tx with a compensate the need for a line of
used to create a phase-shift Gaussian beam-emitting laser sight (LoS) between Tx and Rx. The
profile that spans the IRS source (LS), an IRS, and a Rx effect on the end-to-end channel
with a lens and a photo varies depending on where TX, RIS,
detector (PD) and RX are in relation to each other

(Continues)
502 HASSOUNA ET AL.

TABLE 2 (Continued)

Reference Surface architecture Control Mechanism System Setup Achievement

[31] 102 phase-binary components make The phase shift of the reflected The transfer of an RGB colour The benefit of shaping wireless
up the metasurface wave may be electrically image across a 3-3 MIMO channels. Physical shaping of
controlled for each element system was simulated using propagation media with simple
using a PIN Diode bias wireless image transmission metasurfaces may achieve complete
voltage from an Arduino in an office room orthogonality of wireless channels
microcontroller to be either 0 and excellent channel diversity and
or 𝜋 low crosstalk
[32] RIS with 16 elements A method of encoding SIMO over a quasi-static fading To boost capacity, a method is utilised
information in both the sent channel that encodes data in the sent signal
signal and the RIS as well as the RIS configuration.
configuration Three times quicker than max-SNR
encoding is the joint encoding.
[33] RIS with large reflecting elements The best RIS phase shift Multiple antennas at the Developing an overhead-aware
configuration transmitter and receiver in a resource allocation framework
point-to-point RIS-based where RIS used to improve the
system performance SE/EE of the system
[34] 256 unit cell programmable surfaces A digital to analogue converter RIS-assisted MIMO wireless The proposed prototype implements
based on varactor diodes generates an external control system real-time RIS based MIMO-QAM
signal that controls the phase wireless communication with less
response of the unit cell power consumption and achievable
data rate 20 Mbps

2.1.1 Circuit tuning we can calculate the discrete amplitude and phase-shift values,
respectively. Even though phase-shift control or phase beam-
The EM behaviour of actual, passive transmission lines, anten- forming can achieve better passive beamforming performance
nas, and metamaterials may be modelled as a lumped inductive, than amplitude control or amplitude beamforming, phase-shift
capacitive, and resistive equivalent circuit. This way of breaking control or phase beamforming for RIS is more expensive to
down complex geometric shapes into a known circuit model is construct when the number of control bits and discrete levels
highly useful for predicting how updated designs would behave. for each reflective element are the same. The authors in [24] pro-
Metamaterial circuit tuning is described as methods for intro- posed a practical reflection model by simulating every reflective
ducing, altering, and controlling specific components in the component as a resonant circuit with specific inductance, capac-
metamaterial’s equivalent circuit. Due to their ease of combina- itance, and resistance values and depending on his prototype, it
tion into a variety of metamaterials, varactor diodes are the most was discovered that the reflecting element’s amplitude response
often used tuning method [19]. The incorporation of a varactor and phase shift are in general non-linearly linked, and hence are
in a metamaterial design is frequently referred to as active meta- not separately controllable. The reflection amplitude achieves
material despite the fact that the device remains inactive for RFs a minimum value at zero phase shift, as shown in Figure 4,
and is only active in its desire for a DC bias [20]. but rises uniformly as the phase shift approaches 180 or -180,
The varactors in most cases have been considered as perfect asymptotically approaching one.
or nearly ideal linear capacitors with good accuracy, although
numerous researches have looked at the nonlinearity aspect of
the varactor at various power levels [21]. In terms of applicabil- 2.1.2 Geometrical tuning
ity and simplicity of integration, the use of PIN diodes [22] in
a metamaterial is comparable to the use of varactors, although Many metamaterials depend on conductive components that
these actuators influence resistance rather than capacitance. In may combine with encroaching EM signals to produce the
practice, independent amplitude and phase shift control of every required electric or magnetic resonance or other beneficial
RIS element are preferable for superior reflection design, but behaviour. Because metamaterial characteristics are generally
this needs more intricate architectural patterns and design ele- influenced by the form, size, direction, and closeness of
ments [23] than those listed above for their independent control conducting components, methods that change the geometric
alone. While continually adjusting the reflection coefficient is features of the conductive elements can be a powerful tool for
advantageous for improving communication performance, it is adjusting or switching metamaterial response. By shifting con-
challenging to put into practice because better quality reflective ducting components in respect to each other, metamaterials
components need not only more expensive but also more com- may be geometrically adjusted. Micro electromechanical systems
plicated hardware architecture. For instance, at least log2 (8) = 3 (MEMS) are often used in THz metamaterials to accomplish
PIN diodes are required to allow 8 levels of phase changes per the mechanical movement of conducting components. The cou-
RIS unit. By properly quantizing the intervals [0,1] and [0, 2𝜋], pling between conducting components varies when they are
HASSOUNA ET AL. 503

FIGURE 3 (a) the structure of the RIS including its reflecting element and the equivalent RLC circuit model and (b) the amplitude and phase responses for
different elements and their corresponding capacitance values.

pushed closer or farther away, resulting in variations in res- 2.1.3 Material tuning
onance frequency or resonance strength. Moving conducting
elements can also alter the shape of the element. Stretching While changing the structure of resonant components provides
the substrate to further separate the elements on the substrate, for a number of tuning possibilities, the metamaterial’s proper-
resulting in significant variations in the resonance frequency ties are ultimately determined by the constituent materials used
of the elements, is a unique approach for moving conducting to create the unit cell. Various constituent materials have been
components in a metamaterial. Several tunable high impedance examined and used in the literature for tuning metamaterials by
surfaces (HIS) have been shown to alter the phase of the changing the permittivity, permeability, and conductivity of unit
reflected wave by mechanically sliding an upper plate of ele- cell sections. Several candidate materials, such as Ba0.5 Sr0.5 TiO3
ments along the surface or vertically [19]. Geometrical tuning (BST) ferroelectric films, liquid crystal, and Ga-Sb-Te (GST)
may result in substantial variations in metamaterial features phase-change materials, have been used to tune the permittivity
since the geometry of the conducting elements has such a large of metamaterials. More details about the material tuning subject
influence on the related resonant frequency. Geometrical tun- can be found in [19, 35, 36].
ing, on the other hand, is difficult to execute since it necessitates The purpose of presenting the three tuning approaches
a physical control mechanism. in this review study is to mention that due to its rapid
504 HASSOUNA ET AL.

1 rooms, or carried on aerial vehicles, such as not fixed balloons

in the air and UAVs [37], to achieve the idea of the smart
radio environment. As a result, for modelling and performance
0.8
analysis of RIS-aided wireless communication, scientific anal-
ysis models that consider the geographical placements of RIS
Amplitude

0.6 elements, the RIS’s EM characteristics, and the wave modifi-

cations utilized by adjacent RIS elements in the environment
are necessary. A transmitted radio signal in a typical wireless
0.4 communication environment contacts many objects along the
route, resulting in duplicates of the transmitted wave which
=1
=2
comes across reflection, diffraction, and dispersion. Multipath
0.2
=2.5 components are signal copies that reach the receiver with ran-
domly and unexpectedly different amplitudes, phase shifts, and
- - /2 0 /2
Phase
signal delays, causing considerable distortions in the received
signal due to their relative constructive or destructive addition.
FIGURE 4 Reflected amplitude vs. phase shift for RIS element This is termed as fading in wireless communication systems,
and it is a critical parameter in existing and future wireless
communication systems.
The basic goal of RIS is to establish a controllable wire-
less communication in which the extremely unpredictable radio
channel is turned into a controllable space by carefully modi-
fying EM signal propagation in a software-controlled manner.
A channel model that captures the key characteristics of any
wireless technology is required for an accurate performance
assessment. The most theoretical work on wireless systems
technologies in scattering conditions has been and continues to
be based on the independent and identically (IDD) distributed
Rayleigh fading channel model [39–41]. When using a rectan-
gular RIS, the authors in [38] shows that such a paradigm does
not exist realistically, and present an equivalent physically valid
Rayleigh fading model that may be used as a reference for eval-
uating RIS-assisted communications. The received signal z ∈ ℂ
FIGURE 5 Tuning mechanism comparison can be represented as in literature:
( )
z = V T 𝜃G + hd x + w, (1)
reaction time, minimal reflection loss, low power consumption
and cheap equipment price, the first technique (circuit tuning) where, x is the transmitted signal with power p = 𝔼{|x|2 }
has been widely used in actual applications and implementa- and w ∼ ℂ (0, 𝜎2 ) is the noise variance. The configuration
tions. Figure 3a depicts an example of a reflecting element and of the RIS is given by the diagonal matrix 𝜃 = ΓΦ with Φ =
its corresponding RLC circuit based on a PIN diode installed diag(e− j ∅1 , … , e− j ∅N ) and Γ = diag(𝛾1 , … , 𝛾N ). The direct path
in the centre of the element. Such illustrations of RLC circuit hd ∈ ℂ has a Rayleigh fading distribution. However, the fading
equivalence have been used widely in the literature. Table 2 distribution of the channels V ∈ ℂN and G ∈ ℂN is inde-
provides an overview of the most updated and recent con- pendent and distributed as V ∼ ℂ (0, ADV CV ) and G ∼
cepts and realizations for a smart radio environment taking ℂ (0, ADG CG ) where A is the area of the RIS element, D
into account essential parameters such as surface architecture, is the average attenuation intensity and C is the spatial corre-
control mechanisms, system setup and ending with the out- lation matrix. Figure 6 shows that non of the cases dH = dV =
comes and work achievements. We noticed that high volume d ∈ {𝜆∕8, 𝜆∕4, 𝜆∕2} resemble The I.I.D. Rayleigh fading. More
of researches depend on the circuit tuning more than the other details can be found in [38]
two tuning methods which coincides with Figure 5 survey.

2.2.2 Pathloss
2.2 Signal and channel model
The authors [42] evaluated the two-ray channel model as per
2.2.1 Channel fading Figure 7. In the most perfect propagation scenario, with no user
movement and no unanticipated environmental consequences,
RIS can be coated on the front of buildings in the wire- a single uncontrolled ground reflection might cause substan-
less environment, such as solid structures and top surfaces of tial signal degradation. The authors assumes that a changeable
HASSOUNA ET AL. 505

-50

-100

-150

-200
-80 -60 -40 -20 0 20 40 60 80

FIGURE 6 The eigenvalues of R in decreasing order for an RIS with N

FIGURE 8 Pathloss of the reflected path. Refer to [45] for simulations
= 1600 and dH = dV = d ∈ {𝜆∕8, 𝜆∕4, 𝜆∕2. please, refer to [38] for simulation
parameters
parameters

By researching the physics and EM characteristics of RISs,

the author in [43] develops free-space pathloss models for RIS-
assisted wireless communications for various circumstances
which can be divided into two groups: RIS-assisted beamform-
ing and RIS-assisted broadcasting. The proposed models reveal
the relations between the free-space pathloss of RIS and dis-
tances from both the transmitter and receiver to the RIS. To
explain the free-space pathloss of RIS-assisted beamforming
and broadcasting, three perceptive free-space pathloss models
were developed: the far and near fields beamforming equa-
tions and the near-field broadcasting equation. The authors in
[44] improved the pathloss models derived in [43] by formulat-
ing joint radiation pattern of antennas and unit cells in addition
to clarifying the relation between the scattering gain of the unit
cell and its size.
The authors in [45] calculates the far-field pathloss using
physical optics methods and explain why the surface consists
of multiple reconfigurable elements that individually behaves as
diffuse scatterers but may collectively beamform the signal in a
desirable direction with a certain beamwidth. As a result, an RIS
FIGURE 7 Illustration of the pathloss of Two Rays’ Propagation Model can be thought of as an array of sub-𝜆-sized diffuse scatterers
that aiming the phase of their reflected signals at the receiver,
resulting in anomalous reflection. Figure 8 depicts the pathloss
metasurface covers the whole ground. The RIS may be thought as a function of observation angle for various RIS sizes. The
of as a perfect phase shifter, designed to change the reflected main beamwidth narrows as the RIS surface area grows. The
signal phase such that the LOS and reflected radiations add RIS virtually functions as a diffuse scatterer when the dimension
up constructively, increasing signal strength. It has been shown is sub-wavelength (less than 𝜆∕2). The authors proved that the
that the usage of programmable intelligent surfaces has the pos- power of the received signal is proportional to the square of the
sibility to change the scaling rule that regulates the received RIS area and to 1∕(r1 r2 )2 , where r1 is the distance between the
signal power with distance. Despite the results concluded by [42] transmitter and the RIS and r2 is the distance between the RIS
are promising, they are not realistic and practical because they and the receiver. This refutes [42] hypothesis that the received
were based on several assumptions including the ability to tune power is proportional to 1∕(r1 + r2 )2 . To compensate for the
the reflected phases without any analogue to digital conversion enormous power loss caused by twofold attenuation, a signifi-
error and for any incident and scattered angle, as well as the lack cant number of RIS reflecting components must be employed to
of reflection impairments and comprehensive awareness of the combine their reflection magnitude and phases to obtain higher
phases status at the RIS. Furthermore, for more actual system passive beamforming gains. RIS is made up of several sub-
models, optimizing the phases is typically not a simple process. wavelength components that scatter incident waves with distinct
506 HASSOUNA ET AL.

phase shifts in order to perform constructive beamforming in a critical for link budget analysis and performance assessment
certain direction. of RIS-assisted wireless systems.
Using the scalar theory of diffraction and the Huygens-
Fresnel principle, the authors in [46] describe pathloss in RISs
in both the near and far fields. RISs are represented as homoge-
nous sheets of EM material with negligible depth. The authors
3 PERFORMANCE ANALYSIS AND
OPTIMIZATION
reveal the regimes in which the pathloss relies on the summa-
tion and multiplication of the distances between the RIS and
This section is divided into two parts. In the first part, we
the source, as well as the RIS and the destination, using the
present the performance analysis of the RIS using different per-
stationary phase technique. The analytical technique provided
formance metrics like bit error probability and sum rate for
is proven to be sufficiently generic for use with a consistent
evaluating the overall behaviour of the wireless systems aided
reflecting surface, anomalously behaving reflectors, and lenses
RIS and under various channels and environments. Further-
with highly concentrating and reflecting features.
more, the second part will concentrate on the RIS reflection
The authors in [47] have provided a viewpoint that unites
optimization techniques and algorithms.
RIS opposing behaviour as a scatterer and as a mirror. It
have been proven that the RIS may be seen as a zero, one,
or two-dimensional object, depending on its size and distance
and its radiated power exhibits a dependency on the fourth, 3.1 Performance analysis
third, or second power of the distance, respectively. In addi-
tion, the Fresnel zone decomposition is used to gain a better In this study, we will present the RIS advantages and perfor-
understanding of how the various variables interact. More pre- mance comparisons with other wireless systems. furthermore,
cisely, the importance of phase in determining the eventual we show the behaviour and the usage of the RIS as reflector,
pathloss exponent is discovered and demonstrated how free receiver and transmitter taking into consideration the perfor-
space propagation may be outperformed via smart dephasing. mance metrics proposed in the literature for RIS supported
The findings are calculated numerically, and the received signal wireless communication systems like coverage or outage prob-
in terms of distance has no clear analytical meaning. [48] reports ability, bit error probability, Ergodic capacity, and achievable
similar observations. data rate.

2.3 Discussions and insightful prospect for 3.1.1 RISs versus relay networks and random
section II phase surfaces

∙ Figure 5 shows the volume of research (paper publications) We first present the following unique advantages of RIS-assisted
using the three approaches of tuning however in this paper wireless communications:
we will not list the publications [19] to avoid repeating what
other authors have done and worth mentioning that the rea- ∙ Densely deployed and sustainable operation: If RISs
sons stated above for the more widespread usage of the first are formed of smart metasurfaces, there are a lot of sub-
approach is not enough and more research still has to be con- wavelength unit cells in them. Such sub-wavelength heavy
ducting on the other two approaches to be parallel with the installations of small sub-wavelength dispersing devices are
circuit tuning method. not typically used in radio communications, where mutual
∙ We have noticed a disagreement between the authors in [45] coupling among the reflecting elements is frequently pre-
and the authors in [42] regarding the reflected power in the vented by design, by confirming that the scattering units are
far-field region as a function of the distance between the sufficiently far apart. This paves the way for the develop-
transmitter to the RIS and the RIS to the receiver while other ment of novel wave and propagation scenarios that may have
researches are in agreement with [45]. Moreover, the cor- an impact on wireless networks’ ultimate performance lim-
relation fading revealed in [38] clarifies that using the idd its, as well as the introduction of new scenarios in design in
fading channel is not encouraged to be modelled in the RIS which wireless systems are built to be mutual coupling fully
wireless systems analysis which opens the door for reconsid- cognizant. RIS is easily applied and removed from a variety
ering previous works depending on the IDD fading models. of surfaces, including front of buildings, interior walls, and
Table 3 offer a comparison for different pathloss models in top ceilings. The RIS can be battery-free and remotely pow-
the existing literature. The purpose of narrating the previous ered thanks to RF-based energy harvesting, which eliminates
work is to investigate the pathloss derivations based on strong the need for active equipments that require signal processing
foundation presenting all the scenarios bearing in mind that methods and power consumption.
pathloss due to distance, large and small scale shadowing and ∙ New signal processing is not required for RIS-aided
multipath fading are all factors that affect channel coeffi- communication: The semi-passive feature of RISs opens
cients. The pathloss of an RIS reflected channel, in particular, new possibilities for re-defining communication, allowing
represents the average power of the channel and is therefore data to be transferred without the use of EM waves, instead
HASSOUNA ET AL. 507

TABLE 3 Pathloss comparison for the existing literature

Reference Concept Field Goal Notes

[42] LoS and ground-reflected rays in a Far-field regime Beamforming The received signal power is
two-ray propagation model proportional to the square of the
RIS area N 2 and inversely
proportional to the square sum of
the distances between the TX and
the RIS and the RIS and the RX
1∕(r1 + r2 )2
[43] The physics and the EM nature of the Far and near field regime Beamforming and broadcasting The received signal power is
RIS proportional to the square of the
RIS area and inversely proportional
to the square product 1∕d 12 d 22 of
the distances between transmitter
and RIS and between RIS and
receiver
[45] Physical optics techniques and Far-field regime Beamforming The received signal power is
Antenna Theory proportional to the square of the
RIS area and inversely proportional
to the square product 1∕(r1 r2 )2 of
the distances between transmitter
and RIS and between RIS and
receiver
[46] The Huygens-Fresnel concept and Far and near field regime Anomalous mirrors and RISs act like anomalous mirrors in the
generalized scalar diffraction theory scatterers short distance domain while they
behave as scatterers in the
long-distance regime.
[47] The Fresnel zone decomposition Far and near field regime Anomalous mirrors and The RIS can be seen as a zero, one, or
scatterers two-dimensional object, depending
on its size and distance, and its
radiated power exhibits a
dependency with the fourth, third,
or second power of the distance,
accordingly.
[49] Metallic reflectors using Near field regime Anomalous mirror Over the same link distance, the
measurements, analytical reflected received power is the same
expressions, and ray-tracing as the LoS free space received
simulations power for millimetre wave
communications

of reprocessing existing EM signals. This may be extremely ∙ Flexible reconfiguration and enhanced capacity: The
useful in terms of minimizing EM pollution and lowering RIS may be used to configure the wireless channel to provide
human EM exposure, which is often raised by deploying a larger link capacity while consuming less power for point-
more network equipment and utilizing more spectra. This to-point communications. When the RIS is used, interference
might be crucial for the effective installations of wireless reduction becomes more effective, resulting in improved sig-
technology in areas that are vulnerable to EM fields (e.g. in nal performance for end users at the cell’s edge. Scattering
hospitals). components in multi-user cellular networks can be sepa-
∙ Highly focusing capabilities: This high focusing capabil- rated and shared to optimise data transfer for multiple users.
ity could be used for a variety of purposes, including firstly, As a consequence, the RIS-assisted wireless network may
it enables interference-free communication in densely popu- be able to improve QoS provisioning as well as sum-rate
lated areas, secondly, enabling accurate radio identifications performance or max-min fairness amongst users.
of users and environment modelling, and lastly, fill the bat- ∙ Investigating of new wireless application: The RIS’s
teries of limited power equipment by means of transfer the advancement is likely to open the door for new and excit-
power wirelessly. The intelligent wireless wall (IWW) is a real- ing research avenues. The RIS, for example, was recently
world example of this high focusing characteristic. The IWW presented as a unique technique for avoiding wireless eaves-
consists of a reconfigurable intelligent surface that does beam dropping assaults by regulating the transfer at the source
steering and beamforming, as well as machine learning algo- and the optimized reflections at the RIS at the same time.
rithms that can accurately and automatically identify human The achievable secrecy rate is greatly enhanced by deploying
activity [50]. the RISs close to the legitimate or eavesdropping user and
508 HASSOUNA ET AL.

appropriately configuring the RIS passive beamforming to mode, whereas relays work in DF mode. In comparison to AF
raise or lower the achievable rate of the legitimate or eaves- repeaters, DF relays, as per Figure 9c, can provide superior noise
dropping user [51]. Despite the said feature in RIS-aided immunity and intercell interference mitigation. The DF relay, on
secrecy communication, the channel state information CSI the other hand, necessitates a complex transceiver and might
is still needed between the AP and eavesdroppers as well add to the transmission latency [54, 55]. In RIS-assisted and
as between the RIS and the eavesdropper. The challenge is AF/DF relay-assisted communications, the receiver decodes
obvious when the eavesdroppers intentionally continue to be the source information symbols, but in wireless backscatter
covert, secret or hidden due to the fact that their CSI link can- communications, it seeks to decode the piggybacked informa-
not be estimated properly from their signal leakage, and this tion symbols from the strong interference signal. Backscatter,
thus requires new channel estimation methods and robust as per Figure 9b, reflects an incoming RF signal while also
RIS beamforming taking into account the imperfect CSI of modifying and modulating it for secondary transmission, or
the eavesdropper [52]. Furthermore, in large-scale secrecy backscatter. There is no need to deploy and maintain sepa-
wireless communication networks with thousands of users rate RF sources because already-available RF sources are used,
whether legitimate or eavesdropper as well as highly dense resulting in cost and power savings. Impedance mismatching
RIS deployments, RIS is a key to increase the network secu- is the fundamental concept behind altering and reflecting RF
rity throughput and enhance the physical layer security for signals [56]. It uses the impedance of an antenna to encode
future 6G wireless modern systems where 1000x , as per data into previously existing waves, however it has low data rate
Figure 1 , increase in the data rates yielding a target of 1 Ter- transmission speeds and lacks data security. Massive backscat-
abit/sec is required. Consequently, meeting these challenges ter communication [57] is a new idea that uses a programmable
of channel estimations and robust RIS beamforming in 6G metasurface to alter the propagation environment of stray
massive networks deserve further investigations. Many other ambient waves. The metasurface’s huge aperture and many
developing research fields, including wireless power transfer, degrees of freedom allow for exceptional signal control and,
UAV communications, and MEC take the advantages of RIS as a result, safe and high-speed data transmission. The pro-
technology. posed backscatter wireless communication strategy in which the
transmitter depends on a programmable metasurface to mod-
In order to assess the advantages of RISs technology, it ulate the propagation environment rather than a single or a
should be compared with different types of relay networks and few impedance-modulated dipole antennas open the door for
surfaces which are not coated with RIS. The SNR for RIS and significantly larger control over the wave. More interestingly,
DF relay as per Figure 9 can be represented as follows: Ambient backscatter communication (AmBC) was designed
to solve communication and power consumption concerns in
p |
h + V T 𝜃G || ,
2
SNRRIS = (2) indoor and limited power IoT technologies. The authors in [58]
BN0 | d presents a novel Ambient backscatter communication approach
( ) in the frequency domain using ambient OFDM subcarriers in
p1 G p1 hd p2V
SNRDF = min , + 2 , (3) combination with the RIS. The higher performance in terms of
𝜎2 𝜎2 𝜎
BER and data rate is demonstrated by analytical and numerical
where, p is the transmit power, B is the bandwidth and N0 is analyses.
the noise power spectral density. Consequently, the spectral effi- The comparison between the RIS and relay assisted net-
ciency  can be calculated for RIS and relay supported network works has been studied in the literature. The power transmission
as follows: needed for achieving a certain rate has been studied in [53, 59]
and a comparison between the RIS and the DF relay was inves-
( )
RIS ∕DF = log2 1 + SNRRIS ∕DF . (4) tigated. Figure 10 shows that the transmit power required in the
RIS case reduces as the number of elements grows, and the dis-
Figure 9a shows a typical RIS-assisted wireless communica- tance to the DF relaying scenario is lowest when the receiver is
tion system model. An RIS controller is used to program the either near to the transmitter or the RIS. When the distance is
RIS reflecting elements. Furthermore, the controller communi- 80 (Mtrs), the RIS must have around or fewer than 100 elements
cates with the BS by another wireless signal in order for the BS to outperform DF relaying, demonstrating the RIS importance
to control the RIS reflections by creating a phase shift matrix technology in future wireless generations. The data rate of RIS
𝜃 that results from modifying huge cheap passive reflecting ele- and relay are compared as a function of distance in Figure 11
ments to configure the channel, and thus the concept of passive [59]. The RIS, for example, is setup to work as an adjusting lens
signal reflections is introduced in the research. As per Figure 9d, as well as an irregular reflector. The diagram indicates that an
AF repeater simply amplifies and sends the received RF signal RIS may achieve a rate comparable to that of an ideal FD relay
,including noise, to the users located in a dead spot By introduc- without the use of a power amplifier. This is possible due to
ing amplifying channel coefficient 𝛽. Repeaters are commonly the RIS’s size effective length. For distances up to 25-50 (Mtrs),
employed in places where signal coverage is a problem and to the RIS under analysis acts as an anomalous mirror, and for
extend the cell coverage however, it has a drawback of amplify- distances larger than 75-100 (Mtrs), it behaves as a diffuse scat-
ing the noise as well which by return will degrade the received terer. Figure 11 further illustrates that over long transmission
signal to interference and noise ratio. Repeaters work in the AF lengths ,more than 150 (Mtrs) in the case study, an ideal FD relay
HASSOUNA ET AL. 509

FIGURE 9 Comparison of Different Wireless Systems. (a) RIS assisted wireless communication,(b) wirless backscatter communication, (c) Decode and
forward (DF) Relay assited wireless communication, (d) Amplify and Forward (AF) Repeater Assisted wireless communication. G, hd , and V𝜃 represent the BS-RIS,
BS-user, and RIS-user channels, respectively

beats a RIS. To outperform a perfect FD relay across extended generalised rules of reflection. The analytical findings reveal that
transmission distances, a bigger RIS may be required. the length of a randomly positioned reflector has no effect on
the possibility of reflecting. This approach, however, implies
that all RISs have the same length, which does not reflect the
3.1.2 Measurable metric performance of RISs: real-world network situation. Furthermore, the authors focus
The state-of-the art just on reflection likelihood, with no assessment of how the
large-scale RIS can increase transmission performance.
The authors in [60], with the help of RISs, has proposed The researchers in [62–66] studied the outage probability of
new models for MIMO systems. The schemes of vertical Bell RIS-assisted wireless systems. The first authors proposed that
Labs Layered Space Time (VBLAST) and Alamouti have been The RIS could be used to increase the LoS likelihood for indoor
presented. Both schemes have shown improved BER per- mm-Wave setups. The authors derive a formula for the outage
formance gains and spectral efficiency and do not require probability and then optimizes the RIS’s deployment position to
significant changes to existing MIMO models, particularly in further reduce the outage probability while the second authors
their receiver design, making them workable and viable options mentioned that for non-LoS components, the outage perfor-
for future wireless communications. However, the authors did mance is initially assessed and optimized in the slow fading
not consider a workable phase-shift setup, which take into scenario. The optimal outage probability declines with the size
considerations phase-dependent amplitude differences, in the of the RIS when the LoS components are bigger than the non-
model of the suggested schemes. LoS components. The authors next describe the asymptotically
The authors in [61] determined the precise probability that a ideal outage probability in the high SNR zone, showing that it
randomly located RIS may reflect for a certain transceiver using decreases as the LoS component powers grow.
510 HASSOUNA ET AL.

SISO full duplex two-way communication systems which charac-

RIS (N=25) terizes the performance of overcoming transmitted data loss
25 RIS (N=50) induced by extended deep fades. The authors computed the
RIS (N=100) probability distribution of the cascaded end-to-end equivalent
20 RIS (N=150)
channel using an RIS beamformer of his choice. The number of
DF Relay
reflecting components has a considerable influence on system
15 reliability, according to both authors [65, 66]. A shortcoming of
their work is that the analysis was shown under the assumption
10 of continuous amplitude and phase shifts.
For calculating the BER of the LIS-assisted systems, the
5 authors in [67] suggested an approximation and upper bound
equation. Under the influence of Nakagami-m fading chan-
0 nels, the author looked at BPSK and M-QAM modulations.
40 50 60 70 80 90 For number of elements N < 4, the authors derived the exact
distribution of the channel coefficient and used a Gaussian
approximation for the in-phase and quadrature components
FIGURE 10 The transmit power to achieve the rate 6 bit/s/HZ [53] because the exact calculation becomes too complicated for
large values of N . However, the exact derivation for the BER
was only for two and three elements and the scenario is more
sophisticated when the number of elements is huge.
Unlike the authors in [67], the researchers in [68] estab-
lished a broad mathematical model for calculating the SEP
by determining the distribution of the received SNR, and the
number of elements used in the simulations was in terms of
hundred. On the other hand, the exact BER analysis of a two-
user NOMA system utilizing square QAM is discussed in [69].
Unlike previous work, there are no restrictions on the modu-
lation order of QAM symbols for any user. In Raleigh fading
channels, closed-form formulas for the BER of the successive
interference cancellation (SIC) receiver are developed. The BER
performance of an RIS-assisted NOMA downlink system is
derived in closed form in [70].
In [72], the researchers examine an reconfigurable intelli-
gent surface aided wireless system with and without a direct
link between the AP and the user, using a finite number of
RIS elements. The authors give a BER and average achievable
FIGURE 11 Data rates of RIS and relaying versus distance [59]
rate (AAR) study of RIS-based systems, assuming maximum
received power. The authors constructs a closed-form BER
Unlike the authors in [62], who look at point-to-point approximation that allows forecasting asymptotic performance
mm-Wave communications, the authors in [64] consider a gen- variation as SNRs and RIS elements increase. Simulation results
eralized mm-Wave downlink cellular network with random show more accurate BERs than previous studies [68]. How-
barriers and reflectors. According to the findings, only the ever, the performance of RIS-assisted wireless communications
placement of high-density reflectors can result in a consider- over Rician fading channels is discussed in [73]. For numer-
able increase in mm-Wave coverage while reflected signals must ous performance metrics, such as outage probability, average
travel greater distances than direct signals in low-density net- SEP, and channel capacity, the authors construct new accu-
works, and coverage probability does not differ from that of rate closed-form approximations. Asymptotic equations for the
blockages. A restriction on this work is that the authors did not outage probability at high SNR levels, as well as closed-form
take into consideration the optimized deployment for the reflec- formulations for the system diversity order and coding gain
tors consequently, the performance coverage of the network are provided to give a better explanation of the system behav-
was not promising. ior. The performance analysis in the previous works did not
The performance of outage probability in RIS-assisted vehic- consider practical implementations like imperfect CSI.
ular communication networks is examined in [66]. Using series The asymptotic optimality of achievable rate in a down-
expansion and the central limit theorem, the authors derived the link RIS system is investigated in [71], which takes place
expression for outage probability. The RIS can greatly lower in a real-world RIS environment with all of its limitations.
the probability of outages for vehicles in its area, according Under practical reflection coefficients, a passive beamformer
to numerical results. However, the authors in [65] investigated is proposed that can attain asymptotic optimal performance
the outage probability of intelligent reflecting surface aided by manipulating the incident wave characteristics. To enhance
HASSOUNA ET AL. 511

FIGURE 12 System model for RIS-based MISO [71] FIGURE 13 The actual capacitance of the RIS element at row 10 and
column 10 is correlated with the intended capacitances of itself and the
neighbouring elements

the feasible system sum-rate, a modulating mechanism that may

be applied in a RIS without interacting with current users is deployed, each with more reflecting elements, the system spatial
provided, and its average symbol error rate is asymptotically throughput increases, but at the cost of greater spatially varied
determined. Simulation findings show that the proposed meth- user rates and this is met with the research article in [79] when
ods are in close agreement with their upper bounds in the the author deployed one RIS only consisting of 4096 elements
presence of multiple RIS references. An obvious concern about to enhance the communication data rate for 50 users distributed
this work that a RIS cannot coherently align with all users con- randomly in a room of 13(Mtrs ) × 14( Mtrs ) and far from the
nected to BS via RIS (named RUE users as per Figure 12) RUEs’ RIS 16.5(Mtrs).
intended channels at the same time, which limiting system per- The authors in [77] studied the hardware impairments and its
formance in addition to the highly dense individual RIS surfaces impact on the performance data rate while in [79] the authors
needed per each user. analyzed in his system model the effect of mutual coupling
In an RIS-assisted mm-Wave MIMO system, the authors in between the adjacent reflecting elements on the achievable data
[74] define the RIS’s ideal phase shifts based on limited feedback rate per each user as per Figure 13 and then the authors rep-
from the mobile user to describe the possible data rate from the resented the mutual coupling per RIS elements statistically as
BS to the user. In addition to improving data throughput, simu- follows:
lation results demonstrate that by deploying the RIS with perfect
−dn,i
CSI, the positioning and orientation error bounds can both be ∑
N 100
𝜆
∼
decreased while the authors in [75] show the way that RIS can Cn = Ci,𝜃 ∑N −dn,i
(5)
be employed and optimized to boost the rank of the channel i=1 j =1 100 𝜆
matrix, resulting in significant capacity improvements. Because
∼
the previous works make use of a perfect RIS with unlimited where dn,i is the distance between element n and element i, Ci,𝜃
phase resolution, the capacity related study that arises has an is the capacitance that we assign to element i and Cn is the
undefined mismatch with practical systems. The authors in [76] actual capacitance of element n. Hence, we may expect that
produce an approximation of the feasible data rate and discuss studying parameters like hardware impairments, mutual cou-
performance deterioration when a practical RIS is implemented pling, and discrete phase shifts in the wireless communication
with constrained phase shifts. aided RIS will not only enhance the system performance in gen-
The authors in [77] employ a basic receiver architecture to eral but also establish a new era of communication research
explore the deterioration of attainable rate and discuss the cor- based on robust foundations that will end eventually with
relation design of equipment deficiencies as a function of the reliable technology.
distance between reflective elements. The deployment of the RIS in [79] was near the user equip-
The authors in [78] specified the spatial throughput of a ment’s however, the authors in [80] studied the distributed and
single-cell multiuser system supported by numerous RISs that centralized deployments of the RIS in the wireless network.
are randomly placed in the cell, in contrast to earlier works that Under symmetric channel topology, it has been proved that
concentrate on link-level performance optimization for RIS- centralised deployment outperforms scattered deployment in
aided wireless applications. When the number of RISs surpasses terms of achievable user rates. Nevertheless, the authors in [81]
a specific value, the simulations show the analysis is correct, and analyzed and compared two interesting scenarios in a SISO
the RIS-aided system outperforms the full-duplex relay-aided system, namely a finite number of big RISs and many finite-
counterpart system in terms of spatial throughput. Further- size RISs, to illustrate which implementation technique is more
more, it is demonstrated that alternative deploying procedures favourable. Using the deterministic equivalent (DE) approach,
for RISs and active relays should be used to maximize their the authors calculate the coverage probability in closed form
respective throughput. It is discovered that when fewer RISs are for both instances based on statistical CSI. The ideal coverage
512 HASSOUNA ET AL.

FIGURE 15 Scatter MIMO creates virtual AP for MIMO streams via

FIGURE 14 Spectrum sharing using reflect-array [26] phase shift reflection and its optimization algorithm [11]

probability is then calculated. Numerical results show, among Similar to [83], the authors in [84] not only did investi-
other things, that adding more surfaces beats the design strategy gate the effects of transceiver hardware impairments (T-HWIs)
of adding more elements per surface. and reconfigurable intelligent surfaces hardware impairments
In contrast to the early works, [26] presents a collaborative (RIS-HWIs) on a general RIS-assisted MU-MISO system with
analytical and empirical investigation to introduce a new spec- imperfect CSI and correlated Rayleigh fading, but also proposed
trum sharing solution for interior situations based on the use a novel optimization methodology for reflecting beamform-
of a reconfigurable reflect-array in the wireless channel. The ing matrices (RBM) optimization with low computational cost,
relevant signals for each transmission pair could be improved which is particularly useful in RIS-assisted systems with many
while interferences would be suppressed by managing the phase elements. The authors were able to get the channel’s linear
shift of each element on the reflect-array optimally. As a result, MMSE estimate with T-HWIs and RIS-HWIs. Furthermore,
numerous wireless users in the same room can simultaneously using just large-scale data, the uplink attainable sum SE with
access the same spectrum band without interfering with one MRC was calcultated in closed form and performed high com-
another. Consequently, network capacity can be significantly putationally efficient optimization about the RIS RBM. In
boosted. The smart reflect-array panels are hung on the walls general, the authors presented an approach that produced ana-
in the indoor setting, as shown in Figure 14. Even though the lytical and tractable formulations that were superior to prior
reflected array does not buffer or process any incoming signals; efforts, as demonstrated by simulation results.
it can alter the phase of the reflected wireless signal. Unlike other studies that focus on capacity, the paper in
The authors in [82] studied the uplink rate in the pres- [85] looks into the feasibility of employing an RIS with a
ence of restrictions such as hardware flaws, inaccurate channel large number of scattering components for terminal location.
estimation, and interference caused by device-specific spatially The author goes into details about the effects of deployments
correlated Rician fading. The authors has demonstrated that with a single centralized LIS and numerous smaller distributed
studies can reliably predict the performance of a LIS surface LISs constrained to the same total surface area. Splitting the
without the use of large simulations. Furthermore, it is shown LIS into 16 smaller LISs results in minor benefits, but it also
that in a LIS-based system, a channel hardening occurs, and the increases the overheads for different small LISs to collaborate
authors also found the asymptotic bound for the uplink data with one another.
rate and demonstrated that as the number of elements rises, Scatter-MIMO is a method for delivering MIMO spatial mul-
hardware impairments, noise, and interference due to chan- tiplexing gain that use a smart surface to enhance scattering
nel estimation errors and the NLoS path become insignificant. in the surroundings [11]. Smart surface connects to a wireless
In comparison to conventional massive MIMO, the simulation transmitter device, such as an active AP, and re-radiates the same
results show that a large-scale RIS can achieve greater reliability amount of power as any active AP, resulting in virtual passive
in terms of capacity expectation and variation. We noticed that APs as per Figure 15. By employing virtual passive APs, Scatter-
the number of elements that have been used in the simulations is MIMO eliminates the synchronization, interference, and power
huge in the order of thousands (10000) which raise our concern requirements of traditional dispersed MIMO systems, allowing
about the mutual coupling between the adjacent elements and its smart surface to give spatial multiplexing gain at a cheap
the practical phase-shift model considerations. In LIS systems, cost. According to the simulation results, Scatter-MIMO gives
the authors in [83] looked at the capacity impacts of hardware a median throughput gain of 2 over the active AP alone.
impairments (HWI). The authors created a general model of the From the above-mentioned references and different illus-
HWI based on the distance between a considered point on the trations, we would like to refer that the RIS aided wireless
LIS and its Centre, with the latter serving as a reference point in communication systems can be used as a reflector, receiver
hardware design. To limit the negative impacts of hardware defi- and transmitter [13] by modifying the phase shifts of the RIS’s
ciencies, the analytical and simulation results suggest dividing a scattering units. The RIS’s outgoing waves can generate differ-
largescale RIS into a succession of smaller RIS units. ent radiation paradigms that can convey data if these patterns
HASSOUNA ET AL. 513

can be identified and distinguished at the receiver. This is the strength or destructively at the non-intended receiver to limit
basic design concept of spatial modulation, which is commonly interference by smartly altering the phase shifts of all scattering
implemented using programmable antennas [86, 87]. This is elements, as shown in Figure 9. The experimental demonstra-
very similar to wireless backscatter communications, which use tion and channel measurements in [43] support this, paving the
load modulation to adjust the antenna’s reflection coefficients road for more theoretical research and system optimization. In
[88]. It’s difficult to get a high rate for SU user in CR commu- the following section, we go over the most common optimiza-
nication systems when there is a lot of cross-link interference tion formulations and solutions for RIS-assisted narrow and
with the PU user. The authors in [89] used the recently devel- wideband wireless systems for single and multi-users.
oped RIS to solve this problem in their research. In particular,
the authors investigate an RIS-assisted CR system, in which an
RIS is used to help with spectrum sharing between a PU and an 3.2.1 Passive beamforming techniques
SU link. By combining SU transmit power and RIS reflect beam-
forming to optimise the feasible SU rate for a particular SINR In this part, we will investigate the passive reflection optimiza-
target for the PU connection. The authors used the AO algo- tion for RIS-assisted wireless communications. We will assume
rithm to solve the SU rate maximization problem via the joint the knowledge of direct and indirect channels for the purpose of
transmit power control and RIS reflect beamforming. Simula- exposition however, the channel estimation will be discussed in
tion findings demonstrate that the RIS-assisted CR is effective Section 4. The optimization problem can be expressed generally
for secondary transmissions, even in the tough scenario when as follows:
the secondary transmitter is much closer to the primary user.
RIS-enhanced energy detection is investigated in depth in |( ) |2
(OP) max f (Q, 𝜽) = | V T 𝜃G + hd Q |
[90] for single-user spectrum sensing, cooperative spectrum Q,𝜽 | |
sensing, and diversity reception. For each example, a perfor- s.t:
mance analysis is presented, as well as an analysis for the average
probability of detection and false alarm. Monte Carlo simula- (C 1) ∶ ‖Q‖2 ≤ p (6)
tions are used to verify the validation of his results. It is believed
that RIS can significantly increase detection performance. (C 2) ∶ 𝜙1 , … , 𝜙N ∈ [0, 2𝜋)
The authors in [91] investigate RIS concept in the UAV
enabled communications to expand network coverage and (C 3) ∶ 𝛾1 , … , 𝛾N = 1 (7)
improve communication reliability and spectral efficiency of
IoT networks. The authors also show that RIS-assisted UAV where Q is the transmit beamforming vector, 𝜃 is the RIS matrix
communication systems may achieve ten times the capacity of with unit-modular constraint on each element and f (Q, 𝜃)
traditional UAV communication systems in terms of achievable denotes the objective function. We envision that the joint beam-
ergodic capacity. The impact of imperfect phase knowledge on forming optimization problem is a non-convex problem due to
system capacity and BER analysis for UAVs communications the fact that both Q and 𝜃 are linked to each other. Further-
assisted by flying intelligent reflecting surfaces is examined in more, RIS elements need unit magnitude since, unlike relays,
[92, 93]. The authors in [93] takes into account a multi-layer they do not amplify or decode then transmit a received sig-
UAV network with several hops, whereas [92] only addresses nal. The necessity for suboptimal tractable rate optimization
a single-hop case. The obtained results demonstrate the need solutions is motivated by the non-convex unit modular con-
of correct phase estimation for RIS-based systems, especially straints in the above mentioned optimization issue. Previous
for systems with a limited number of reflecting elements. The works on narrowband [94–107] and wideband [108–113] RIS-
results are also demonstrating the importance of the number of assited wireless communication systems attempt to solve the
elements in getting a reliable performance. non-convex problem where the main challenge includes the unit
modulus constraint. Several techniques have been used in the
literature to address this constraint.
3.2 Optimization techniques and
algorithms ∙ The alternating optimization method:
The famous alternating optimization method is examined
The impact of multipath on the received signal is determined extensively in the literature. The method style of switching
by how great or small the spread of time delays associated with between the active transmit beamforming and passive reflec-
the LoS and other multipath components are concerning the tion gives it an advantage to deal with the active transmit
inverted signal bandwidth. The LoS and other multipath com- beamforming as a conventional problem when the passive
ponents are often non-resolvable if the channel delay spread is beamforming is fixed; however, under the given active beam-
minimal, leading to the narrowband fading model. The LoS and forming the passive beamforming is still a non-trivial exercise
all multipath components are often resolvable into several dis- to tackle including the unit modular constraint on each ele-
crete components if the delay spread is large, resulting in the ment of the RIS. There are possible ways have been shown
wideband fading model [5]. The reflected signals can be merged in the literature to deal with this constraint. For exam-
coherently at the intended receiver to boost the received signal ple, the SDR method is applied to relax the non-convex
514 HASSOUNA ET AL.

rank-one constraint to a standard convex semidefinite pro-

gram (SDP). The SDP is used then to solve the formulated
non-convex quadratically constrained quadratic program
(QCQP) problem. However, the relaxed problem may not
result to a rank one solution which by return require
obtaining the eigenvalue decomposition by using circularly
symmetric complex Gaussian (CSCG) distribution methods.
The SDR with Large number of gaussian randomizations will
lead eventually to approximation of 𝜋∕4 of the optimal objec-
tive value [100]. Consequently, the SDR approach can only
provide an approximation and solving an SDP program is
computationally expensive for large number of antennas and
RIS elements.
∙ The iterative methods:
The concept of the iterative algorithms is to find a locally
or near optimal solution for the objective problem at a rea-
sonable computational complexity and acceptable run time.
For example the low-complexity successive refinement algo- FIGURE 16 Achievable rate of RIS-aided MISO system versus AP-user
rithm [94, 98] to determine the optimal discrete phase shifts horizontal distance, d [11, 114]
of different elements at RIS one by one in an iterative way.
Another technique based on the fixed point iteration and
manifold optimization is investigated [99]. The authors in coverage when compared to a traditional setup without the RIS.
[107] used the projected gradient method (PGM) to jointly For example, Figure 16 [3, 114] investigates four schemes (1)
optimize the covariance transmitted matrix and the phases AO’s joint transmit and passive beamforming design, (2) AP-
of the RIS elements. The PGM method achieved the same user MRT transmission, (3) AP-RIS MRT, and (4) MRT without
data rate as the alternating optimization method but with RIS. When the user is near to the AP, the AP-user MRT system
less number of iterations and lower computational complex- works similarly to the AO method, but suffers a considerable
ity. The authors in [113] extended the work of [107] in the rate loss when the user is nearer to the RIS, as illustrated in
wideband MIMO communication. The paper in [109] pro- Figure 16. This is to be expected, given the AP-user direct
posed the low complexity power method to compute the link dominates the user received signal in the former scenario,
dominant eigenvector of the reflection coefficient matrix to but the RIS-user link dominates in the latter. Furthermore, the
configure the RIS phases. In general, all these iterative meth- AP-RIS MRT system operates in the opposite manner when
ods have shown good system performance with acceptable compared to the AP-user MRT counterpart. Consequently, as
computational complexity and reasonable run time. shown in Figure 16, if the transmit beamforming is not ade-
quately built according to the whole channels, the attainable rate
utilizing the RIS may be poorer than the conventional MRT
3.2.2 Active and passive beamforming: The without the RIS, for example, in the application of the AP-RIS
state-of-the art MRT scheme for distance d ≤ 40(Mtrs). This emphasises the
significance of combining active and passive beamforming to
The authors in [114] investigate an RIS-assisted point-to-point achieve the best balance of direct transmission to the user and
MISO wireless system in which one RIS is used to aid transmis- indirect transmission via RIS reflection in order to maximise the
sion from a multi-antenna access point AP to a single-antenna received signal strength at the user. For a desired SNR of 8 dB,
user. As a consequence, the user simultaneously receives both the network coverage without the RIS is roughly 33 (Mtrs) ,
the signal directly from the AP and the signal reflected by the but when the suggested combined beamforming designs with
RIS. By combining the active transmitted beam at the AP with an RIS are utilised, this value is increased to about 50(Mtrs)
the passive reflected beam at the RIS using phase shifters, the [114]. Another important result in [114] is that the user’s receive
authors aimed to optimize overall received signal power at the power grows quadratically as N grows, by a factor of (N 2 ).
user. Assuming that the RIS contains global CSI, a centralised The transmit power can be reduced at the AP in the order of
solution based on the SDR technique was proposed. Because 1∕N 2 without sacrificing the receive SNR. This is due to the fact
the centralised approach includes excessive channel estimation that the RIS catches an extra aperture gain of (N ) in the AP-
and signal exchange overheads, a low-complexity distributed RIS link in addition to the reflect beamforming gain of (N ) in
technique is used, in which the AP and RIS modify the trans- the RIS-user link.
mit beamforming and phase shifts in alternating fashion until The power scaling in reconfigurable intelligent surfaces and
convergence is achieved. When compared to benchmark sys- its comparison with m-MIMO, are studied in the reference [115]
tems, simulation results suggest that the proposed techniques in which the authors proved analytically that the gap between
can obtain high performance gains. Furthermore, it has been m-MIMO and RIS is enormous; an RIS requires a high number
proven that the RIS can significantly improve link quality and of reflecting elements to achieve SNRs equivalent to massive
HASSOUNA ET AL. 515

m-MIMO and by concluding such solid fact the authors disap- them more complicated and challenging to solve. The fun-
proves of the myths and the wrong understanding of the RIS damental capacity limit of RIS-assisted point-to-point MIMO
fundamental concepts. communication systems with multi-antenna transmitter and
A point-to-point MISO communication system with RIS receiver is quantified in general by simultaneously optimising
assistance is also investigated in [116]. The beamformer at the RIS reflection coefficients and the MIMO transmit covari-
the AP and the RIS phase shifts are tuned simultaneously to ance matrix, which differs from previous MISO systems [118].
improve spectral efficiency. The resultant non-convex optimiza- For frequency-flat channels, an alternate optimization approach
tion problem is handled with the help of two efficient methods was devised to discover a locally optimum solution by optimis-
that employ fixed-point iteration and manifold optimization ing one of the reflected coefficients or transmitted covariance
approaches, respectively. The proposed techniques not only matrix at a time while leaving the others constant, and the
improve spectral efficiency but also reduce computational com- best possible solution were found in closed-form. Alterna-
plexity as compared to the current state-of-the-art approach. tive, less difficult algorithms for asymptotically low and high
The RIS-assisted MISO downlink model described above, as SNR circumstances, as well as MISO and SIMO channels,
well as the heuristic alternating optimization described in [114, were devised. Furthermore, for frequency-selective channels, a
116], provide a generic framework for the optimum design of MIMO-OFDM system was investigated, in which a collection
RIS-assisted systems that may be used in a variety of network of reflection coefficients for all subcarriers must be designed. A
scenarios. Despite the fact that RISs with continuously phase novel alternating optimization approach used the convex relax-
shifts and maximum reflection magnitudes were explored, it ation technique to successively optimise a set of transmitting
was discovered that realistic RIS with discrete phase shifts and covariance matrices across multiple subcarriers or a common
discrete reflection amplitudes satisfied the hardware criteria. reflection coefficient for all subcarriers. Extensive numerical
Different from the preceding sections, the authors in [117] results reveal that the suggested algorithms outperform several
look at an RIS-based wireless communication, in which an RIS benchmark methods with and without RIS in terms of rate per-
with a limited figure of phase shifts at each element is used to formance. The performance gain improves as the number of
enhance communication between a multi-antenna AP and many elements increases.
individual-antenna users. Particularly, the continuously transmit- The authors in [119] make the first efforts to investigate
ted precoder at the AP and the discrete phase shifts at the RIS the wideband beamforming for RIS-assisted mm-Wave massive
were tuned together to lower transmit power at the AP while MIMO using a different design. For RIS-assisted millimetre-
still achieving user SINR requirements. For single-user and mul- wave (mm-Wave) hybrid MIMO systems, the authors present
tiuser instances, both optimum and successive refinement based a geometric mean decomposition-based beamforming method.
suboptimal solutions were investigated. Furthermore, when With the help of RIS, simulation results show that the proposed
the figure of reflecting elements gets asymptotically large, the strategy can achieve good BER performance in a wideband
performance degradation of RIS due to discrete phase shifts hybrid MIMO system. To increase the system sum-rate, the
against the ideal scenario with continuous phase changes was authors [120] updated the source precoders and RIS phase shift
investigated. Surprisingly, it was discovered that utilizing RIS matrix in the full-duplex MIMO two-way communication sys-
with even 1-bit phase shifters can attain the same asymptotic tem. To maximise the system sum rate, the RIS phase shift
squared power increase as continuous phase shifts when only matrix and source precoders are tuned jointly. The Arimoto-
a constant power loss in dB is applied. In comparison to the Blahut approach is used to divide the non-convex optimization
situation without RIS, simulation results revealed that employ- issue into three sub-problems that are tackled alternately.
ing RIS with discrete phase shifts can save significant transmit Closed-form solutions can be used to solve all the sub-problems
power. Furthermore, it was shown that immediately quantiz- quickly. RIS provides a performance boost equal to a relay oper-
ing the optimised continuous phase shifts to generate discrete ating at a transmission power of only 40 dBm to 35 dBm. This
phase shifts offers near-optimal performance in the single-user is related to the RIS’s concern with double-fading. It’s worth
scenario, but that performance is significantly reduced in the noting, though, that RIS does not need any transmission power.
multiuser case owing to substantial co-channel interference. All of the aforementioned research studies in [116, 118–120]
Eventually, owing to the multiuser channel rank enhancement rely on an ideal RIS with infinite phase resolution, that is,
provided by the RIS’s extra signal routes, it was proved that the each scattering element’s phase shift may be entirely con-
ZF precoder-based algorithm performs almost as well as the trolled. However, this is difficult to accomplish in practice, and
MMSE precoder-based method. devising precise phase control algorithms is similarly tough.
We looked at narrow-band communication systems with Furthermore, for the RIS controller to provide precise phase
frequency-flat fading channels and a single and multiple antenna control, full CSI is usually required. This implies that data shar-
for the AP and user in the aforementioned works. Conse- ing might be incredibly expensive, particularly for RIS that is
quently, the wideband communication is important to explore self-sustaining due to wireless energy harvesting.
in the literature. Passive RIS reflection optimization prob- The authors in [121] presented a novel way to use the
lems for MIMO systems with multiple antennas at both the RIS to improve the attainable rate of an OFDM system. The
AP and the user, as well as broadband OFDM systems with authors developed a workable transmission protocol by assem-
frequency-selective fading channels, must cater to multi-antenna bling the RIS elements as shown in Figure 17 and calculating
channels and multi-path channels with different delays, making the joint channel for every group, with data transfer based on a
516 HASSOUNA ET AL.

FIGURE 18 Achievable rate versus transmit power for RIS-aided OFDM

system [122]

RIS proposes a unique reflection pattern to facilitate channel

estimation at the AP based on received pilot signals from the
user, for which the channel estimation error is calculated in
FIGURE 17 Illustration of the RIS reflect array and proposed elements closed-form under the unit-modulus constraint. After that, the
grouping [121] reflection coefficients are optimised using the estimated CSI
and a low-complexity approach based on the temporal domain’s
resolved strongest signal route. It is observed from Figure 18
shared RIS reflection coefficient. Then, based on the predicted that in comparison to the OFDM system without RIS and the
channels under the recommended protocol with any defined OFDM system with random RIS phase shifts and the related
grouping, the authors constructed a combined optimization optimal transmit power allocation, the SCA-based algorithm
problem of transmit power allocation and RIS reflection coeffi- suggested in [121] achieves a much higher rate. Furthermore,
cients. The non-concave rate function was estimated utilising its the most powerful CIR maximization algorithm provides similar
concave lower limit based on the first-order taylor expansion results to the SCA-based method, making it a cost-effective and
in an efficient sequential convex approximation (SCA)-based low-complexity option. The figure also sets a possible upper rate
technique. The SCA-based approach guarantees that the RIS’s limit by proposing that various RIS reflection coefficients can be
combined reflection and transmission power optimization issue created ideally for various sub-carriers, resulting in a frequency-
will converge to a stationary point, and it only has polynomial selective RIS reflection system. With the realistic frequency-flat
complexity over the number of elements and sub-carriers. The RIS reflection, this upper rate limit greatly outperforms the
simulation results show that the improvements in the data rate SCA-based approach, and the rate difference expands as the
is connected with the selection of the group ratio and the coher- number of sub-carriers rises. As a result of its passive func-
ence time of the channel however, we have seen a performance tioning, RIS-aided OFDM systems have a basic weakness in the
degradation in the low SNR regime as the design parameters absence of frequency selectivity of the RIS reflection. Due to the
are adjusted based on the calculated CSI, the CSI obtained at necessity to cater to additional channels in both space and fre-
low SNR is inaccurate, resulting in a larger performance degra- quency, the RIS reflection design for rate increase is entailed in
dation at both low and high grouping ratios so, we think that more general RIS-aided MIMO-OFDM systems where the user
to maximize the obtainable rate, the grouping approach as well and AP are kitted out with more antennas. Additionally, numer-
as the training sequence can be further improved. In addition, ous transmit covariance matrices at distinct sub-carriers must
the author considers continuous phase and maximum amplitude be optimised together with the RIS reflection. Consequently,
equal one in the optimization schemes rather than proposing the authors in [118] expanded the narrow-band MIMO situ-
practical discrete phase shift. ation and using the convex relaxation technique, the authors
To reduce complexity even more, the authors in [122] suggested an effective AO-based solution. Despite the absence
proposed a simpler approach called highest channel impulse of frequency selectivity, the results in [118] revealed that RIS
response (CIR) maximisation, in which RIS phase changes are is still useful in enhancing the systems rate of MIMO-OFDM
only aligned with the time-domain channel with the strongest with well-planned RIS reflection coefficients when compared
path power. The authors proposes a feasible transmission to a typical system without considering RIS.
protocol for an RIS-enhanced OFDM system that conducts Different from the above researches, the authors in [79]
channel estimation and reflection optimization sequentially. The investigated a wideband OFDM system supported by a
HASSOUNA ET AL. 517

150 sum-rate maximization and MSE minimization. In Both scenar-

140 Lossless (Amplitude=1)
ios complexity and convergence are explored. When compared
130 to the ideal RIS reflection model, simulation results demonstrate
120
110
that the proposed technique may provide considerable average
1 Bit RIS: 180 Degree
100 NLOS USERS
Phase Spacing sum-rate augmentation, underscoring the importance of using
90 the realistic model in wideband system design. However, the
80 LOS USERS authors put the assumption that the channels were well known,
70 but in practice, a channel estimating phase is required. Because
60
50
Random 90 Degree the RIS is a passive device, the estimation must be done at the
Phase Spacing
40 receiver which was not considered in addition to the proper
30 employment of the RIS.
20 Uniform Surface (RIS OFF)
10
0
0 5 10 14 20 25 30 35 40 45 50 3.3 Discussions and insightful prospect for
section III
FIGURE 19 Achieved Data Rate using dataset in [123] ∙ In summary of the performance analysis, we would like
to refer that the bulk of the research papers presume that
the channel state information and discrete phase shifts are
practical RIS configuration with two binary states per each ele- known at both transmitter and receiver [68]. As a result,
ment as well as mutual coupling between adjacent elements. more practical setups are needed to evaluate the effectiveness
Using the dataset from [123], new channel estimation and of implementing large-scale RIS cellular communications
configuration techniques have been proposed and analyzed. networks in practice. Furthermore, neither for indoor nor
Because of the tradeoff between multiple subcarriers, finding a outdoor communication setups, the above-mentioned litera-
good RIS configuration with reasonable complexity is difficult ture pays little attention to user mobility. User mobility causes
in OFDM systems. not just handoffs between RIS units, but also a geographical
In [122, 129], the strongest tap maximization approach is correlation in user distribution, which could have a significant
studied, to maximize the magnitude of the channel’s greatest impact on system performance. Consequently, incorporating
entry. This works well for LoS channels but not for NLoS various mobility models into performance will become a cru-
channels [16] and extending it to the scenario where each RIS cial study path in the future. Furthermore, the maximization
element has just two states is difficult. Instead, the authors of performance gain for a certain network structure prob-
aimed to optimize the total signal power received by creat- lem involves a combined tuning of active transceivers and
ing a heuristic algorithm based on the iterative power method, RIS passively scattering components. Passively reflected sig-
which finds the dominating eigenvalue by iterating the com- nals, which are intimately related with the active transceivers’
putation until it converges. The simulation results show the control variables, can be thought of as the combined phase
rates achieved by the 50 Users in dataset as per Figure 19, control of scattering components. Because the scattering
which are arranged by increasing rates using the proposed objects in wireless environment scenarios are unpredictable
technique. Two benchmarks are used: (1) choose the best and omnipresent, performance evaluation of RIS-assisted
configuration from among the N examined during pilot trans- wireless systems is complicated, but it also generates new
mission; and (2) turn off all coefficients to simulate a uniform optimization challenges that demand creative methods to
metal surface. For all Users, including NLoS, the suggested account for active-passive device communication [3, 16].
technique gives higher rates. When compared to a uniform sur- Table 4 is a summary of some of the work done in the perfor-
face, the sum rate with an optimized RIS is 3.3 times higher, mance system analysis considering the usage of the intelligent
demonstrating the technology’s considerable benefits even in reflecting surfaces as a reflector, receiver, and transmitter in
complex configurations. addition to the design objectives. We believe that in order to
Similar to [79], the authors in [130] investigated a wideband construct genuinely widespread wireless networks that can
orthogonal frequency-division multiplexing OFDM system for provide continuous communication and good quality of ser-
MISO multi-users, the authors first look at the dual-phase and vice (QoS) to numerous users in such a challenging wireless
amplitude squint effect of reflected signals before presenting a environment, new and radical solutions are still required.
simplified RIS reflection model for wideband signals. Then, a ∙ In summary of the optimization techniques and algorithms,
wideband RIS enhanced MU MISO-OFDM system is exam- we have presented the different optimization techniques
ined. When using both continuous and discrete phase shifters, used in the literature for both narrow and wideband sys-
the transmit beamformer and RIS reflection are developed tems taking into consideration many communication systems
jointly to maximise the average sum-rate over all subcarriers. parameters like capacity, rate, sum rate and SNR to study the
The original problem is turned into a multi-block variable prob- performance and efficiency. Different system models were
lem that may be solved effectively using the block coordinate illustrated using single and multi-users under various chan-
descent (BCD) approach by leveraging the relationship between nels and environments however, there are still huge subjects
518 HASSOUNA ET AL.

TABLE 4 Summary of existing work in the field of RIS systems performance analysis

Criterion for measuring

Reference Communication setup IRS functionality performance Design goal

[60] MIMO Transmitter BER Enhance the performance and boost

the spectral efficiency
[61] SISO in the presence of random Reflector Probability of being a reflector If an item is coated in meta surfaces,
objects the chance that it is a reflector is
unaffected by its length, but it is
strongly affected if Snell’s law of
reflection must be applied
[62] SISO mm-wave communications with Reflector Outage probability Even when the links are impeded by
Blocked LoS barriers, a reflect-array deployment
may provide reliable mm-Wave
connections for indoor
communications
[63] MISO Reflector Outage Probability The effects of several critical system
factors on the ideal outage
probability are analysed to uncover
crucial design insights
[64] SISO in the presence of both Reflector Coverage probability Improving the coverage in
line-of-sight signal blockages and high-density networks
reflectors
[67] SISO under double Nakagami-m Reflector Bit error probability Performance improvement
channels
[68] SISO under Rayleigh fading channel Reflector and Transmitter SEP Increasing the received SNR
[70] Multi-users NOMA system Reflector BER Improving system performance and
reliability
[71] Multi-users MISO Reflector SER Optimal SNR and increase Sum rate
[74] mm-wave MIMO Reflector Achievable Rate High rate in low SNR
[75] Point to point MIMO Reflector Achievable Rate Rate by choosing proper RIS
deployment and phases
[76] Narrow band SISO Reflector Achievable data rate Maximize data rate
[78] Single BS-Multiusers Reflector Spatial throughput Maximize spatial throughput for users
[79] Single BS-Multiusers Reflector Achievable data rate Sum Rate enhancement
[80] Single AP- Multiuser Reflector Achievable user Rate Superior rate performance of
centralized over distributed
[81] SISO Reflector Coverage Probability Increasing the number of surfaces
surpasses the design technique of
increasing the number of elements
per surface.
[82] Uplink LIS-based large antenna-array Receiver Ergodic Rate Produce performance comparable to
system for single Antenna traditional massive MIMO
multiusers
[83] Uplink single user to signal processing Receiver Capacity Reducing the impact of hardware
unit impairments
[84] MU-MISO Reflector Sum SE Enhance sum SE
[85] Single antenna radiating to LIS Receiver Coverage positioning Distributed deployments have the
ability to expand terminal
placement coverage and deliver
superior average Cramer–Rao lower
bound (CRLBs) in all dimensions.
[124] MISO Reflector Coverage probability and Improve system coverage probability
average throughput and throughput without consuming
more energy.
[125] MISO-OFDM Reflector Downlink rate Downlink rate enhancement despite
the lack of independent RIS phase
control

(Continues)
HASSOUNA ET AL. 519

TABLE 4 (Continued)

Criterion for measuring

Reference Communication setup IRS functionality performance Design goal

[126] Single user SISO Reflector Array Outage probability and the The RIS-assisted system outperforms
average bit-error probability the AF relay system with fewer
reflecting elements
[127] Single source and two wireless sensor Reflector Average symbol error As the number of reflecting elements
nodes probability (ASER) and the grows, the performance improves
outage probability
[128] RIS-assisted NOMA Forward Relay Outage probability and ergodic Enhance energy efficiency compared
rate to conventional cooperative
communications.

that need to be investigated which are not yet covered prop- To address the challenges, let us consider the downlink wide-
erly in the above research like the doppler shift and mobility, band SISO communication. Recall (1) but for the wideband
practical phase shift models, proper and fast channel esti- system, The received signal can be represented as:
mations, and mutual coupling for the adjacent RIS elements.
Table 5 summarizes the work of RIS reflection optimization ( )
studies based on the system settings analysis and optimization z = F HeqT 𝜃 + hd ⊙ x + w (8)
methodologies used.

Where F ∈ ℂK ×K is the DFT matrix and K is the number of

4 CHANNEL ESTIMATION subcarriers, Heq = [heq1 , heq2 , … , heqK ] ∈ ℂN ×K is the cascaded
channels V and G to and from the RIS, respectively, the con-
4.1 Channel estimation protocol and pilot figuration of the RIS is given by the diagonal matrix 𝜃 = ΓΦ
transmission with Φ = diag(e− j ∅1 , … , e− j ∅N ) and Γ = diag(𝛾1 , … , 𝛾N ), hd =
[hd 1 , … … , hdK ] ∈ ℂK ×1 gathers all the uncontrollable chan-
To effectively appreciate the performance advantages given by nel components and w ∼ ℂ (0, 𝜎2 ) is the receiver noise.
RIS, accurate CSI is necessary, which is a practical obstacle. The Accordingly, the downlink CSI includes the number of chan-
challenges can be summarized as follows: nel coefficients to and from the RIS (i.e. V and G ) equal to
(NK + NK × U ) where U = 1 for single user antenna and
∙ Aside from having significantly more RIS-induced channel the number of channel coefficients for the direct link hd equal
coefficients than a conventional system without RIS, RIS to (U × K ). The total number of coefficients is different for
channel estimation has a hurdle in that its cheap reflecting TDD and FDD systems. Furthermore, they are substantially
components lack active RF chains and hence are unable to more as compared to the conventional systems. It is sufficient
broadcast pilot and training signals to facilitate channel esti- to estimate the cascaded BS-RIS-User channel Heq for any RIS
mate, in contrast to traditional systems’ active BSs and user configuration 𝜙. Suppose pilot signal x is transmitted on each
equipment. of the S subcarriers. Let the RIS configuration 𝜙i where i is the
∙ The RIS channel estimation is constructing high-dimensional index of the OFDM block. The received signal:
channels from low-dimensional channels with approximated
CSI.
∙ In broadband systems with frequency-selective fading chan- z̄ (i ) = FS HeqT 𝜃𝜙i x + w(i
̄ ) ∈ ℂS , (9)
nels, extra channel coefficients must be estimated due to
the multi-path delay spread and subsequent convolution
where FS contains the S rows of F . There are SN unknown
of the User-RIS and RIS-BS multi-path channels in each
coefficients in Heq but we only get S observations from Z̄ (i ).
cascaded User-RIS-BS channel. Additionally, although the
One possible way to get SN linearly independent observations,
channels are frequency-selective, the RIS reflection coef-
is to consider a sequence of N OFDM blocks with different
ficients are frequency-flat, making it impossible to create
configurations (𝜙1 , … , 𝜙N ). So, the joint received signal [139]:
flexible various frequencies, such as separate sub-carriers
in OFDM communications. In all SISO, MISO, MIMO,
single and multi-user scenarios, forecasting the RIS chan- [
z̄ [1], … , z[N
̄ ] = FS HeqT 𝜃1 , … , 𝜃N ]x + w[1],
̄ … , w[N
̄ ] (10)
nel for broadband frequency-selective fading channels is
much more challenging than estimating the RIS channel for
narrow-band flat-fading channels due to these characteristics Let Z̄ = Z̄ [1], … , Z̄ [N ], w𝜃 = [𝜃1 , … , 𝜃N ] and W̄ =
[3]. ̄
w[1], … , w[N
̄ ]. If w𝜃 is invertible then we can apply the
520 HASSOUNA ET AL.

TABLE 5 Summary of existing work of optimization techniques for narrow and wideband systems

Reference System model Algorithm Algorithm Performance Work Objective

[79] Wideband Multiuser SISO-OFDM Power Method Heuristic Maximize Data Rate for 50 Users
[114] Narrowband Single/Multiuser MISO AO and SDR Near Optimal In an RIS-assisted multiuser
system, reduce the transmit
power as much as possible
[116] Narrowband Single User MISO Fixed point iteration, manifold Locally Optimal Maximize Spectral efficiency and
optimization lower computational complexity
[117] Narrowband Single and Multiuser ZF precoder and MMSE precoder Near Optimal Combine the AP’s continuous
MISO based algorithms transmit precoding with the
RIS’s discrete reflect phase
changes to reduce transmit
power at the AP
[118] Wideband Single User MIMO-OFDM AO and convex relaxation Locally Optimal Capacity enhancement
[119] Wideband Single user mm-wave Geometric mean decomposition Heuristic Improve BER without
MIMO-OFDM (GMD) based on beamformer and sophisticated bit/power
combiner allocation
[120] Narrowband Full-duplex MIMO Arimoto-Blahut algorithm Near Optimal Maximize sum rate
[121] Wideband Single user SISO-OFDM SCA, AO Near Optimal To optimise the achievable rate,
combine the transmit power
allocation and the RIS passive
array reflection coefficients
[122] Wideband Single user SISO-OFDM The strongest-CIR maximization Near Optimal Maximize the achievable date rate
(SCM) method
[129] Wideband Single user SIMO-OFDM SDR, STM Near Optimal Improve the maximum achievable
rate and data transfer latency
[130] Wideband Multiuser MISO-OFDM Three-phase one-dimensional search Near Optimal Determine the wideband
method MU-MISO-OFDM system’s
highest average sum-rate
[131] Wideband Single user SISO-OFDM AO / The necessity of considering
practical RIS model for channel
estimation
[132] Wideband Multiuser SISO-OFDM Quadratic unconstrained binary / Assigning each RIS to a maximum
optimization formulation (QUBO) of one UE using the allocation
scheduling method for each UE
[133] Wideband Single-user downlink AO High Quality Suboptimal Optimize the transmission power
OFDM allocation at the BS and the
passive array reflection
coefficients at the IRS to boost
the user’s downlink achievable
rate
[134] Wideband Multi-antenna for AO Approximation and Maximizing the sum secrecy rate
eavesdropper MIMOME-OFDM Suboptimal
[135] Wideband Multicell multiuser MISO Successive Convex Approximation Iterative Suboptimal Under QoS constraints, maximise
OFDMA ultra-reliable low latency and iterative rank maximization the weighted total throughput
(ULLC) approach (IRMA)
[136] Wideband RIS-Assisted UAV AO Approximation The employment of an RIS in UAV
OFDMA OFDMA transmission can boost
the system’s sum-rate
dramatically
[137] Wideband Multiuser RIS assisted UAV Successive Convex Approximation Iterative Suboptimal Maximize the lowest possible
with the Rate restriction penalty average rate for all users
[138] Narrowband Single user MISO Branch and Bound (BnB) Globally Optimal Maximize spectral efficiency
HASSOUNA ET AL. 521

in designing OFDM-based pilot symbols to effectively estimate

the channels of multiple users simultaneously. For instance, by
exploiting the common BS-RIS channel and the LoS dominant
RIS-User channels. It was demonstrated that N + 1 OFDM
symbols are required to estimate the cascaded channels of up
(N +1)(K −L)
to [ ⌋ + 1 users at the same time. Where L is the
N +L
number of multipath and K is the number of subcarriers.

4.2 Channel estimation: The state-of-the art

The authors in [149] examines the progression of the reflecting

radio idea to RISs, as well as the RIS-assisted MISO communi-
cation model and how it differs from traditional multi-antenna
communication models. For the design and study of RIS-
assisted systems, a MMSE based channel estimate technique
was proposed. The BS instructs the microcontroller to keep
FIGURE 20 RIS assisted wireless communication with sensing elements
(nearly passive RIS) all RIS elements turned off throughout the channel estimation
phase, and the BS estimates the direct channel for all users.
The BS then sends a signal to the microcontroller to turn ele-
least square (LS) estimate to compute Heq as follows: ment n of the RIS ON while leaving the other elements OFF,
allowing the BS to begin estimating the cascaded channel. As
1 −1 −1 1
FS Z̄ w𝜃 = HeqT + FS−1W̄ w𝜃−1 (11) a result, the microcontroller instructs the RIS’s control circuit
x ⏟⏟⏟ x board in Figure 3a to carry out the needed sequence, and so on.
⏟⎴⎴⏟⎴⎴⏟ ⏟ ⎴⎴ ⏟ ⎴⎴ ⏟
Known Estimated Noise The MMSE estimating method is used to calculate the estima-
Signal Composite Signal tions. The BS then computes the optimal beamforming vector
Channel using the channel estimates and transmits it to the RIS micro-
controller. Using the estimated channels (direct and cascaded),
The challenge is when w𝜃 is non-invertible and the system is the phase shifts of the RIS are tuned by utilizing a gradient
overdetermined to solve. ascent algorithm.
There are two basic methodologies for RIS channel estimates Similar to [149] the authors in [150] describe a unique pas-
in the available literature, which are based on two alternative sive intelligent surface PIS-assisted energy transfer mechanism
RIS configurations, depending on whether it is installed with from a multi-antenna power beacon (PB) to a single-antenna
sensing devices (receive RF chains as per Figure 20) or not, and energy harvesting user. A controlled LS Channel estimate pro-
are referred to as semi-passive RIS and completely passive RIS, tocol with binary reflection was proposed. This binary model is
respectively. For more details, refer to [140, 141]. One example used because it accounts for the fact that a passive intelligent
for the full passive channel estimation method is by employing surface (PIS) lacks active components, forcing PB to estimate
[142] the ON/OFF based RIS reflection pattern but it requires all the channel vectors on its own. The above-aforementioned
N + 1 channel coefficients and substantial reflection power works require at least N + 1 pilot symbols for predicting the
loss. An other workable method by grouping the neighbouring total N + 1 channel coefficients in the system. Because only one
RIS elements into smaller RIS surfaces so, only the effective cas- element is turned on at a time, the ON/OFF-based RIS reflec-
caded BS-RIS-User channel associated with each smaller surface tion setup suffers from considerable reflection power losses,
needs to be estimated, hence significantly reducing the training resulting in a feeble reflected signal.
overhead [108, 143]. Different from the above research of work, as shown in
In case of MIMO/MISO multi-user systems, the training Figure 17 (over which the RIS channels are typically spatially
overhead will be more as compared with single user SISO. correlated), the authors in [122, 148] developed an effective
To handle this, some RIS channel characteristics (such as spa- method of grouping adjacent RIS elements into a sub-surface,
tial correlation, sparsity, and low rank) can be used to make referred to as RIS element grouping; consequently, only the
the cascaded channel division easier and decrease the training effective cascaded user RIS-BS channel connected with each
overhead [144]. The authors in [145] used deep learning and sub-surface needs to be calculated, significantly decreasing the
searching algorithms to accelerate the training for channel esti- training overhead and simplifying RIS reflection configuration
mation. The authors in [146] exploited the fact that all users for transmitting data. However, it is crucial to mention that the
share the same RIS-BS channel in the uplink to get minimum channel coherence time has a significant impact on the appro-
(U −1)N
training overhead U + N + max(U − 1, [ ]) where U is priate grouping ratio for practical implementation. The upward
AB
the number of active users and AB is the number of antennas and downward trend of the achievable data rate in Figure 21 is
at base station. Moreover, [147] utilized the redundancy of hav- an obvious example of the effect of the grouping ratio on the
ing more OFDM subcarriers than the number of delayed paths data rate at a high SNR regime.
522 HASSOUNA ET AL.

respective channels to RIS, whereas in the second mode or

called reflection or transmission mode, the sensors are deacti-
vated, and the RIS reflecting elements are turned on to reflect
the data signals from the BS or users for improving downlink
and uplink communication in both.
The authors in [168] presented a new deep reinforcement
learning framework for estimating RIS reflection matrices by
teaching the RIS how to anticipate optimum interaction matri-
ces using sampling channel knowledge on its own. Unlike
supervised learning-based methods, this technique does not
require an initial dataset gathering step.

FIGURE 21 Achievable rate versus RIS Grouping [148]

4.3 Discussions and insightful prospect for
section IV
Similar to [122, 148] the authors in [151, 152] stud- ∙ In summary of the above section, a detailed explanation for
ied the channel estimation for the broadband system but the various types of channel estimation in different wireless
for multi-users. The authors in [151] proposed two efficient communication system environments was presented. Table 6
channel estimate algorithms for various channel configura- summarize to some extent good literature of channel esti-
tions in an RIS-assisted multi-user broadband communication mation taking into consideration different communication
system using orthogonal frequency division multiple access parameters. The pilot transmission and channel estimation
(OFDMA). The sequential-user channel estimation (SeUCE) are not fully covered in the literature.
scheme can sustain more users for channel estimation than
the simultaneous-user channel estimation (SiUCE) scheme by
leveraging the advantage that all users have the same RIS-AP 5 RIS DEPLOYMENT
channel, but at the cost of increased channel estimation com-
plexity and some decayed channel estimation performance. The 5.1 Why RIS location is important?
authors in [152] describes a compressive sensing (CS)-based
CE solution for RIS-aided mm-Wave massive MIMO systems, RIS works in FD mode with just passive reflection, avoid-
whereby the angular channel sparsity of large-scale array at mm- ing amplification, processing noise and self-interference. These
Wave is used for improved CE with reduced pilot overhead. The appealing features of RIS have prompted extensive research
authors develop pilot signals based on the previous knowing of into using it to largely enhance the performance of wire-
the line-of-sight-dominated transmitter-to-RIS channel as well less systems in a variety of scenarios, including multi-user
as previous experience of the high-dimensional transmitter-to- NOMA, WPT, physical-layer security, multi-carrier communi-
receiver and RIS-to-receiver channels using compressed sensing cations, multi-antenna, and MEC [3, 6, 13]. In the existing
techniques. To take advantage of the channel sparsity, a dis- literature, the RIS is usually deployed near the distributed users
tributed orthogonal matching pursuit method is used. As a for the purpose of enhancement the local coverage which is
result, several writers [153–159] leverage RIS channel features completely different from that for the active relay, almost placed
(such as low-rank, sparsity, and spatial correlation) to simplify in the middle of the transmitter and receiver for balancing the
cascaded channel decomposition and decrease training over- SNRs of the two-hop links, that process and amplify the source
head in RIS-assisted single-user MIMO/MISO systems. On the signal before forwarding it to the receiver. Alternatively, the
other hand, the authors in [160–163] used the same channel other deployment method is to position the RIS near the base
properties to investigate the channel estimation in the multi-user station bearing in mind that both methods of deployments min-
MISO/MIMO systems. imize the product distance path loss as per Figures 23a and 23b.
To speed up the training process in passive RIS-aided MIMO It’s worth mentioning that RIS might be deployed much more
and MISO systems, deep learning and hierarchical search- widely across the network, due to its cheap price, to effectively
ing algorithms have been developed for channel estimation modify signal propagation. However, this creates a consider-
[164–167]. ably larger-scale deployment optimization challenge that is far
Different from the above-aforementioned works, the authors more difficult to address in addition to the fact that because
in [168–171] exploited some sensing elements interlaced with RIS is passive equipment, the strength of their reflected signals
the passive RIS elements to provide RIS with sensing capabili- decays fast with distance; consequently, if RISs are placed far
ties for the channel estimation process as per Figure 20. Unlike enough away, reciprocal interference is essentially non-existent,
the fully passive RIS channel estimation, the nearly passive RIS considerably simplifying their deployment design so, thanks to
has two modes of operation. All the reflecting elements are kept the RIS passivity.
OFF in the first mode or called sensing mode, moreover the Another important issue is that for a RIS to be genuinely suc-
sensing elements are powered up to receive pilot signals from cessful, it must be properly deployed. It should be installed at a
the BS or users in the downlink or uplink for estimating their place with LoS to the access point, and it may then be set to
HASSOUNA ET AL. 523

TABLE 6 Summary of existing work of channel estimation methods and techniques

Practical
phase shift
Reference Communication setup model RIS passivity Channel estimation protocol

[149] Multi-user MISO Discrete Full Channel estimation procedure based on the minimum mean
squared error (MMSE)
[162] Point to Point MISO Discrete Full LS CE protocol with binary-reflection (full or no)
[88] Single user RIS-enhanced OFDM Continuous Full Interpolation based on DFT/IDFT to estimate the channel
[87] Single user RIS based OFDM Discrete Full The RIS’s on-and-off state control of reflecting elements
[151] Multi-user OFDMA Uplink Continuous Full The sequential-user channel estimation
[152] Multi-user mm-Wave massive MIMO Continuous Full Compressive sensing (CS)-based CE solution
[153] Single user mm-wave MISO Continuous Full Compressed-sensing-based channel estimation
[154] Single-userLIM-assisted massive Continuous Full a two-stage mixed channel estimate. The JBF-MC method (joint
MIMO system bilinear factorization and matrix completion)
[155] Single-user RIS-assisted MIMO Continuous Full a tensor model with parallel factors (PARAFAC) that may be used to
system estimate the communication channels involved
[156] Single-user RIS-enabled MIMO Continuous Full A bilinear adaptive vector approximate message passing
system (BADVAMP) method with a traditional training uplink technique
[157] Single-user RIS-aided mm-Wave Continuous Full The channel parameters are successively estimated using an iterative
MIMO system reweighted technique
[158] RIS assisted SISO Continuous Full a low complexity channel information acquisition method using the
channel sparsity and the position of the UE
[159] Single-user MISO Continuous Full Method for channel estimate based on compressed sensing
[160] A RIS aided multi-user MIMO system Continuous Full CS approaches are used to solve a sparse channel matrix recovery
problem
[161] An RIS aided multi-user MIMO Continuous Full a dual-link pilot transmission scheme
system
[162] Multi-user MIMO Continuous Full a matrix-calibration based sparse matrix factorization
[163] Multi-user MISO Continuous Full a method based on the parallel factor (PARAFAC) decomposition
[164] Single-user massive MIMO assisted Continuous Full Design of a hierarchical search codebook (low-complexity basis of
RIS beam training)
[165] Multi-user massive MIMO assisted Continuous Full a beam training-based cooperative channel estimation approach for
RIS RIS-assisted MIMO systems
[166] Single-user RIS assisted MISO Continuous Full Deep Learning DL-based detector, called (Deep RIS) for channel
estimates
[167] Multi-user mm-Wave massive MIMO Continuous Full Deep learning bases scheme (A twin convolutional neural network
(CNN))
[168–170] Single-user OFDM based system Continuous Nearly passive Deep Reinforcement Learning and compressive sensing Based RIS
[171] Single user SISO Discrete Nearly Passive a method for explicit channel estimation that uses alternating
optimization

substantially boost the information rate for users inside LoS. 5.2 RIS deployment: The sate-of-the art
The authors in [139] compared the cases of LoS and NLoS
between the transmitter and the RIS. By considering the In this part of the paper, we look at the new RIS deployment
strongest tap maximization approach to select the value of 𝜃l problems in a variety of scenarios to get valuable insights into
that maximizes the magnitude of each channel tap l : practical design, beamforming performance, and coverage.
The authors in [172] tackles the critical topic of how to place
| T | RISs in a wireless communication network for maximum per-
𝜃𝜄 = argmax|hd [𝜄] + Heq𝜄 𝜃|, 𝜄 = 0, … , M − 1, (12)
𝜃
| | formance. In terms of different communication performance
metrics, the two traditional techniques of installing RIS at the
where M is the channel taps. The authors proved in the case of BS or at distributed users are evaluated, and then suggest a
LoS that the RIS can boost the data rate by 2.7–2.9 times, which novel hybrid RIS deployment approach that combines their
make a huge improvements when there is NLoS case. cooperative benefits. An inter-RIS reflective link between both
524 HASSOUNA ET AL.

FIGURE 22 RIS Deployment methods, (a) Centralized RIS Deployment, (b) Distributed RIS Deployment

FIGURE 23 The three typical users’ achievable rates, as well as their minimum rate, given various RIS deployment techniques [172], (a) System simulation
setup,(b) Achievable Rate versus Deployment method

the BS-side RIS and each of the user-side RIS is included in than one RIS near the users or base stations and to over-
the hybrid RIS installation proposal. When both direct and come the standalone RIS which suffers from gain loss if it is
single-reflection links between the BS and its served users are located far away from the base station or user equipment. As
significantly obstructed, the double-reflection links can be lever- per Figure 24, in the case of standalone RIS, when situating the
aged to provide alternative accessible LoS pathways between RIS near the user or AP it provides the highest SNR however,
them. The double-reflection path may achieve a significantly placing it in the midway between the user and the AP yields the
larger asymptotic passive beamforming gain than the single- lowest SNR meanwhile the two cooperative RIS case obtains a
reflection channel Given the same total number of reflecting significant SNR boost When compared to the highest SNR in
elements N despite the fact that N increase of (N 4 ) versus the single-RIS scenario.
(N 2 ). The simulation results show the superior performance The authors in [173] contribute to current research by
of the proposed deployment over the conventional ones as per proposing and assessing a wireless communication system with
Figure 23. Because all users may be served by at least one the double-RIS communication system. On the reasonable
RIS, located whether user or BS side, under the hybrid RIS premise that the reflection channel from the first RIS to the
deployment, it demonstrates its increased efficacy in terms of second RIS is of rank one, a combined passive beamform-
network coverage. ing design for the two RISs was developed. A power increase
The authors [173–176] employed the cooperative RIS of order (N 4 ), may be achieved by deploying two coopera-
method in their work to reap the advantages of placing more tive RIS with a total of N components, which is superior than
HASSOUNA ET AL. 525

and when BS has a larger transmit power budget, the proposed

multi-RIS with continuous phase shifts outperforms the sin-
gle RIS with continuous phase shifts in terms of WSR. This
is because, in multi-RIS systems, having the RIS close to the
users minimises pathloss propagation, allowing the users’ data
rate to increase due to the RIS’ passive beamforming gain.
This exemplifies the value of using several RISs to increase
system performance.
More inter-RIS reflections are created as the number of
RISs in the User-AP connection grows, resulting in increased
pathloss, backhaul cost, and complexity in the aforemen-
tioned activities. However, bigger multiplicative beamforming
benefits may be obtained by performing cooperative passive
beamforming across a large number of RISs.
The researchers in [80, 177] considered two workable RIS
FIGURE 24 Received SNR versus d(m) [3] deployment methods that correspond to various effective chan-
nels between users and the AP: scattered deployment, in which
the N elements form two RISs, each placed near one user,
installing one ordinary RIS with a power gain of level (N 2 ). and centrally managed deployment, in which all of the N
Simulated findings demonstrate that establishing two coopera- elements are deployed near the AP. The capacity and achiev-
tive RISs performs much better than deploying one RIS with able rate regions for both deployment methods using various
the same total number of elements. The simulation shows that multiple-access algorithms are calculated considering the uplink
when two cooperative RISs are deployed, their performance is multiple access channel (MAC). It is shown that in symmet-
lower than when only one RIS is deployed. However, when the ric channel settings, the centrally controlled deployment beats
total number of elements is big, such as N = 1600, using two the scattered deployment in terms of achievable user rates and
cooperatives’ RISs instead of one can result in substantial per- capacity region. By exploiting the MAC broadcast (BC) duality,
formance gains. Another point to notice is that increasing the the results were extended to the downlink RIS-aided broadcast
total number of elements by a factor of two, for example, from channel BC, where the performance advantage of centralized
N = 800 to N = 1600, raises the received SNR of the bench- over distributed RIS deployment was also proven to be true.
mark case with one RIS by 6 dB, while increasing the received The advantages of centralized vs scattered RIS deployment
SNR of the two-RIS case with N1 = N2 = N ∕2 by 12 dB. is particularly apparent when the two users have asymmetric
Similar to [173], the authors in [175] investigate effec- rate demands and channel conditions, according to numerical
tive channel estimation and passive beamforming solutions results. However, in practical scenarios, the assumption of the
for a single and multiuser communication system helped by uplink and downlink channel duality for both RIS deployments
a double-RISs. In [174] the authors consider a cooperative may change in terms of the LoS and the NLoS which it will by
passive beamforming design for a double-RIS assisted multi- return impact performance gain of the system.
user MIMO that catches the inter-RIS channel’s multiplication The uplink power control of an RIS-aided IoT network
beamforming gain. The constructive received beamforming at under the QoS limitations at each user is investigated in [178].
the BS and the cooperative passive reflected beamforming at the The objective is to reduce total user power by simultaneously
two distributed RISs, positioned near the BS and users, respec- optimizing RIS reflecting element phase shifts and receiving
tively, are optimised in a general channel structure with both beamforming at the BS, while considering each user’s unique
dual and single-reflection links to increase the minimum SINR SINR restriction. The authors used the environment diversity
of all users equipment. In the simulation, the maximum power by installing several RISs to produce considerable improve-
P = 30 dBm was adopted to investigate the influence of channel ments in energy efficiency through combined optimization of
ranking and spatially multiplexing gain on system performance RIS phase shifts and reception beamforming. Specifically, Each
for multi-users and found the superiority of the doubled RIS RIS is made up of N reflecting components that may indepen-
over the single in relation to achievable max-min rate data rate dently reflect the incident signal with an adjustable phase shift.
for the users. The simulation results reveal that when the number of RIS units
The researchers in [176] proposed a wireless network con- rises from 1 to 8, the Riemannian manifold based alternating
sists of multiple reconfigurable intelligent surfaces RISs, which optimization (RM-AO) algorithm saves about 4 dBm transmit
would help transmission between a multi-antenna base sta- power in the single-user situation. Furthermore, as the num-
tion and a large number of single-antenna cell-edge users. The ber of users grows, more transmission power can be conserved.
authors aim to optimise the weighted sum rate of all cell-edge Despite the above-mentioned efforts and advancements in link-
users by adjusting the BS’s transmit beamforming and RIS’s level performance optimization for different RIS-aided wireless
phase shifts together. For several schemes, the simulation results systems, the large-scale deployment of RISs in large size wire-
display the transmit power versus the weighted sum rate (WSR). less networks require methods and tools to optimize the huge
To begin, all systems’ WSR grows in lockstep with BS power, deployments of the multiple RIS units in the wireless network.
526 HASSOUNA ET AL.

TABLE 7 Summary of existing work of RIS deployment

Reference RIS deployment method Users Achievement

[168] Hybrid by adopting IA-MS design (distributed near the Multi-users SISO Minimum data rate significant improvement over the methods of
users and base station) placing the RIS near users or near the base station
[173] Cooperative double RIS (near base station and user) Single-user SISO Improvement of the power gain of order(N 4 ) instead of (N 2 )
[174] Cooperative double RIS (near base station and users) Multi-user MIMO maximize the minimum SINR among all users
[175] Cooperative double RIS (near base station and user) Single-user SISO The training overhead and channel estimate error are taken into
consideration, resulting in significant rate improvement
[176] Four numbers of RIS distributed uniformly at the cell Multi-user MISO maximisation of the cell-edge users’ weighted sum rate
edge of the base station vicinity
[80, 177] Both Distributed and Centralized Two users SISO the centralized deployment beats the distributed deployment in terms
of possible user rates Under symmetric channel setups
[178] Multiple RIS near the users Multi-user SIMO By combining the phase shifts of RIS reflecting components with
receiving beamforming at the BS, the overall user power is reduced
[179] Hybrid active and passive OFDMA wireless network Multi-user maximizes the hybrid network throughput

The authors in [78, 179, 180] offer an analytical framework coordinated signal focusing and interference elimination at
for the RIS assisted hybrid network relied on stochastic geom- the RIS’s location. Meanwhile, whether or not an associated
etry while the authors in [181, 182] rely on machine learning RIS is located near every user, the RIS passive reflections
algorithms in resolving the problems joint RIS deployments, must be built in tandem with the BSs or users’ transmissions
phase shift design, and power allocation in a MISO NOMA net- in order to enhance their end-to-end communications across
work to increase energy efficiency while considering the data the RISs re-designed wireless channels.
needs of each individual user. 2) Lacking RF chains makes it difficult to obtain the CSI
between RIS and its feeding BSs or users, which is necessary
for the RIS reflection optimization mentioned earlier. This
5.3 Discussions and insightful prospect for is especially true given that RIS typically has a high percent-
section V age of reflecting elements and consequently related channel
coefficients to calculate.
Table 7 summarize the different RIS deployments for various 3) Due to their different array structures, passive versus active,
communication system models for single and multiple users in operating mechanisms, and reflect versus transmit or receive,
addition to the purpose and the achievement of the RIS deploy- the best possible implementation method for RISs in wire-
ments. It was shown that different deployments methods can less networks to achieve maximum network capacity is
achieve tradeoff between performance improvements from one observed to be extremely different from that for traditional
side and complexity/budget/training overhead/optimization wireless networks with active BSs, APs, and relays, and
from the other side. Moreover, the dimentional size (number of thus needs to be thoroughly re-tested. In conclusion, inte-
RIS elements) of RIS and the LoS path bewteen the transmitter grating RISs into wireless networks effectively brings both
and the RIS are essential parameters that should be taken into new possibilities and problems, both of which need further
account in the RIS designs and deployments to ensure achieving exploration.
performance gain and better channel conditions.
In this section, we will try to present some challenges
of the RIS that we think that still need more studying and
6 FUTURE RESEARCH DIRECTIONS investigations.

Various techniques, examined in this research, represent that ∙ Scaling laws and beyond far-field regime: When narrow-
RIS-assisted wireless networks can strongly improve the band capacity increased, the SNR with an RIS grows as
received signal power, boost the capacity and sum rate, expand N 2 G 𝜃V at the time when N paths have the same propa-
network coverage, minimize transmit power, reduce interfer- gation loss [16]. The quadratic SNR scaling does not imply
ence, and provide better security layer and Quality of service that a higher SNR may be achieved than if the RIS were
supply to multiple users. However, from a communication replaced with an equitable antenna array broadcasting at the
aspect, the design of RIS-aided wireless communication has same power level as the RIS. The SNR would be propor-
novel and unique issues, which are outlined below: tional to NV in the latter scenario. The SNR scaling can be
factorized obtained by the RIS as (NG 𝜃)(NV ) to recognize
1) The passive reflections of all reflective elements at each the difference. The first term accounts for the proportion of
RIS must be constructed in such a way that they enable the transmitter’s signal power reflected by the RIS, which is
HASSOUNA ET AL. 527

when the RIS is densified cannot be determined without ade-

quate modelling of the mutual impedance. As a result, to
capture the behaviour of RISs with closely spaced compo-
nents, new modelling approaches that do not rely on the
canonical minimum-scattering (CMS) assumption must be
devised. Machine learning methods could be useful to solve
the problem of system identification if accurate models are
difficult to develop [16]. The mutual coupling will affect
both algorithmic design from one side and communication
and localization performance from the other side. The effect
of Mutual coupling on RIS-aided IoT-based applications is
inevitable. The RIS is most beneficial when the number of
its elements is large however when the surface is quite big
(e.g. 4096 elements, i.e. 2Mtrs X 2Mtrs), the mutual coupling
between the adjacent elements will impact the actual response
of the reflection coefficients of all elements which will end
FIGURE 25 The frequency response adjustment when employing a eventually in losing the reward of RIS technology in beam-
varactor to change the effective capacitance
forming or steering the beam direction towards the intended
receiver. For example, if the RIS is deployed in LoS with the
energy transmitters and receivers in addition to utilising their
a relatively tiny amount even when N is big. As a result, the big aperture and excellent passive beamforming gains, the
RIS can’t get a higher SNR than NV . Consequently, the RIS received power of neighbouring IoT devices can be greatly
must be large to be competitive with another enabling tech- improved nevertheless, such power enhancement cannot be
nology [183]. Far-field assumptions have been used in the achieved when mutual coupling affecting the beamforming
majority of current research on RIS-enabled SRE. RISs, on capability of the RIS.
the other hand, can be composed of a few square metres ∙ Electromagnetic Interference (EMI): Many researches
of geometrically large surfaces. This means that inappropri- are ignoring the electromagnetic interference (EMI) that is
ate application settings, such as interior environments, RISs unavoidably present in any environment and instead focus
can function in the near-field regime. The usage of large RISs only on the signals produced by the system [186]. The EMI
allows for the creation of novel wireless networks that func- may result from a range of natural, purposeful, or uninten-
tion in the near-field domain, which is not a common design tional factors, such as man-made devices and background
assumption in wireless communications since the RIS ele- radiation from the environment. In general, every unregu-
ments incur substantial variations in propagation loss due to lated wireless transmission generates EMI [187]. The energy
the different propagation distances and angles. More signifi- of the EMI waves that strike the RIS in the space in front
cantly, the near field allows the RIS to focus signals not only of it is proportional to its area. When the EMI energy is
in a certain direction but also at a specific location within radiated again by the RIS, it will reach the intended receiver
that direction, making it superior to a mirror [184]. This fea- resulting in degrading the end-to-end SNR of the wireless
ture can also be exploited for enhanced localization [185]. networks which are unaware of such interference effects. The
The basic performance restrictions, design requirements, and SNR decline is attributed to the fact that such wireless net-
possible uses and advantages of near-field transmissions in works/systems are simply designed to deliberately act against
RIS-enabled SREs have received little attention thus far. The only the thermal noise generated by the receiver. As a result,
near-field regime is deserving of greater research because the RIS must be aware of such damaging EMI effect by
of the possible uses that may be unlocked, such as highly designing beamforming RIS asisted algorithms that take into
focused capabilities. consideration the uncontrollable EMI. For example, many
∙ Mutual coupling: When the RIS sections on the substrate authors [188] studied the potential of an RIS assisted wireless
material are near together, separating them appropriately powered sensor networks (WPSN) where the RIS is deployed
is challenging. This causes mutual coupling, in which one to help multiple IOT devices to enhance their energy har-
element’s impedance is linked to the impedances of its neigh- vesting and data transmission capabilities with intelligently
bours. As a result, whereas Figure 25 depicts how a RIS adjustable phase shifts. Consequently, we will not get consid-
element acts in isolation, the frequency response will change erable improved throughput performance with the presence
based on the arrangement of the adjacent elements. The of EMI and mutual coupling effects.
mutual impedance is determined using full-wave simulations, ∙ Deployment of large-scale wireless networks: Due to
such as the method of moments, and is reliant on the phys- the extensive deployment of reconfigurable metasurfaces on
ical characteristics of the components. A RIS constructed diverse objects in the future smart radio environment, wire-
of patch or slot antennas effectively decouples the reflected less networks can benefit from a decentralized RIS system
wave’s amplitude and phase, providing complete two-phase with individually programmable RIS units. The great bulk
control. The trade-off between complexity and performance of previous research activities have focused on optimizing
528 HASSOUNA ET AL.

small scale networks, which are often networks consisting calculating the convex optimizations. For example, the
of a single RIS. This is a natural growth to start when authors in [192] optimizes beamforming without CSI at the
evaluating the potential advantages of new technology. Fur- BS and RIS by decreasing transmit power while maintaining
thermore, the evaluation of these fundamental conditions is a minimal signal-to-noise ratio SNR by proposing a particle
dependent on modelling hypotheses that may or may not be swarm optimization (PSO) technique which is a stochastic
realistic enough for the evaluation of sub-wavelength meta- optimization method inspired by some animal species’ social
materials. However, quantifying the performance constraints behaviour based on updating the varying velocity and posi-
of SREs in large-scale installations is critical. This creates a tions of the particles and then , for each particle, comparing
difficult scenario for real-time allocation and optimization the SNR at the user equipment with threshold to update its
of different RISs in dynamic and heterogeneous networks state and so on. While the researcher in [194] solved the opti-
to service diverse data streams. Individual transceivers can mization problem by proposing cosine similarity based low
adjust their operating parameters in response to the channel complexity algorithm to avoid the complex iterations and
state, which is determined by a stochastic model and can be the huge overhead needed when using convex optimization
predicted or estimated via a training procedure. Because of techniques. Despite, the proposed algorithm is not power-
the RIS’s reconfigurability, the radio environment becomes ful in comparison to convex, it is still very simple and does
programmable and non-stationary. As a result, understanding not require many iterations to adjust the phases of the RIS
the CSI via training becomes more challenging for individual elements.
transceivers. This suggests that the RIS-assisted networks, at ∙ RIS configuration under mobility: Many characteristics
least for the scattered RIS units, will be coordinated from a of communication systems have been investigated under
central location. The wireless environment becomes tractable the idea that the channel is approximately piecewise time-
and controlled as a result. To efficiently allocate and associate invariant, allowing the use of linear time-invariant (LTI)
RIS units to service numerous users at the same time, a joint system theory, in the majority of the research employing
control method is necessary [13]. Consequently, deploying the RIS to aid transmission however, the wireless channels
large density RISs in a big industrial plant or even in a city’s are time-variant because of the mobility of the transmit-
Centre to improve coverage probability or energy efficiency ter/receiver. The authors in [16, 195, 196] revealed the
are becoming necessary in future wireless systems. importance of using the RIS in reducing the delay spread
∙ Sensing and estimation of energy-efficient channels: and eliminating additional Doppler spread. However, the
The RIS’s supremacy is based on its ability to reconfigure results in [195] were based on hypothetical RISs which create
the phase shift of each scattering element based on the chan- specular reflections with a single and very large conducting
nel circumstances from the transmitter to the receiver. This elements. While in [196] the inclusion of statistical channel
necessitates channel sensing and signal processing capabili- model that accounts for atmospheric effects and line of sight
ties, which are difficult to come by without specific signal outage were not considered in the system model and opti-
processing capacity at the passive scattering components. mization of the RIS assisted LEO satellite communications.
Channel estimation in an RIS-assisted system is often per- Consequently, the literature is still lacking to be enriched with
formed at one side of the communication process, such as practical communication models that take into consideration
the BS with superior computing capabilities or the receiver real and practical assumptions and parameters.
end. In channel estimation section IV, we overview the dif-
ferent existing RIS channel estimate techniques, For instance, In the above-mentioned points, we tried to summarize some
the paper in [150], assumes that only one scattering element challenges that have been noticed while preparing this survey
is active at any one time, with the rest inert. An element- and the purpose of referring to them is to ensure that the
by-element ON/OFF-based channel estimation technique is research is still in its infancy and herculean efforts are required
fundamentally too costly for a sizable RIS with massive scat- to establish strong foundation research to create strong enabling
tering components. The RIS is underutilized because only RIS technologies for beyond 5G and 6 G wireless standards so,
a tiny percentage of the scattering components is active at this paper suggests some study directions for future investiga-
any given moment. This reduces the precision of the chan- tion based on the existing literature review and the thorough
nel estimate and causes a significant estimation delay. Despite system analysis of the present volume of researches.
the tremendous efforts exerted in [122, 151, 154, 189–191]
in creating robust channel estimation algorithms to reduce ∙ Machine learning for passive beamforming: Machine
the training overhead and enhancing the estimation accuracy learning techniques, rather than the alternating optimiza-
by deviating from the ON/OFF technique and proposing tion techniques often used in the literature, may be more
other effective ones, we think that still feasible, efficient, appealing to the RIS in terms of achieving flexible and light-
and long-term channel estimation methods are needed to weight phase management dependent on locally obtained
alleviate the pressure on requesting more energy consump- radio environment data. This might reduce the quantity of
tion for exchange of information, signal processing, and data shared between the RIS and active transceivers [13, 160,
computing. Consequently, the authors in [192–194] resolved 161, 173–175].
the optimization problems by considering schemes with- ∙ mm-Wave and THz communications: Mm-Wave com-
out the need for computing the channel state information munications cannot give gigabit-per-second data rates in the
HASSOUNA ET AL. 529

[210–212] in real time and in programmable way; and on the

other hand, they can directly modulate the digital informa-
tion. More recently, time-domain digital coding metasurfaces
have been developed, which increase an additional degree
of freedom to control the EM waves - the frequency spec-
tral distributions [213, 214]. In this manner, the general
space-time-coding digital metasurfaces have the capabilities
in engineering both spatial beams and the frequency spectra
simultaneously and independently, and directly modulating
the digital information [215, 216]. Hence, the digital coding
and programmable metasurfaces are also called as informa-
FIGURE 26 3G Metasurfaces [6] tion metasurfaces [217–219], which have set up a broad link
between the EM physical world and the digital world. The
future, enabling rate-demanding wireless applications (aug- information metasurfaces can be regarded as advanced RISs,
mented/virtual reality (AR/VR) and online gaming, because which will have deeper applications in wireless communi-
of the large bandwidth available at mm-Wave frequencies (i.e. cations. For example, the information metasurfaces can be
30–300 GHz). However, because of the growing figure of utilized to construct new architectures of the wireless com-
operating antennas and RF chains, mm-Wave communi- munication systems, in which the digital-analog converters,
cations have significantly higher energy consumption and mixers, RF devices, and antennas in the traditional transmit-
hardware costs [3]. Furthermore, mm-Wave communication ters are no longer required [220–226]; and the information
channels are more prone to obstruction and suffer from metasurfaces can also be used to develop the EM information
greater propagation loss overall. When RISs are installed theory to combine the Shannon information theory and the
between base stations and end-users, the RIS’s two key EM Maxwell’s equations, in which both EM information and bit-
features of reflection and refraction may be leveraged to stream information are considered simultaneously [227, 228].
address the critical problem of dead spots by establishing line- Based on the information metasurfaces, ambient backscat-
of-sight channels between the AP and the user equipment, ter communications were demonstrated by manipulating the
allowing for high SE and data rates [197–199]. It is predicted commodity Wi-Fi signals, making it possible to realize secure
that the RIS will be integrated with the beyond 5G, Figure 26 wireless communications without any active radio compo-
, networks and the current massive MIMO and the mm-wave nents [224]. In the future, we expect that the information
and THz networks to enhance the network coverage and metasurfaces, or the advanced RISs, will push revolutionary
increase the capacity. advances of the wireless communications in basic theories,
∙ Multiple access MEC: The core premise of the MEC is system levels and communication security for future wireless
that network congestion is minimized, and applications func- generations.
tion better when applications and related processing activities
are executed closer to consumers. The MEC technology is The literature is full of interesting and promising future
intended for use at cellular base stations and other edge research directions, however, this study concentrates on the
nodes. It helps clients to deploy new applications and ser- conceptual and the techniques of analysis and optimization for
vices in a flexible and timely manner, in addition, to assisting RIS more than investigating each RIS assisted application indi-
in offloading computing workloads from mobile devices to vidually.
reduce latency and energy consumption. Some academics
have been looking into the use of RISs in this setting recently
[200–204] to enhance the computing performance gains and 7 CONCLUSION
optimize the computing and communication operations.
∙ Advanced applications of RISs in wireless communi- This paper provided a thorough overview of the RIS’s archi-
cations: In the above discussions, the applications of RISs tecture and uses in wireless communication networks. In the
in wireless communications are focused on controlling the beginning, we have presented the RIS principles to throw new
wireless channels and improving the wireless environment light on the RIS design and different types of control mech-
by reconfiguring the beam directions and beam coverages, anisms that show the applied tuning methods used in various
in which the metasurfaces are not involved in the digital communication models. In addition to the channel correlation
information modulations. Recently, digital coding and pro- fading and practical pathloss models which characterize the sig-
grammable metasurfaces have been presented [205–212], nal and channel model in RIS-aided communications. Then, the
in which the metasurfaces are characterized digital states optimization frameworks and performance analysis methodolo-
and controlled by FPGA. On one hand, the digital coding gies for RIS, are discussed. The methods have shown promises
and programmable metasurfaces can manipulate the beam in increasing the spectral efficiency of wireless networks because
directions, beam coverages, number of beams [205–207], of their capacity to modify the behaviours of the interacting
polarization states [208], and the waveforms like cone EM waves through intelligent manipulations of the reflections
beams [209] and orbital-angular-momentum vortex beams phase shifts under different wireless communication scenarios,
530 HASSOUNA ET AL.

including the SNR/data rate/secrecy rate maximization, trans- PWE Programmable Wireless Environments
mit power minimization and EE/SE maximization. The paper SMM Spatial Microwave Modulators
afterwards gives various relevant RIS deployments strategies FSO Free Space Optical
and channel estimation algorithms to teach the RIS how to rec- MU Multiuser
ognize the surrounding environments and enhance the system UAV Unmanned Aerial Vehicle
performance despite its passiveness. Finally, this work paves the MEC Mobile Edge Computing
way for some research limitations and future directions. FPGA Field Programmable Gate Array
This survey encompasses a wide range of research top- MEMS Micro Electromechanical Systems
ics related to RIS and its uses in wireless communication, HIS High Impedance Surfaces
from its physical tuning, channel modelling, digital coding and LOS Line of Sight
information modulations to research issues from a wireless PEC Perfect Electric Conductor
communication viewpoint, with a key goal on the optimization IID Independent and identically distributed
techniques and solution approaches for RIS-assisted wireless VBLAST Vertical Bell Labs Layered Space Time
systems. The presented methods in the current research are NOMA Non-Orthogonal Multiple Access
still limited to certain assumptions and dropping some practical MRT Maximum Ratio Transmission
parameters and limitations. For example, the mutual coupling LIS Large Intelligent Surfaces
between the adjacent RIS elements, the impractical phase shift ZF Zero Forcing
resolution models that describe the reflection coefficients of all MMSE Minimum Mean Square Error
RIS elements and the uncontrollable EM interference are vital, CIR Channel Impulse Response
crucial and essential factors in limiting and degrading the RIS SCA Successive Convex Approximation
performance. In addition to the scaling laws in the near/far- BCD The Block Coordinate Descent
field regime and the time-variant channel mobility which are MSE Mean Square Error
still required to be explored thoroughly due to their signifi- AO Alternating Optimization
cant effects on the performance of RIS-aided communication. SDR Semidefinite Relaxation
The current solutions mentioned in the literature is depending STM Strongest Tap Maximization
mainly on the alternating optimization methods and group- GMD Geometric Mean Decomposition
ing the elements to reduce the overhead and guarantee the LS Least Square
convergence of the algorithms to locally optimal solutions nev- CE Channel Estimate
ertheless, the computational complexity and the elapsed run SeUCE Sequential User Channel Estimation
time are still substantially high. In the future work, by analyz- SiUCE Simultaneous User Channel estimation
ing low complex and run time RIS-based algorithms using the DFT/IDFT Discrete Fourier Transform
machine learning methods for acceleration the phase manage- PARAFAC Parallel Factor
ment, the RIS-aided wireless systems are predicted to attain JBF MC Joint Bilinear and Matrix Completion
not only a higher performance gain but also it will establish BADV AMP Bilinear Adaptive Vector Approximate Message
for more refined and realistic communication models. Fur- Passing
thermore, the RIS must be aware of the mutual coupling, FD Full Duplex
the uncontrollable electromagnetic interference and the prac- WPT Wireless Power Transfer
tical phase shift models for the RIS coefficients to guarantee IoT Internet of Things
remarkable and optimum RIS performance outcome in their MISO Multiple-Input Single-Output
beamforming algorithms. Besides, the RISs provide LOS chan- SISO single-input and single-output
nels which can be exploited to enhance the coverage in the RF Radio frequency
dead spot zones of the mm-wave communication channels and BER Bit Error Rate
the MEC computing performance gains in congested networks. UE User Equipment
However, the research of RIS-enhanced networks is still full of SNR Signal-to-noise-ratio
challenges and future directions not only from controlling the SINR Signal-to-Interference-plus-Noise ratio
wireless channel and the smart radio environment perspective QAM Quadrature Amplitude Modulation
but also when the RIS is involved in digital coding and informa- BPSK Binary Phase Shift Keying
tion modulations which will push for revolutionary advances in SEP Symbol Error Probability
security communication based RIS systems. CR Cognitive Radio
AP Access Point
SU/PU Secondary/Primary Users
LIST OF ABBREVIATIONS BS Base Station
PIN A Positive-Intrinsic-Negative
SRE Smart Radio Environments LoS Line of Sight
AF/DF Amplify Forward/Decode Forward EM Electromagnetic
CSI Channel State Information MRC Maximum Ratio Combining
FSS Frequency Selective Surface FDD/TDD Frequency/Time Division Duplexing
HASSOUNA ET AL. 531

AUTHOR CONTRIBUTIONS wireless environments. IEEE/ACM Trans. Netw. 27(4), 1696–1713

Saber Hassouna: Writing - original draft. Muhammad Ali (2019)
11. Dunna, M., Zhang, C., Sievenpiper, D., Bharadia, D.: Scattermimo:
Jamshed: Investigation, supervision. James Rains : Data cura-
Enabling virtual mimo with smart surfaces. In: Proceedings of the 26th
tion. Jalil ur Rehman Kazim: Data curation. Masood Ur Annual International Conference on Mobile Computing and Networking,
Rehman: Supervision. Muhammad abualhayja: Visualization. pp. 1–14. ACM, New York (2020)
Lina mohjazi: Visualization. Tiejun Cui: Conceptualization, data 12. Liu, Y., Liu, X., Mu, X., Hou, T., Xu, J., Di Renzo, M., Al-Dhahir, N.:
curation. Muhammad Imran: Supervision. Qammer Abbasi: Reconfigurable intelligent surfaces: Principles and opportunities. IEEE
Commun. Surv. Tutorials 23(3), 1546–1577 (2021)
Conceptualization, supervision, validation.
13. Gong, S., Lu, X., Hoang, D.T., Niyato, D., Shu, L., Kim, D.I., Liang, Y.-C.:
Toward smart wireless communications via intelligent reflecting surfaces:
A contemporary survey. IEEE Commun. Surv. Tutorials 22(4), 2283–
ACKNOWLEDGEMENT 2314 (2020)
This work was supported in parts by Engineering and Physical 14. Alghamdi, R., Alhadrami, R., Alhothali, D., Almorad, H., Faisal, A., Helal,
S., Shalabi, R., Asfour, R., Hammad, N., Shams, A., Saeed, N., Dahrouj,
Sciences Research Council (EPSRC) grants: EP/T021063/1.
H., Al-Naffouri, T.Y., Alouini, M.-S.: Intelligent surfaces for 6g wireless
networks: A survey of optimization and performance analysis techniques.
IEEE Access 8, 202795–202818 (2020)
CONFLICT OF INTEREST 15. Zhao, J.: A survey of intelligent reflecting surfaces (irss): Towards
The authors declare that there is no conflict of Interest. 6g wireless communication networks. arXiv preprint arXiv:1907.04789,
2019.
16. Björnson, E., Wymeersch, H., Matthiesen, B., Popovski, P., Sanguinetti,
DATA AVAILABILITY STATEMENT L., de Carvalho, E.: Reconfigurable intelligent surfaces: A sig-
nal processing perspective with wireless applications. arXiv preprint
Data sharing not applicable to this article as no datasets were
arXiv:2102.00742, 2021.
generated or analyzed during the current study. 17. Chen, H.-T., Taylor, A.J., Yu, N.: A review of metasurfaces: physics and
applications. Rep. Prog. Phys. 79(7), 076401 (2016)
18. Liu, F., Pitilakis, A., Mirmoosa, M.S., Tsilipakos, O., Wang, X.,
ORCID Tasolamprou, A.C., Abadal, S., Cabellos-Aparicio, A., Alarcón, E.,
Saber Hassouna https://orcid.org/0000-0002-3178-1677 Liaskos, C., Kantartzis, N.V., Kafesaki, M., Economou, E.N., Soukoulis,
Masood Ur Rehman https://orcid.org/0000-0001-6926-7983 C.M., Tretyakov, S.: Programmable metasurfaces: State of the art and
prospects. In: 2018 IEEE International Symposium on Circuits and
Qammer H. Abbasi https://orcid.org/0000-0002-7097-9969 Systems (ISCAS), pp. 1–5. IEEE, Piscataway (2018)
19. Turpin, J.P., Bossard, J.A., Morgan, K.L., Werner, D.H., Werner, P.L.:
Reconfigurable and tunable metamaterials: a review of the theory and
REFERENCES applications. Int. J. Antennas Propag. 2014, 429837 (2014)
1. 3GPP: First 5G nr specs approved. https://www.3gpp.org/news- 20. Costa, F., Monorchio, A., Talarico, S., Valeri, F.M.: An active high-
events/3gpp-news/1929-nsa. Accessed January 2022 impedance surface for low-profile tunable and steerable antennas. IEEE
2. Saad, W., Bennis, M., Chen, M.: A vision of 6g wireless systems: Applica- Antennas Wirel. Propag. Lett. 7, 676–680 (2008)
tions, trends, technologies, and open research problems. IEEE Network 21. Huang, D., Poutrina, E., Smith, D.R.: Analysis of the power dependent
34(3), 134–142 (2019) tuning of a varactor-loaded metamaterial at microwave frequencies. Appl.
3. Wu, Q., Zhang, S., Zheng, B., You, C., Zhang, R.: Intelligent reflecting sur- Phys. Lett. 96(10), 104104 (2010)
face aided wireless communications: A tutorial. IEEE Trans. Commun. 22. Katko, A.R., Hawkes, A.M., Barrett, J.P., Cummer, S.A.: Rf limiter meta-
(2021) material using pin diodes. IEEE Antennas Wirel. Propag. Lett. 10,
4. Samsung NewsRoom. (2020, July) The vision of 6g, samsungresearch. 1571–1574 (2011)
https://news.samsung.com/global/samsungs-6g-white-paper-lays- 23. Nayeri, P., Yang, F., Elsherbeni, A.Z.: Reflectarray antennas: theory,
out-the-companys-vision-for-the-nextgeneration-of-communications- designs, and applications. (2018)
technology. Accessed 11 July 2020. 24. Abeywickrama, S., Zhang, R., Wu, Q., Yuen, C.: Intelligent reflecting sur-
5. Goldsmith, A.: Wireless Communications. Cambridge University Press, face: Practical phase shift model and beamforming optimization. IEEE
Cambridge (2005). Trans. Commun. 68(9), 5849–5863 (2020)
6. Di Renzo, M., Zappone, A., Debbah, M., Alouini, M.-S., Yuen, C., 25. Kaina, N., Dupré, M., Lerosey, G., Fink, M.: Shaping complex microwave
De Rosny, J., Tretyakov, S.: Smart radio environments empowered by fields in reverberating media with binary tunable metasurfaces. Sci. Rep.
reconfigurable intelligent surfaces: How it works, state of research, 4(1), 1–8 (2014)
and the road ahead. IEEE J. Sel. Areas Commun. 38(11), 2450–2525 26. Tan, X., Sun, Z., Jornet, J.M., Pados, D.: Increasing indoor spectrum
(2020) sharing capacity using smart reflect-array. In: 2016 IEEE International
7. Subrt, L., Pechac, P.: Controlling propagation environments using intel- Conference on Communications (ICC), pp. 1–6, IEEE (2016)
ligent walls. In: 2012 6th European Conference on Antennas and 27. Hu, S., Rusek, F., Edfors, O.: The potential of using large antenna
Propagation (EUCAP), pp. 1–5. IEEE, Piscataway (2012) arrays on intelligent surfaces. In: 2017 IEEE 85th Vehicular Technology
8. Welkie, A., Shangguan, L., Gummeson, J., Hu, W., Jamieson, K.: Pro- Conference (VTC Spring), pp. 1–6. IEEE (2017)
grammable radio environments for smart spaces. In: Proceedings of the 28. Liaskos, C., Nie, S., Tsioliaridou, A., Pitsillides, A., Ioannidis, S., Akyildiz,
16th ACM Workshop on Hot Topics in Networks, 2017, pp. 36–42. ACM, I.: A novel communication paradigm for high capacity and security via
New York (2017) programmable indoor wireless environments in next generation wireless
9. Liaskos, C., Nie, S., Tsioliaridou, A., Pitsillides, A., Ioannidis, systems. Ad Hoc Networks 87, 1–16 (2019)
S., Akyildiz, I.: A new wireless communication paradigm through 29. Nie, S., Akyildiz, I.F.: Beamforming in intelligent environments based
software-controlled metasurfaces. IEEE Commun. Mag. 56(9), 162–169 on ultra-massive mimo platforms in millimeter wave and terahertz
(2018) bands. In: ICASSP 2020-2020 IEEE International Conference on Acous-
10. Liaskos, C., Tsioliaridou, A., Nie, S., Pitsillides, A., Ioannidis, S., Akyildiz, tics, Speech and Signal Processing (ICASSP), pp. 8683–8687. IEEE,
I.F.: On the network-layer modeling and configuration of programmable Piscataway (2020)
532 HASSOUNA ET AL.

30. Najafi, M., Schmauss, B., Schober, R.: Intelligent reflecting surfaces for 49. Khawaja, W., Ozdemir, O., Yapici, Y., Erden, F., Guvenc, I.: Coverage
free space optical communication systems. IEEE Trans. Commun. 69(9), enhancement for nlos mmwave links using passive reflectors. IEEE Open
6134–6151 (2021) J. Commun. Soc. 1, 263–281 (2020)
31. del Hougne, P., Fink, M., Lerosey, G.: Optimally diverse communica- 50. Usman, M., Rains, J., Cui, T.J., Khan, M.Z., Imran, M.A., Abbasi, Q.H.:
tion channels in disordered environments with tuned randomness. Nat. Intelligent wireless walls for contactless in-home monitoring. Light Sci.
Electron. 2(1), 36–41 (2019) Appl. 11(1), 1–13 (2022)
32. Karasik, R., Simeone, O., Di Renzo, M., Shitz, S.S.: Beyond max-snr: Joint 51. Cui, M., Zhang, G., Zhang, R.: Secure wireless communication via intel-
encoding for reconfigurable intelligent surfaces. In: 2020 IEEE Interna- ligent reflecting surface. IEEE Wireless Commun. Lett. 8(5), 1410–1414
tional Symposium on Information Theory (ISIT), pp. 2965–2970. IEEE, (2019)
Piscataway (2020) 52. Yu, X., Xu, D., Sun, Y., Ng, D.W.K., Schober, R.: Robust and secure wire-
33. Zappone, A., Di Renzo, M., Shams, F., Qian, X., Debbah, M.: Overhead- less communications via intelligent reflecting surfaces. IEEE J. Sel. Areas
aware design of reconfigurable intelligent surfaces in smart radio Commun. 38(11), 2637–2652 (2020)
environments. IEEE Trans. Wireless Commun. 20(1), 126–141 (2020) 53. Björnson, E., Özdogan, Ö., Larsson, E.G.: Intelligent reflecting sur-
34. Tang, W., Dai, J.Y., Chen, M.Z., Wong, K.-K., Li, X., Zhao, X., Jin, S., face versus decode-and-forward: How large surfaces are needed to beat
Cheng, Q., Cui, T.J.: Mimo transmission through reconfigurable intelli- relaying? IEEE Wireless Commun. Lett. 9(2), 244–248 (2019)
gent surface: System design, analysis, and implementation. IEEE J. Sel. 54. Sharma, S.K., Patwary, M., Chatzinotas, S., Ottersten, B., Abdel-Maguid,
Areas Commun. 38(11), 2683–2699 (2020) M.: Repeater for 5g wireless: A complementary contender for spec-
35. Kern, D., Wilhelm, M., Werner, D., Werner, P.: A novel design technique trum sensing intelligence. In: 2015 IEEE International Conference on
for ultra-thin tunable ebg amc surfaces. In: IEEE Antennas and Propaga- Communications (ICC), pp. 1416–1421. IEEE, Piscataway (2015)
tion Society Symposium, 2004, vol. 2, pp. 1167–1170. IEEE, Piscataway 55. Liu, R., Wu, Q., Di Renzo, M., Yuan, Y.: A path to smart radio envi-
(2004) ronments: An industrial viewpoint on reconfigurable intelligent surfaces.
36. Sheng, Z., Varadan, V.V.: Tuning the effective properties of metamate- IEEE Wireless Commun. 29(1), 202–208 (2022)
rials by changing the substrate properties. J. Appl. Phys. 101(1), 014909 56. Xu, C., Yang, L., Zhang, P.: Practical backscatter communication sys-
(2007) tems for battery-free internet of things: A tutorial and survey of recent
37. Lu, H., Zeng, Y., Jin, S., Zhang, R.: Enabling panoramic full-angle reflec- research. IEEE Signal Process Mag. 35(5), 16–27 (2018)
tion via aerial intelligent reflecting surface. In: 2020 IEEE International 57. Zhao, H., Shuang, Y., Wei, M., Cui, T.J., del Hougne, P., Li, L.:
Conference on Communications Workshops (ICC Workshops), pp. 1–6. Metasurface-assisted massive backscatter wireless communication with
IEEE, Piscataway (2020) commodity wi-fi signals. Nat. Commun. 11(1), 1–10 (2020)
38. Björnson, E., Sanguinetti, L.: Rayleigh fading modeling and channel hard- 58. Nemati, M., Ding, J., Choi, J.: Short-range ambient backscatter communi-
ening for reconfigurable intelligent surfaces. IEEE Wireless Commun. cation using reconfigurable intelligent surfaces. In: 2020 IEEE Wireless
Lett. 10(4), 830–834 (2020) Communications and Networking Conference (WCNC), pp. 1–6. IEEE,
39. Wu, Q., Zhang, R.: Towards smart and reconfigurable environment: Intel- Piscataway (2020)
ligent reflecting surface aided wireless network. IEEE Commun. Mag. 59. Di Renzo, M., Ntontin, K., Song, J., Danufane, F.H., Qian, X., Lazarakis,
58(1), 106–112 (2019) F., De Rosny, J., Phan-Huy, D.-T., Simeone, O., Zhang, R., Debbah, M.,
40. Yu, X., Xu, D., Schober, R.: Miso wireless communication systems via Lerosey, G., Fink, M., Tretyakov, S., Shamai, S.: Reconfigurable intel-
intelligent reflecting surfaces. In: 2019 IEEE/CIC International Con- ligent surfaces vs. relaying: Differences, similarities, and performance
ference on Communications in China (ICCC), pp. 735–740. IEEE, comparison. IEEE Open J. Commun. Soc. 1, 798–807 (2020)
Piscataway (2019) 60. Khaleel, A., Basar, E.: Reconfigurable intelligent surface-empowered
41. Huang, C., Zappone, A., Alexandropoulos, G.C., Debbah, M., Yuen, mimo systems. IEEE Syst. J. 15(3), 4358–4366 (2020)
C.: Reconfigurable intelligent surfaces for energy efficiency in wire- 61. Di Renzo, M., Song, J.: Reflection probability in wireless networks with
less communication. IEEE Trans. Wireless Commun. 18(8), 4157–4170 metasurface-coated environmental objects: An approach based on ran-
(2019) dom spatial processes. Eurasip J. Wirel. Commun. Netw. 2019(1), 1–15
42. Basar, E., Di Renzo, M., De Rosny, J., Debbah, M., Alouini, M.-S., Zhang, (2019)
R.: Wireless communications through reconfigurable intelligent surfaces. 62. Tan, X., Sun, Z., Koutsonikolas, D., Jornet, J.M.: Enabling indoor
IEEE Access 7, 116 753–116 773 (2019) mobile millimeter-wave networks based on smart reflect-arrays. In: IEEE
43. Tang, W., Chen, M.Z., Chen, X., Dai, J.Y., Han, Y., Di Renzo, M., Zeng, Y., INFOCOM 2018-IEEE Conference on Computer Communications,
Jin, S., Cheng, Q., Cui, T.J.: Wireless communications with reconfigurable pp. 270–278. IEEE, Piscataway (2018)
intelligent surface: Path loss modeling and experimental measurement. 63. Guo, C., Cui, Y., Yang, F., Ding, L.: Outage probability analysis and min-
IEEE Trans. Wireless Commun. 20(1), 421–439 (2020) imization in intelligent reflecting surface-assisted miso systems. IEEE
44. Tang, W., Chen, X., Chen, M.Z., Dai, J.Y., Han, Y., Di Renzo, M., Jin, S., Commun. Lett. 24(7), 1563–1567 (2020)
Cheng, Q., Cui, T.J.: Path loss modeling and measurements for reconfig- 64. Narayanan, A., Sreejith, T., Ganti, R.K.: Coverage analysis in millime-
urable intelligent surfaces in the millimeter-wave frequency band. arXiv ter wave cellular networks with reflections. In: GLOBECOM 2017-2017
preprint arXiv:2101.08607 (2021) IEEE Global Communications Conference, pp. 1–6. IEEE, Piscataway
45. Özdogan, Ö., Björnson, E., Larsson, E.G.: Intelligent reflecting surfaces: (2017)
Physics, propagation, and pathloss modeling. IEEE Wireless Commun. 65. Lu, B., Wang, R., Liu, Y.: Outage probability of intelligent reflecting sur-
Lett. 9(5), 581–585 (2019) face assisted full duplex two-way communications. IEEE Commun. Lett.
46. Di Renzo, M., Danufane, F.H., Xi, X., De Rosny, J., Tretyakov, 26(2), 286–290 (2021)
S.: Analytical modeling of the path-loss for reconfigurable intelli- 66. Wang, J., Zhang, W., Bao, X., Song, T., Pan, C.: Outage analysis for intel-
gent surfaces–anomalous mirror or scatterer? In: 2020 IEEE 21st ligent reflecting surface assisted vehicular communication networks. In:
International Workshop on Signal Processing Advances in Wireless GLOBECOM 2020-2020 IEEE Global Communications Conference,
Communications (SPAWC), pp. 1–5. IEEE, Piscataway (2020) pp. 1–6. IEEE, Piscataway (2020)
47. Garcia, J.C.B., Sibille, A., Kamoun, M.: Reconfigurable intelligent sur- 67. Ferreira, R.C., Facina, M.S., De Figueiredo, F.A., Fraidenraich, G., De
faces: Bridging the gap between scattering and reflection. IEEE J. Sel. Lima, E.R.: Bit error probability for large intelligent surfaces under
Areas Commun. 38(11), 2538–2547 (2020) double-nakagami fading channels. IEEE Open J. Commun. Soc. 1,
48. Ellingson, S.W.: Path loss in reconfigurable intelligent surface-enabled 750–759 (2020)
channels. In: 2021 IEEE 32nd Annual International Symposium on 68. Basar, E.: Transmission through large intelligent surfaces: A new frontier
Personal, Indoor and Mobile Radio Communications (PIMRC), pp. 829– in wireless communications. In: 2019 European Conference on Networks
835. IEEE, Piscataway (2021) and Communications (EuCNC), pp. 112–117. IEEE, Piscataway (2019)
HASSOUNA ET AL. 533

69. Assaf, T., Al-Dweik, A.J., El Moursi, M.S., Zeineldin, H., Al-Jarrah, M., Hélard, M., Gati, A., Sarrebourse, T., Di Renzo, M.: Single-carrier spatial
Exact bit error-rate analysis of two-user noma using qam with arbitrary modulation for the internet of things: Design and performance evalua-
modulation orders. IEEE Commun. Lett. 24(12), 2705–2709 (2020) tion by using real compact and reconfigurable antennas. IEEE Access 7,
70. Thirumavalavan, V.C., Jayaraman, T.S.: Ber analysis of reconfigurable 18 978–18 993 (2019)
intelligent surface assisted downlink power domain noma system. 88. Van Huynh, N., Hoang, D.T., Lu, X., Niyato, D., Wang, P., Kim, D.I.:
In: 2020 International Conference on COMmunication Systems & Ambient backscatter communications: A contemporary survey. IEEE
NETworkS (COMSNETS), pp. 519–522. IEEE, Piscataway (2020) Commun. Surv. Tutorials 20(4), 2889–2922 (2018)
71. Jung, M., Saad, W., Debbah, M., Hong, C.S.: Asymptotic optimal- 89. Guan, X., Wu, Q., Zhang, R.: Joint power control and passive beam-
ity of reconfigurable intelligent surfaces: Passive beamforming and forming in irs-assisted spectrum sharing. IEEE Commun. Lett. 24(7),
achievable rate. In: ICC 2020-2020 IEEE International Conference on 1553–1557 (2020)
Communications (ICC), pp. 1–6. IEEE, Piscataway (2020) 90. Wu, W., Wang, Z., Yuan, L., Zhou, F., Lang, F., Wang, B., Wu,
72. Li, J., Hong, Y.: Intelligent reflecting surface aided communication sys- Q.: Irs-enhanced energy detection for spectrum sensing in cognitive
tems: Performance analysis. In: 2021 IEEE 32nd Annual International radio networks. IEEE Wireless Commun. Lett. 10(10), 2254–2258
Symposium on Personal, Indoor and Mobile Radio Communications (2021)
(PIMRC), pp. 519–524. IEEE, Piscataway (2021) 91. Mahmoud, A., Muhaidat, S., Sofotasios, P., Abualhaol, I., Dobre, O.A.,
73. Salhab, A.M., Samuh, M.H.: Accurate performance analysis of recon- Yanikomeroglu, H.: Intelligent reflecting surfaces assisted uav commu-
figurable intelligent surfaces over rician fading channels. IEEE Wireless nications for iot networks: Performance analysis. IEEE Trans. Green
Commun. Lett. 10(5), 1051–1055 (2021) Commun. Networking 5(3), 1029–1040 (2021)
74. He, J., Wymeersch, H., Sanguanpuak, T., Silvén, O., Juntti, M.: Adap- 92. Al-Jarrah, M., Alsusa, E., Al-Dweik, A., So, D.K.: Capacity analysis of irs-
tive beamforming design for mmwave ris-aided joint localization and based uav communications with imperfect phase compensation. IEEE
communication. In: 2020 IEEE Wireless Communications and Net- Wireless Commun. Lett. 10(7), 1479–1483 (2021)
working Conference Workshops (WCNCW), pp. 1–6. IEEE, Piscataway 93. Al-Jarrah, M., Al-Dweik, A., Alsusa, E., Iraqi, Y., Alouini, M.-S.: On the
(2020) performance of irs-assisted multi-layer uav communications with imper-
75. Özdogan, Ö., Björnson, E., Larsson, E.G.: Using intelligent reflecting fect phase compensation. IEEE Trans. Commun. 69(12), 8551–8568
surfaces for rank improvement in mimo communications. In: ICASSP (2021)
2020-2020 IEEE International Conference on Acoustics, Speech and 94. Zhang, S., Zhang, R.: Capacity characterization for intelligent reflecting
Signal Processing (ICASSP), pp. 9160–9164. IEEE, Piscataway (2020) surface aided mimo communication. IEEE J. Sel. Areas Commun. 38(8),
76. Zhang, H., Di, B., Song, L., Han, Z.: Reconfigurable intelligent surfaces 1823–1838 (2020)
assisted communications with limited phase shifts: How many phase 95. Huang, C., Alexandropoulos, G.C., Yuen, C., Debbah, M.: Indoor
shifts are enough? IEEE Trans. Veh. Technol. 69(4), 4498–4502 (2020) signal focusing with deep learning designed reconfigurable intelligent
77. Alegría, J.V., Rusek, F.: Achievable rate with correlated hardware surfaces. In: 2019 IEEE 20th International Workshop on Signal Pro-
impairments in large intelligent surfaces. In: 2019 IEEE 8th Interna- cessing Advances in Wireless Communications (SPAWC), pp. 1–5. IEEE,
tional Workshop on Computational Advances in Multi-Sensor Adaptive Piscataway (2019)
Processing (CAMSAP), pp. 559–563. IEEE, Piscataway (2019) 96. Ye, J., Guo, S., Alouini, M.-S.: Joint reflecting and precoding designs
78. Lyu, J., Zhang, R.: Spatial throughput characterization for intelligent for ser minimization in reconfigurable intelligent surfaces assisted mimo
reflecting surface aided multiuser system. IEEE Wireless Commun. Lett. systems. IEEE Trans. Wireless Commun. 19(8), 5561–5574 (2020)
9(6), 834–838 (2020) 97. Huang, C., Zappone, A., Alexandropoulos, G.C., Debbah, M., Yuen,
79. Björnson, E., Optimizing a binary intelligent reflecting surface for ofdm C.: Reconfigurable intelligent surfaces for energy efficiency in wire-
communications under mutual coupling. arXiv preprint arXiv:2106.04280 less communication. IEEE Trans. Wireless Commun. 18(8), 4157–4170
(2021) (2019)
80. Zhang, S., Zhang, R.: Intelligent reflecting surface aided multi-user 98. Wu, Q., Zhang, R.: Beamforming optimization for wireless network aided
communication: Capacity region and deployment strategy. IEEE Trans. by intelligent reflecting surface with discrete phase shifts. IEEE Trans.
Commun. 69(9), 5790–5806 (2021) Commun. 68(3), 1838–1851 (2019)
81. Papazafeiropoulos, A., Pan, C., Elbir, A., Kourtessis, P., Chatzinotas, 99. Yu, X., Xu, D., Schober, R.: Miso wireless communication systems via
S., Senior, J.M.: Coverage probability of distributed irs systems under intelligent reflecting surfaces. In: 2019 IEEE/CIC International Con-
spatially correlated channels. IEEE Wireless Commun. Lett. 10(8), ference on Communications in China (ICCC), pp. 735–740. IEEE,
1722–1726 (2021) Piscataway (2019)
82. Jung, M., Saad, W., Jang, Y., Kong, G., Choi, S.: Performance analy- 100. Wu, Q., Zhang, R.: Intelligent reflecting surface enhanced wireless net-
sis of large intelligent surfaces (liss): Asymptotic data rate and channel work: Joint active and passive beamforming design. In: 2018 IEEE
hardening effects. IEEE Trans. Wireless Commun. 19(3), 2052–2065 Global Communications Conference (GLOBECOM), pp. 1–6. IEEE,
(2020) Piscataway (2018)
83. Hu, S., Rusek, F., Edfors, O.: Capacity degradation with model- 101. Chen, W., Ma, X., Li, Z., Kuang, N.: Sum-rate maximization for intelli-
ing hardware impairment in large intelligent surface. In: 2018 IEEE gent reflecting surface based terahertz communication systems. In: 2019
Global Communications Conference (GLOBECOM), pp. 1–6. IEEE, IEEE/CIC International Conference on Communications Workshops in
Piscataway (2018) China (ICCC Workshops), pp. 153–157. IEEE, Piscataway (2019)
84. Papazafeiropoulos, A., Pan, C., Kourtessis, P., Chatzinotas, S., Senior, 102. Guo, H., Liang, Y.-C., Chen, J., Larsson, E.G.: Weighted sum-rate max-
J.M.: Intelligent reflecting surface-assisted mu-miso systems with imper- imization for reconfigurable intelligent surface aided wireless networks.
fect hardware: Channel estimation, beamforming design. arXiv preprint IEEE Trans. Wireless Commun. 19(5), 3064–3076 (2020)
arXiv:2102.05333 (2021) 103. Huang, C., Zappone, A., Debbah, M., Yuen, C.: Achievable rate max-
85. Hu, S., Rusek, F., Edfors, O.: Beyond massive mimo: The potential of imization by passive intelligent mirrors. In: 2018 IEEE International
positioning with large intelligent surfaces. IEEE Trans. Signal Process. Conference on Acoustics, Speech and Signal Processing (ICASSP),
66(7), 1761–1774 (2018) pp. 3714–3718. IEEE, Piscataway (2018)
86. Basar, E., Wen, M., Mesleh, R., Di Renzo, M., Xiao, Y., Haas, H.: Index 104. Mu, X., Liu, Y., Guo, L., Lin, J., Al-Dhahir, N.: Exploiting intelligent
modulation techniques for next-generation wireless networks. IEEE reflecting surfaces in noma networks: Joint beamforming optimization.
Access 5, 16 693–16 746 (2017) IEEE Trans. Wireless Commun. 19(10), 6884–6898 (2020)
87. Phan-Huy, D.-T., Kokar, Y., Rachedi, K., Pajusco, P., Mokh, A., 105. Guo, H., Liang, Y.-C., Chen, J., Larsson, E.G.: Weighted sum-rate opti-
Magounaki, T., Masood, R., Buey, C., Ratajczak, P., Malhouroux-Gaffet, mization for intelligent reflecting surface enhanced wireless networks.
N., Conrat, J.-M., Prévotet, J.-C., Ourir, A., De Rosny, J., Crussière, M., arXiv preprint arXiv:1905.07920 (2019)
534 HASSOUNA ET AL.

106. Yu, X., Xu, D., Schober, R.: Optimal beamforming for miso communica- 127. El Bouanani, F., Muhaidat, S., Sofotasios, P.C., Dobre, O.A., Badarneh,
tions via intelligent reflecting surfaces. In: 2020 IEEE 21st International O.S.: Performance analysis of intelligent reflecting surface aided wireless
Workshop on Signal Processing Advances in Wireless Communications networks with wireless power transfer. IEEE Commun. Lett. 25(3), 793–
(SPAWC), pp. 1–5. IEEE, Piscataway (2020) 797 (2020)
107. Perović, N.S., Tran, L.-N., Di Renzo, M., Flanagan, M.F.: Achievable rate 128. Yue, X., Liu, Y.: Performance analysis of intelligent reflecting sur-
optimization for mimo systems with reconfigurable intelligent surfaces. face assisted noma networks. IEEE Trans. Wireless Commun. 21(4),
IEEE Trans. Wireless Commun. 20(6), 3865–3882 (2021) 2623–2636 (2021)
108. Yang, Y., Zheng, B., Zhang, S., Zhang, R.: Intelligent reflecting sur- 129. Lin, S., Zheng, B., Alexandropoulos, G.C., Wen, M., Chen, F., Smumtaz,
face meets ofdm: Protocol design and rate maximization. IEEE Trans. S.: Adaptive transmission for reconfigurable intelligent surface-assisted
Commun. 68(7), 4522–4535 (2020) ofdm wireless communications. IEEE J. Sel. Areas Commun. 38(11),
109. Björnson, E.: Optimizing a binary intelligent reflecting surface for ofdm 2653–2665 (2020)
communications under mutual coupling. arXiv preprint arXiv:2106.04280 130. Li, H., Cai, W., Liu, Y., Li, M., Liu, Q., Wu, Q.: Intelligent reflecting
(2021) surface enhanced wideband mimo-ofdm communications: From prac-
110. Yang, Y., Zhang, S., Zhang, R.: Irs-enhanced ofdma: Joint resource allo- tical model to reflection optimization. IEEE Trans. Commun. 69(7),
cation and passive beamforming optimization. IEEE Wireless Commun. 4807–4820 (2021)
Lett. 9(6), 760–764 (2020) 131. Yang, W., Li, H., Li, M., Liu, Y., Liu, Q.: Channel estimation for practi-
111. Yang, Y., Zhang, S., Zhang, R.: Irs-enhanced ofdm: Power allocation cal irs-assisted ofdm systems. In: 2021 IEEE Wireless Communications
and passive array optimization. In: 2019 IEEE Global Communications and Networking Conference Workshops (WCNCW), pp. 1–6. IEEE,
Conference (GLOBECOM), pp. 1–6. IEEE, Piscataway (2019) Piscataway (2021)
112. Björnson, E., Wymeersch, H., Matthiesen, B., Popovski, P., Sanguinetti, 132. Ohyama, T., Kawamoto, Y., Kato, N.: Intelligent reflecting surface (irs)
L., de Carvalho, E.: Reconfigurable intelligent surfaces: A signal pro- allocation scheduling method using combinatorial optimization by quan-
cessing perspective with wireless applications. IEEE Signal Process Mag. tum computing. IEEE Trans. Emerging Top. Comput. 10(3), 1633–1644
39(2), 135–158 (2022) (2021)
113. Nuti, P., Balti, E., Evans, B.L.: Spectral efficiency optimization for 133. Yang, Y., Zhang, S., Zhang, R.: Irs-enhanced ofdm: Power alloca-
mmwave wideband mimo ris-assisted communication. arXiv preprint tion and passive array optimization. In: 2019 IEEE Global Com-
arXiv:2201.01739 (2021) munications Conference (GLOBECOM) pp. 1–6. IEEE, Piscataway
114. Wu, Q., Zhang, R.: Intelligent reflecting surface enhanced wireless net- (2019)
work via joint active and passive beamforming. IEEE Trans. Wireless 134. Jiang, W., Chen, B., Zhao, J., Xiong, Z., Ding, Z.: Joint active and
Commun. 18(11), 5394–5409 (2019) passive beamforming design for the irs-assisted mimome-ofdm secure
115. Björnson, E., Sanguinetti, L.: Demystifying the power scaling law of communications. IEEE Trans. Veh. Technol. 70(10), 10 369–10 381
intelligent reflecting surfaces and metasurfaces. In: 2019 IEEE 8th (2021)
International Workshop on Computational Advances in Multi-Sensor 135. Ghanem, W.R., Jamali, V., Schober, R.: Joint beamforming and phase shift
Adaptive Processing (CAMSAP), pp. 549–553. IEEE, Piscataway (2019) optimization for multicell irs-aided ofdma-urllc systems. In: 2021 IEEE
116. Yu, X., Xu, D., Schober, R.: Miso wireless communication systems via Wireless Communications and Networking Conference (WCNC), pp. 1–
intelligent reflecting surfaces. In: 2019 IEEE/CIC International Con- 7. IEEE, Piscataway (2021)
ference on Communications in China (ICCC), pp. 735–740. IEEE, 136. Wei, Z., Cai, Y., Sun, Z., Ng, D.W.K., Yuan, J., Zhou, M., Sun, L.: Sum-rate
Piscataway (2019) maximization for irs-assisted uav ofdma communication systems. IEEE
117. Wu, Q., Zhang, R.: Beamforming optimization for wireless network aided Trans. Wireless Commun. 20(4), 2530–2550 (2020)
by intelligent reflecting surface with discrete phase shifts. IEEE Trans. 137. Pan, Y., Wang, K., Pan, C., Zhu, H., Wang, J.: Uav-assisted and intelligent
Commun. 68(3), 1838–1851 (2019) reflecting surfaces-supported terahertz communications. IEEE Wireless
118. Zhang, S., Zhang, R.: Capacity characterization for intelligent reflecting Commun. Lett. 10(6), 1256–1260 (2021)
surface aided mimo communication. IEEE J. Sel. Areas Commun. 38(8), 138. Yu, X., Xu, D., Schober, R.: Optimal beamforming for miso communica-
1823–1838 (2020) tions via intelligent reflecting surfaces. In: 2020 IEEE 21st International
119. Ying, K., Gao, Z., Lyu, S., Wu, Y., Wang, H., Alouini, M.-S.: Gmd- Workshop on Signal Processing Advances in Wireless Communications
based hybrid beamforming for large reconfigurable intelligent surface (SPAWC), pp. 1–5. IEEE, Piscataway (2020)
assisted millimeter-wave massive mimo. IEEE Access 8, 19 530–19 539 139. Björnson, E., Wymeersch, H., Matthiesen, B., Popovski, P., Sanguinetti,
(2020) L., de Carvalho, E.: Reconfigurable intelligent surfaces: A signal pro-
120. Zhang, Y., Zhong, C., Zhang, Z., Lu, W.: Sum rate optimization for two cessing perspective with wireless applications. IEEE Signal Process Mag.
way communications with intelligent reflecting surface. IEEE Commun. 39(2), 135–158 (2022)
Lett. 24(5), 1090–1094 (2020) 140. Wu, Q., Zhang, S., Zheng, B., You, C., Zhang, R.: Intelligent reflect-
121. Yang, Y., Zheng, B., Zhang, S., Zhang, R.: Intelligent reflecting sur- ing surface-aided wireless communications: A tutorial. IEEE Trans.
face meets ofdm: Protocol design and rate maximization. IEEE Trans. Commun. 69(5), 3313–3351 (2021)
Commun. 68(7), 4522–4535 (2020) 141. Björnson, E., Özdogan, Ö., Larsson, E.G.: Reconfigurable intelligent sur-
122. Zheng, B., Zhang, R.: Intelligent reflecting surface-enhanced ofdm: faces: Three myths and two critical questions. IEEE Commun. Mag.
Channel estimation and reflection optimization. IEEE Wireless Com- 58(12), 90–96 (2020)
mun. Lett. 9(4), 518–522 (2019) 142. Mishra, D., Johansson, H.: Channel estimation and low-complexity beam-
123. Björnson, E.: Configuring an intelligent reflecting surface for wireless forming design for passive intelligent surface assisted miso wireless
communications. arXiv preprint arXiv:2106.03497 (2021) energy transfer. In: ICASSP 2019-2019 IEEE International Conference
124. Yang, Z., Feng, L., Zhou, F., Qiu, X., Li, W.: Analytical performance on Acoustics, Speech and Signal Processing (ICASSP), pp. 4659–4663.
analysis of intelligent reflecting surface aided ambient backscatter com- IEEE, Piscataway (2019)
munication network. IEEE Wireless Commun. Lett. 10(12), 2732–2736 143. Zheng, B., Zhang, R.: Intelligent reflecting surface-enhanced ofdm:
(2021) Channel estimation and reflection optimization. IEEE Wireless Com-
125. He, M., Xu, W., Shen, H., Xie, G., Zhao, C., Di Renzo, M.: Cooperative mun. Lett. 9(4), 518–522 (2019)
multi-ris communications for wideband mmwave miso-ofdm systems. 144. He, Z.-Q., Yuan, X.: Cascaded channel estimation for large intelligent
IEEE Wireless Commun. Lett. 10(11), 2360–2364 (2021) metasurface assisted massive mimo. IEEE Wireless Commun. Lett. 9(2),
126. Du, H., Zhang, J., Cheng, J., Ai, B.: Millimeter wave communica- 210–214 (2019)
tions with reconfigurable intelligent surfaces: Performance analysis and 145. Elbir, A.M., Papazafeiropoulos, A., Kourtessis, P., Chatzinotas, S.:
optimization. IEEE Trans. Commun. 69(4), 2752–2768 (2021) Deep channel learning for large intelligent surfaces aided mm-wave
HASSOUNA ET AL. 535

massive mimo systems. IEEE Wireless Commun. Lett. 9(9), 1447–1451 165. Ning, B., Chen, Z., Chen, W., Du, Y., Fang, J.: Terahertz multi-user mas-
(2020) sive mimo with intelligent reflecting surface: Beam training and hybrid
146. Wang, Z., Liu, L., Cui, S.: Channel estimation for intelligent reflecting beamforming. IEEE Trans. Veh. Technol. 70(2), 1376–1393 (2021)
surface assisted multiuser communications: Framework, algorithms, and 166. Khan, S., Khan, K.S., Haider, N., Shin, S.Y.: Deep-learning-aided
analysis. IEEE Trans. Wireless Commun. 19(10), 6607–6620 (2020) detection for reconfigurable intelligent surfaces. arXiv preprint
147. Zheng, B., You, C., Zhang, R.: Intelligent reflecting surface assisted arXiv:1910.09136 (2019)
multi-user ofdma: Channel estimation and training design. IEEE Trans. 167. Elbir, A.M., Papazafeiropoulos, A., Kourtessis, P., Chatzinotas, S.: Deep
Wireless Commun. 19(12), 8315–8329 (2020) channel learning for large intelligent surfaces aided mm-wave massive
148. Yang, Y., Zheng, B., Zhang, S., Zhang, R.: Intelligent reflecting sur- mimo systems. IEEE Wireless Commun. Lett. 9(9), 1447–1451 (2020)
face meets ofdm: Protocol design and rate maximization. IEEE Trans. 168. Taha, A., Zhang, Y., Mismar, F.B., Alkhateeb, A.: Deep reinforce-
Commun. 68(7), 4522–4535 (2020) ment learning for intelligent reflecting surfaces: Towards standalone
149. Nadeem, Q.-U.-A., Kammoun, A., Chaaban, A., Debbah, M., Alouini, operation. In: 2020 IEEE 21st International Workshop on Signal Pro-
M.-S.: Intelligent reflecting surface assisted wireless communication: cessing Advances in Wireless Communications (SPAWC), pp. 1–5. IEEE
Modeling and channel estimation. arXiv preprint arXiv:1906.02360 (2020)
(2019) 169. Taha, A., Alrabeiah, M., Alkhateeb, A.: Deep learning for large intelligent
150. Mishra, D., Johansson, H.: Channel estimation and low-complexity beam- surfaces in millimeter wave and massive mimo systems. In: 2019 IEEE
forming design for passive intelligent surface assisted miso wireless Global Communications Conference (GLOBECOM), pp. 1–6. IEEE,
energy transfer. In: ICASSP 2019-2019 IEEE International Conference Piscataway (2019)
on Acoustics, Speech and Signal Processing (ICASSP), pp. 4659–4663. 170. Taha, A., Alrabeiah, M., Alkhateeb, A.: Enabling large intelligent surfaces
IEEE, Piscataway (2019) with compressive sensing and deep learning. IEEE Access 9, 44 304–
151. Zheng, B., You, C., Zhang, R.: Intelligent reflecting surface assisted 44 321 (2021)
multi-user ofdma: Channel estimation and training design. IEEE Trans. 171. Alexandropoulos, G.C., Vlachos, E.: A hardware architecture for
Wireless Commun. 19(12), 8315–8329 (2020) reconfigurable intelligent surfaces with minimal active elements for
152. Wan, Z., Gao, Z., Alouini, M.-S.: Broadband channel estimation for intel- explicit channel estimation. In: ICASSP 2020-2020 IEEE International
ligent reflecting surface aided mmwave massive mimo systems. In: ICC Conference on Acoustics, Speech and Signal Processing (ICASSP),
2020-2020 IEEE International Conference on Communications (ICC), pp. 9175–9179. IEEE, Piscataway (2020)
pp. 1–6. IEEE, Piscataway (2020) 172. You, C., Zheng, B., Zhang, R.: How to deploy intelligent reflecting sur-
153. Wang, P., Fang, J., Duan, H., Li, H.: Compressed channel estimation faces in wireless network: Bs-side, user-side, or both sides? arXiv preprint
for intelligent reflecting surface-assisted millimeter wave systems. IEEE arXiv:2012.03403 (2020)
Signal Process Lett. 27, 905–909 (2020) 173. Han, Y., Zhang, S., Duan, L., Zhang, R.: Cooperative double-irs aided
154. He, Z.-Q., Yuan, X.: Cascaded channel estimation for large intelligent communication: Beamforming design and power scaling. IEEE Wireless
metasurface assisted massive mimo. IEEE Wireless Commun. Lett. 9(2), Commun. Lett. 9(8), 1206–1210 (2020)
210–214 (2019) 174. Zheng, B., You, C., Zhang, R.: Double-irs assisted multi-user mimo:
155. de Araújo, G.T., de Almeida, A.L.: Parafac-based channel estimation Cooperative passive beamforming design IEEE Trans. Wireless Com-
for intelligent reflective surface assisted mimo system. In: 2020 IEEE mun. 20(7), 4513–4526 (2021)
11th Sensor Array and Multichannel Signal Processing Workshop (SAM), 175. You, C., Zheng, B., Zhang, R.: Wireless communication via double irs:
pp. 1–5. IEEE, Piscataway (2020) Channel estimation and passive beamforming designs. IEEE Wireless
156. Mirza, J., Ali, B.: Channel estimation method and phase shift Commun. Lett. 10(2), 431–435 (2020)
design for reconfigurable intelligent surface assisted mimo net- 176. Li, Z., Hua, M., Wang, Q., Song, Q.: Weighted sum-rate maximization for
works. IEEE Trans. Cognit. Commun. Networking 7(2), 441–451 multi-irs aided cooperative transmission. IEEE Wireless Commun. Lett.
(2021) 9(10), 1620–1624 (2020)
157. He, J., Leinonen, M., Wymeersch, H., Juntti, M.: Channel estimation 177. Zhang, S., Zhang, R.: Intelligent reflecting surface aided multi-user
for ris-aided mmwave mimo systems. In: GLOBECOM 2020-2020 communication: Capacity region and deployment strategy. IEEE Trans.
IEEE Global Communications Conference, pp. 1–6. IEEE, Piscataway Commun. (2021)
(2020) 178. Wu, J., Shim, B.: Power minimization of intelligent reflecting surface-
158. Cui, Y., Yin, H.: An efficient csi acquisition method for intelligent reflect- aided uplink iot networks. In: 2021 IEEE Wireless Communications and
ing surface-assisted mmwave networks. arXiv preprint arXiv:1912.12076 Networking Conference (WCNC), pp. 1–6. IEEE, Piscataway (2021)
(2019) 179. Lyu, J., Zhang, R.: Hybrid active/passive wireless network aided by intelli-
159. Wang, P., Fang, J., Duan, H., Li, H.: Compressed channel estimation gent reflecting surface: System modeling and performance analysis. IEEE
for intelligent reflecting surface-assisted millimeter wave systems. IEEE Trans. Wireless Commun. 20(11), 7196–7212 (2021)
Signal Process Lett. 27, 905–909 (2020) 180. Kishk, M.A., M.-S. Alouini,: Exploiting randomly located blockages
160. Chen, J., Liang, Y.-C., Cheng, H.V., Yu, W.: Channel estimation for for large-scale deployment of intelligent surfaces. IEEE J. Sel. Areas
reconfigurable intelligent surface aided multi-user mimo systems. arXiv Commun. 39(4), 1043–1056 (2020)
preprint arXiv:1912.03619 (2019) 181. Liu, X., Liu, Y., Chen, Y., Poor, H.V.: Ris enhanced massive non-
161. Hu, C., Dai, L., Han, S., Wang, X.: Two-timescale channel estimation for orthogonal multiple access networks: Deployment and passive beam-
reconfigurable intelligent surface aided wireless communications. IEEE forming design. IEEE J. Sel. Areas Commun. 39(4), 1057–1071 (2020)
Trans. Commun. 69(11), 7736–7747 (2021) 182. Jia, C., Gao, H., Chen, N., He, Y.: Machine learning empowered beam
162. Liu, H., Yuan, X., Zhang, Y.-J.A.: Matrix-calibration-based cascaded chan- management for intelligent reflecting surface assisted mmwave networks.
nel estimation for reconfigurable intelligent surface assisted multiuser China Commun. 17(10), 100–114 (2020)
mimo. IEEE J. Sel. Areas Commun. 38(11), 2621–2636 (2020) 183. Björnson, E., Sanguinetti, L.: Power scaling laws and near-field behav-
163. Wei, L., Huang, C., Alexandropoulos, G.C., Yuen, C.: Parallel factor iors of massive mimo and intelligent reflecting surfaces. IEEE Open J.
decomposition channel estimation in ris-assisted multi-user miso com- Commun. Soc. 1, 1306–1324 (2020)
munication. In: 2020 IEEE 11th Sensor Array and Multichannel Signal 184. Björnson, E., Özdogan, Ö., Larsson, E.G.: Reconfigurable intelligent sur-
Processing Workshop (SAM), pp. 1–5. IEEE, Piscataway (2020) faces: Three myths and two critical questions. IEEE Commun. Mag.
164. Ning, B., Chen, Z., Chen, W., Du, Y.: Channel estimation and transmission 58(12), 90–96 (2020)
for intelligent reflecting surface assisted thz communications. In: ICC 185. Elzanaty, A., Guerra, A., Guidi, F., Alouini, M.-S.: Reconfigurable intel-
2020-2020 IEEE International Conference on Communications (ICC), ligent surfaces for localization: Position and orientation error bounds.
pp. 1–7. IEEE, Piscataway (2020) IEEE Trans. Signal Process. 69, 5386–5402 (2021)
536 HASSOUNA ET AL.

186. Loyka, S.: Electromagnetic interference in wireless communica- 205. Cui, T.J., Qi, M.Q., Wan, X., Zhao, J., Cheng, Q.: Coding metamaterials,
tions: behavioral-level simulation approach. In: IEEE 60th Vehicular digital metamaterials and programmable metamaterials. Light Sci. Appl.
Technology Conference, VTC2004-Fall, vol. 6, pp. 3945–3949. IEEE, 3(10), e218–e218 (2014)
Piscataway (2004) 206. Liu, S., Cui, T.J., Zhang, L., Xu, Q., Wang, Q., Wan, X., Gu, J.Q., Tang,
187. de Jesus Torres, A., Sanguinetti, L., Björnson, E.: Electromagnetic inter- W.X., Qing Qi, M., Han, J.G., et al.: Convolution operations on coding
ference in ris-aided communications. IEEE Wireless Commun. Lett. metasurface to reach flexible and continuous controls of terahertz beams.
11(4), 668–672 (2021) Adv. Sci. 3(10), 1600156 (2016)
188. Chu, Z., Zhu, Z., Zhou, F., Zhang, M., Al-Dhahir, N.: Intelligent reflect- 207. Li, L., Jun Cui, T., Ji, W., Liu, S., Ding, J., Wan, X., Bo, Li, Y., Jiang, M., Qiu,
ing surface assisted wireless powered sensor networks for internet of C.-W., Zhang, S.: Electromagnetic reprogrammable coding-metasurface
things. IEEE Trans. Commun. 69(7), 4877–4889 (2021) holograms. Nat. Commun. 8(1), 1–7 (2017)
189. Guan, X., Wu, Q., Zhang, R.: Anchor-assisted intelligent reflecting sur- 208. Ma, Q., Shi, C.B., Bai, G.D., Chen, T.Y., Noor, A., Cui, T.J.: Beam-editing
face channel estimation for multiuser communications. In: GLOBECOM coding metasurfaces based on polarization bit and orbital-angular-
2020-2020 IEEE Global Communications Conference, pp. 1–6. IEEE, momentum-mode bit. Adv. Opt. Mater. 5(23), 1700548 (2017)
Piscataway (2020) 209. Liu, S., Cui, T.J.: Flexible controls of terahertz waves using coding and
190. Xia, S., Shi, Y.: Intelligent reflecting surface for massive device con- programmable metasurfaces. IEEE J. Sel. Top. Quantum Electron. 23(4),
nectivity: Joint activity detection and channel estimation. In: ICASSP 1–12 (2016)
2020-2020 IEEE International Conference on Acoustics, Speech 210. Zhang, L., Liu, S., Li, L., Cui, T.J.: Spin-controlled multiple pencil
and Signal Processing (ICASSP), pp. 5175–5179. IEEE, Piscataway beams and vortex beams with different polarizations generated by
(2020) pancharatnam-berry coding metasurfaces. ACS Appl. Mater. Interfaces
191. Jensen, T.L., De Carvalho, E.: An optimal channel estimation scheme 9(41), 36447–36455 (2017)
for intelligent reflecting surfaces based on a minimum variance unbi- 211. Shuang, Y., Zhao, H., Ji, W., Cui, T.J., Li, L.: Programmable high-order
ased estimator. In: ICASSP 2020-2020 IEEE International Conference oam-carrying beams for direct-modulation wireless communications.
on Acoustics, Speech and Signal Processing (ICASSP), pp. 5000–5004. IEEE J. Emerging Sel. Top. Circuits Syst. 10(1), 29–37 (2020)
IEEE, Piscataway (2020) 212. Zhang, X.G., Jiang, W.X., Jiang, H.L., Wang, Q., Tian, H.W., Bai, L., Luo,
192. Souto, V.D.P., Souza, R.D., Uchôa-Filho, B.F., Li, A., Li, Y.: Beamforming Z.J., Sun, S., Luo, Y., Qiu, C.-W., et al.: An optically driven digital meta-
optimization for intelligent reflecting surfaces without csi. IEEE Wireless surface for programming electromagnetic functions. Nat. Electron. 3(3),
Commun. Lett. 9(9), 1476–1480 (2020) 165–171 (2020)
193. Abrardo, A., Dardari, D., Di Renzo, M.: Intelligent reflecting surfaces: 213. Zhao, J., Yang, X., Dai, J.Y., Cheng, Q., Li, X., Qi, N.H., Ke, J.C., Bai,
Sum-rate optimization based on statistical position information. IEEE G.D., Liu, S., Jin, S., et al.: Programmable time-domain digital-coding
Trans. Commun. 69(10), 7121–7136 (2021) metasurface for non-linear harmonic manipulation and new wireless
194. Yigit, Z., Basar, E., Altunbas, I.: Low complexity adaptation for recon- communication systems. Natl. Sci. Rev. 6(2), 231–238 (2019)
figurable intelligent surface-based mimo systems. IEEE Commun. Lett. 214. Dai, J.Y., Zhao, J., Cheng, Q., Cui, T.J.: Independent control of harmonic
24(12), 2946–2950 (2020) amplitudes and phases via a time-domain digital coding metasurface.
195. Basar, E.: Reconfigurable intelligent surfaces for doppler effect Light Sci. Appl. 7(1), 1–10 (2018)
and multipath fading mitigation. arXiv preprint arXiv:1912.04080 215. Zhang, L., Chen, X.Q., Liu, S., Zhang, Q., Zhao, J., Dai, J.Y., Bai,
(2019) G.D., Wan, X., Cheng, Q., Castaldi, G., et al.: Space-time-coding digital
196. Matthiesen, B., Björnson, E., De Carvalho, E., Popovski, P.: Intelligent metasurfaces. Nat. Commun. 9(1), 1–11 (2018)
reflecting surface operation under predictable receiver mobility: A con- 216. Zhang, L., Chen, X.Q., Shao, R.W., Dai, J.Y., Cheng, Q., Castaldi, G.,
tinuous time propagation model. IEEE Wireless Commun. Lett. 10(2), Galdi, V., Cui, T.J.: Breaking reciprocity with space-time-coding digital
216–220 (2020) metasurfaces. Adv. Mater. 31(41), 1904069 (2019)
197. Tan, X., Sun, Z., Koutsonikolas, D., Jornet, J.M.: Enabling indoor 217. Cui, T.-J., Liu, S., Li, L.-L.: Information entropy of coding metasurface.
mobile millimeter-wave networks based on smart reflect-arrays. In: IEEE Light Sci. Appl. 5(11), e16 172–e16 172 (2016)
INFOCOM 2018-IEEE Conference on Computer Communications, 218. Cui, T.J., Liu, S., Zhang, L.: Information metamaterials and metasurfaces.
pp. 270–278. IEEE, Piscataway (2018) J. Mater. Chem. C 5(15), 3644–3668 (2017)
198. Jamali, V., Tulino, A.M., Fischer, G., Müller, R.R., Schober, R.: Intel- 219. Cui, T.J., Li, L., Liu, S., Ma, Q., Zhang, L., Wan, X., Jiang, W.X.,
ligent surface-aided transmitter architectures for millimeter-wave ultra Cheng, Q.: Information metamaterial systems. Iscience 23(8), 101403
massive mimo systems. IEEE Open J. Commun. Soc. 2, 144–167 (2020)
(2020) 220. Dai, J.Y., Tang, W.K., Zhao, J., Li, X., Cheng, Q., Ke, J.C., Chen, M.Z., Jin,
199. Wang, P., Fang, J., Yuan, X., Chen, Z., Li, H.: Intelligent reflecting S., Cui, T.J.: Wireless communications through a simplified architecture
surface-assisted millimeter wave communications: Joint active and pas- based on time-domain digital coding metasurface. Adv. Mater. Technol.
sive precoding design. IEEE Trans. Veh. Technol. 69(12), 14 960–14 973 4(7), 1900044 (2019)
(2020) 221. Wan, X., Zhang, Q., Chen, T.Y., Zhang, L., Xu, W., Huang, H., Xiao,
200. Mao, Y., You, C., Zhang, J., Huang, K., Letaief, K.B.: A survey on mobile C.K., Xiao, Q., Jun Cui, T.: Multichannel direct transmissions of near-field
edge computing: The communication perspective. IEEE Commun. Surv. information. Light Sci. Appl. 8(1), 1–8 (2019)
Tutorials 19(4), 2322–2358 (2017) 222. Cui, T.J., Liu, S., Bai, G.D., Ma, Q.: Direct transmission of digital mes-
201. Cao, Y., Lv, T.: Intelligent reflecting surface enhanced resilient sage via programmable coding metasurface. Research 2019, 2584509
design for mec offloading over millimeter wave links. arXiv preprint (2019)
arXiv:1912.06361 (2019) 223. Dai, J.Y., Tang, W., Yang, L.X., Li, X., Chen, M.Z., Ke, J.C., Cheng, Q.,
202. Liu, Y., Zhao, J., Xiong, Z., Niyato, D., Yuen, C., Pan, C., Huang, B.: Jin, S., Cui, T.J.: Realization of multi-modulation schemes for wireless
Intelligent reflecting surface meets mobile edge computing: Enhanc- communication by time-domain digital coding metasurface. IEEE Trans.
ing wireless communications for computation offloading. arXiv preprint Antennas Propag. 68(3), 1618–1627 (2019)
arXiv:2001.07449 (2020) 224. Zhao, H., Shuang, Y., Wei, M., Cui, T.J., Hougne, P.D., Li, L.: Metasurface-
203. Bai, T., Pan, C., Deng, Y., Elkashlan, M., Nallanathan, A., Hanzo, L.: assisted massive backscatter wireless communication with commodity wi-
Latency minimization for intelligent reflecting surface aided mobile edge fi signals. Nat. Commun. 11(1), 1–10 (2020)
computing. IEEE J. Sel. Areas Commun. 38(11), 2666–2682 (2020) 225. Zhang, L., Chen, M.Z., Tang, W., Dai, J.Y., Miao, L., Zhou, X.Y., Jin,
204. Hua, S., Zhou, Y., Yang, K., Shi, Y., Wang, K.: Reconfigurable intelli- S., Cheng, Q., Cui, T.J.: A wireless communication scheme based on
gent surface for green edge inference. IEEE Trans. Green Commun. space-and frequency-division multiplexing using digital metasurfaces.
Networking 5(2), 964–979 (2021) Nat. Electron. 4(3), 218–227 (2021)
HASSOUNA ET AL. 537

226. Chen, M.Z., Tang, W., Dai, J.Y., Ke, J.C., Zhang, L., Zhang, C., Yang, J.,
Li, L., Cheng, Q., Jin, S., et al.: Accurate and broadband manipulations of
harmonic amplitudes and phases to reach 256 qam millimeter-wave wire- How to cite this article: Hassouna, S., Jamshed, M.A.,
less communications by time-domain digital coding metasurface. Natl. Rains, J., Kazim, J.R., Rehman, M.U., Abualhayja, M.,
Sci. Rev. 9(1), nwab134 (2022) Mohjazi, L., Cui, T.J., Imran, M.A., Abbasi, Q.H.: A
227. Wu, H., Bai, G.D., Liu, S., Li, L., Wan, X., Cheng, Q., Cui, T.J.: Information survey on reconfigurable intelligent surfaces: Wireless
theory of metasurfaces. Natl. Sci. Rev. 7(3), 561–571 (2020)
communication perspective. IET Commun. 17,
228. Wu, H., Gao, X.X., Zhang, L., Bai, G.D., Cheng, Q., Li, L., Cui, T.J.: Har-
monic information transitions of spatiotemporal metasurfaces. Light Sci. 497–537 (2023). https://doi.org/10.1049/cmu2.12571
Appl. 9(1), 1–13 (2020)