wu2021.pdf

IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 69, NO. 5, MAY 2021 3313
Intelligent Reflecting Surface-Aided Wireless
Communications: A Tutorial
Qingqing Wu , Member, IEEE, Shuowen Zhang , Member, IEEE, Beixiong Zheng , Member, IEEE,
Changsheng You , Member, IEEE, and Rui Zhang , Fellow, IEEE
(Invited Paper)
Abstract—Intelligent reflecting surface (IRS) is an enabling
technology to engineer the radio signal propagation in wireless
networks. By smartly tuning the signal reflection via a large
number of low-cost passive reflecting elements, IRS is capable of
dynamically altering wireless channels to enhance the communi-
cation performance. It is thus expected that the new IRS-aided
hybrid wireless network comprising both active and passive com-
ponents will be highly promising to achieve a sustainable capacity
growth cost-effectively in the future. Despite its great potential,
IRS faces new challenges to be efficiently integrated into wireless
networks, such as reflection optimization, channel estimation,
and deployment from communication design perspectives. In
this paper, we provide a tutorial overview of IRS-aided wireless
communications to address the above issues, and elaborate
its reflection and channel models, hardware architecture and
practical constraints, as well as various appealing applications in
wireless networks. Moreover, we highlight important directions
worthy of further investigation in future work.
Index Terms—Intelligent reflecting surface (IRS), smart and
reconfigurable environment, IRS-aided wireless communication,
IRS channel model, IRS hardware architecture and practical
constraints, IRS reflection optimization, IRS channel estimation,
IRS deployment, IRS applications.
I. INTRODUCTION
A. Motivation
ALTHOUGH the fifth-generation (5G) wireless network
is still under deployment worldwide, both academia
and industry have been enthusiastically looking into future
beyond 5G (B5G) such as the sixth-generation (6G) wireless
Manuscript received July 5, 2020; revised November 27, 2020; accepted
December 28, 2020. Date of publication January 18, 2021; date of current
version May 18, 2021. The work of Qingqing Wu is supported in part by
SRG2020-00024-IOTSC and FDCT 0108/2020/A. This work is supported
in part by the National University of Singapore under Research Grant
R-261-518-005-720. The associate editor coordinating the review of this
article and approving it for publication was D. W. K. Ng. (Corresponding
author: Rui Zhang.)
Qingqing Wu was with the Department of Electrical and Computer Engi-
neering, National University of Singapore, Singapore 117583. He is now with
the State Key Laboratory of Internet of Things for Smart City and Department
of Electrical and Computer Engineering, University of Macau, Macau 999078,
China (e-mail: qingqingwu@um.edu.mo).
Shuowen Zhang was with the Department of Electrical and Computer
Engineering, National University of Singapore, Singapore 117583. She
is now with the Department of Electronic and Information Engineering,
The Hong Kong Polytechnic University, Hong Kong SAR, China (e-mail:
shuowen.zhang@polyu.edu.hk).
Beixiong Zheng, Changsheng You, and Rui Zhang are with the Depart-
ment of Electrical and Computer Engineering, National University of Sin-
gapore, Singapore 117583 (e-mail: elezbe@nus.edu.sg; eleyouc@nus.edu.sg;
elezhang@nus.edu.sg).
Color versions of one or more figures in this article are available at
https://doi.org/10.1109/TCOMM.2021.3051897.
Digital Object Identifier 10.1109/TCOMM.2021.3051897
network that targets at meeting more stringent requirements
than 5G, such as ultra high data rate and energy efficiency,
global coverage and connectivity, as well as extremely high
reliability and low latency [1]. These requirements, however,
may not be fully achieved with the existing technology
trends for accommodating 5G services (e.g., enhanced mobile
broadband (eMBB), ultra-reliable and low latency communi-
cation (URLLC), and massive machine-type communication
(mMTC)), which mainly include [2]–[5]
• deploying increasingly more active nodes such as base
stations (BSs), access points (APs), relays, and distributed
antennas/remote radio heads (RRHs) to shorten the com-
munication distance for achieving enhanced network cov-
erage and capacity, which, however, incurs higher energy
consumption and deployment/backhaul/maintenance cost,
as well as the more severe and complicated network
interference issues;
• packing substantially more antennas at the
BSs/APs/relays to harness the enormous massive
multiple-input-multiple-output (M-MIMO) gains, which
requires increased hardware and energy cost as well as
signal processing complexity;
• migrating to higher frequency bands such as millimeter
wave (mmWave) and even terahertz (THz) frequencies
to utilize their large and available bandwidth, which
inevitably results in deploying even more active nodes
and mounting them even more antennas (i.e., super
MIMO) so as to compensate for their higher propagation
loss over distance.
In view of the above issues and limitations, it is imperative
to develop disruptively new and innovative technologies to
achieve a sustainable capacity growth of future wireless net-
works with low and affordable cost, complexity, and energy
consumption.
On the other hand, the fundamental challenge for achieving
ultra-reliable wireless communications arises from the time-
varying wireless channels due to user mobility. Traditional
approaches for tackling this challenge either compensate for
the channel fading by exploiting various modulation, coding
and diversity techniques, or adapt to it via adaptive power/rate
control and beamforming techniques [6], [7]. However, they
not only need additional overhead but also have limited control
over the largely random wireless channels, thus leaving the
ultimate barrier to achieving high-capacity and ultra-reliable
wireless communications unconquered.
0090-6778 © 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: City, University of London. Downloaded on May 17,2021 at 20:50:28 UTC from IEEE Xplore. Restrictions apply.

3314 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 69, NO. 5, MAY 2021
Fig. 1. Main functions of IRS for wireless channel reconfiguration.
B. What Is IRS?
Motivated by the above, intelligent reflecting surface (IRS)
has recently emerged as a promising new paradigm to achieve
smart and reconfigurable wireless channels/radio propagation
environment for B5G/6G wireless communication systems
[8]–[10]. Generally speaking, IRS is a planar surface com-
prising a large number of passive reflecting elements, each
of which is able to induce a controllable amplitude and/or
phase change to the incident signal independently. By densely
deploying IRSs in wireless network and smartly coordinat-
ing their reflections, the signal propagation/wireless channels
between transmitters and receivers can be flexibly recon-
figured to achieve desired realizations and/or distributions,
which thus provides a new means to fundamentally tackle
the wireless channel fading impairment and interference issue,
and potentially achieves a quantum leap improvement for
wireless communication capacity and reliability. As shown
in Fig. 1, IRS is able to achieve several appealing functions
in wireless channel reconfiguration, such as creating virtual
line-of-sight (LoS) link to bypass obstacles between trans-
ceivers via smart reflection, adding extra signal paths toward
desired direction to improve the channel rank condition,
refining the channel statistics/distribution by e.g. transforming
Rayleigh/fast fading to Rician/slow fading for ultra-high reli-
ability, suppressing/nulling co-channel/inter-cell interference
and so on.
Not only conceptually appealing, IRS also possesses various
practical advantages for implementation. First, its reflecting
elements (e.g., low-cost printed dipoles) only passively reflect
the impinging signals without requiring any transmit radio-
frequency (RF) chains, thus can be implemented/operated with
orders-of-magnitude lower hardware/energy cost as compared
to traditional active antenna arrays or the recently proposed
active surfaces [11]. Besides, IRS operates in full-duplex
(FD) mode and is free of any antenna noise amplification
as well as self-interference, which thus offers competitive
advantages over traditional active relays, e.g., half-duplex
(HD) relay that suffers from low spectral efficiency as well
as FD relay that needs sophisticated techniques for self-
interference cancellation. Furthermore, since IRS is generally
of low profile, light weight, and conformal geometry, it can
be easily mounted on/removed from environment objects for
deployment/replacement. Finally, IRS serves as an auxiliary
device in wireless networks and can be integrated into them
transparently, thus providing great flexibility and compatibility
with existing wireless systems (e.g., cellular or WiFi).
Due to the above promising advantages, IRS is suitable to
be massively deployed in wireless networks to significantly
enhance its spectral and energy efficiency cost-effectively.
As such, it is envisioned that IRS will lead to fundamental
paradigm shifts of wireless system/network designs, namely,
from the existing M-MIMO system without IRS to the new

WU et al.: IRS-AIDED WIRELESS COMMUNICATIONS: A TUTORIAL 3315
Fig. 2. Potential paradigm shifts of wireless system/network designs with IRS.
IRS-aided small/moderate MIMO system, as well as from
the existing heterogenous wireless network to the new IRS-
aided hybrid network in the future, as shown in Figs. 2 (a)
and (b), respectively. On one hand, different from M-MIMO
that leverages tens and even hundreds of active antennas to
generate sharp beams directly, an IRS-aided MIMO system
allows the BS to be equipped with substantially less antennas
without compromising the users’ quality-of-service (QoS),
by exploiting the large aperture of IRS to create fine-grained
reflect beams via smart passive reflections [9]. As such,
the system hardware cost and energy consumption can be sig-
nificantly reduced, especially for wireless systems migrating
to higher frequency bands in the future. On the other hand,
although existing wireless networks rely on a heterogenous
multi-tier architecture consisting of macro and small BSs/APs,
relays, distributed antennas, etc., they are all active nodes that
generate new signals in the network, thus requiring sophisti-
cated coordination and interference management among them
in order to achieve the premise of enhanced network spatial
capacity with more active nodes deployed. However, this
approach inevitably aggravates the network operation overhead
and thus may not be able to sustain the wireless network
capacity growth cost-effectively in the future. In contrast,
integrating IRSs into wireless network will shift the existing
heterogeneous network with active components only to a
new hybrid architecture comprising both active and passive
components co-working in an intelligent way. Since IRSs are
of much lower cost as compared to their active counterparts,
they can be more densely deployed in wireless network at even
lower cost, yet without the need of sophisticated interference
management between IRSs thanks to their passive reflection
and resultant local coverage. By optimally setting the ratios
between active BSs and passive IRSs deployed in the hybrid
network given their total cost, a sustainable network capacity
scaling with cost can be achieved [12].
In Fig. 3, we show an envisioned future wireless network
aided by IRSs with a variety of promising applications. For
instance, IRS is particularly useful for the coverage extension
in mmWave and THz communications that are highly vulner-
able to blockage. Besides, deploying IRSs at the cell edge not
only helps improve the desired signal power at cell-edge users
but also facilitates the suppression of co-channel interference
from neighboring cells to them. Moreover, to improve the
efficiency of simultaneous wireless information and power
transfer (SWIPT) from the AP to wireless devices in e.g.,
smart office/home, the large aperture of IRS can be leveraged
to compensate for the significant power loss over long distance
via reflect beamforming to its nearby devices. In indoor

Fig. 3. Illustration of IRS applications in future wireless network.
environment, IRS can also be attached to the ceilings, walls,
furniture, and even behind the paintings/decorations, to help
achieve enhanced coverage and high-capacity hot-spot, which
is particularly appealing for eMBB and mMTC applications
in smart factories, stadiums, shopping centers, airports, etc.
In particular, for mMTC scenarios usually with only a small
fraction of devices that are active for communication at each
time instant, IRS can be leveraged to effectively improve the
device activity detection accuracy and efficiency by exploiting
its extra controllable paths, especially when massive devices
are distributed in different propagation conditions [13]. While
in outdoor environment, IRS can be coated on the building
facade, lamppost, advertising board, and even the surface of
high-speed moving vehicles, to support various applications,
e.g., URLLC for remote control and smart transportation
by effectively compensating the Doppler and delay spread
effects [14]–[16]. For instance, the communication reliability
can be greatly improved by exploiting IRS to turn typically
random wireless channels into more deterministic ones, thus
helping reduce packet retransmissions and minimize the delay
which is critical for URLLC applications. Therefore, IRS
is a disruptive technology for making our current “dumb”
environment intelligent, which can potentially benefit a wide
range of vertical industries in 5G/6G such as transportation,
manufacturing, smart city, etc. Recently, IRS has been recog-
nized as one promising technology for the future 6G ecosystem
[1], [17], [18]. In addition, there has been an upsurge of
interest in industry on implementing and commercializing
IRS-like technologies to create new value chains, and in
the meanwhile, several pilot projects have been launched
to advance the research in this new field, with their more
information given in Table I. It is worth noting that there
have been other terminologies similar to IRS named in the
literature, such as intelligent wall [19], smart reflectarray [20],
and reconfigurable metasurface/intelligent surface (RIS) [21],
[22], large intelligent surface/antennas (LISA) [23], RFocus
[24], among others, which, despite different terminologies, are
essentially based on the same principle of passive and tunable
reflecting/refracting surfaces.1
C. What’s New?
Although IRS can be regarded as a reconfigurable metasur-
face, it extends the traditional applications of metasurface via
controlling electromagnetic (EM) waves such as invisibility
cloaking, imaging, radar sensing, and hologram [33], to the
new frontier of wireless communication as an innovative
enabler for smart and reconfigurable propagation environment.
Moreover, compared with the traditional reflectarray [34]
where a passive mirror/lens with fixed/reconfigurable beam
patterns is placed in the near field of the wireless transceiver
for saving the active antennas/RF chains, IRS is flexibly
located in the network to help alter the wireless communi-
cation channel via smart reflection. As such, the design of
IRS-aided wireless systems/networks faces new and unique
challenges that arise from a communication standpoint, which
are elaborated as follows.
Firstly, the passive reflections of all reflecting elements at
each IRS need to be properly designed to achieve cooper-
ative signal focusing and/or interference cancellation in its
local proximity. Meanwhile, to serve all the users in the
network regardless of whether there is any associated IRS
nearby each user, the IRS passive reflections also need to be
jointly designed with the BSs’/users’ transmissions so as to
optimize their end-to-end communications over the reconfig-
ured wireless channels by IRSs. Secondly, as IRS does not
1Also note that such surfaces can be practically manufactured as a mir-
ror or lens for signal reflection and refraction, respectively, depending on the
application scenarios where wireless transmitter and receiver are located on
the same side of the surface, or on its opposite sides.

TABLE I
LIST OF MAIN INDUSTRY PROGRESS, PROTOTYPES, AND PROJECTS RELATED TO IRS
possess RF chains in general, how to acquire the channel state
information (CSI) between IRS and its serving BSs/users that
is essential to the aforementioned IRS reflection optimization
becomes a practically difficult task, especially considering that
IRS typically has a large number of reflecting elements and
thus associated channel coefficients to be estimated. Thirdly,
the optimal deployment strategy of IRSs in wireless network to
maximize the network capacity is expected to be significantly
different from that for the conventional wireless networks with
active BSs/APs and relays only due to their different array
architectures (passive versus active) and operating mechanisms
(reflect versus transmit/receive), which thus needs to be re-
investigated thoroughly. In a nutshell, the efficient integration
of IRSs into wireless network brings both new opportunities as
well as challenges, which deserve a new and dedicated study.
D. Objective, Contribution, and Organization
The promising prospects of IRS for future wireless networks
have spurred extensive research recently. A handful of articles
have appeared in the literature providing overview or survey
of the research work on IRS and its variants from different
perspectives, such as IRS implementation, channel modelling,
applications, etc. [10], [21]–[23], [35]–[42], which are sum-
marized in Table II for ease of reference.
Compared with the above works, this paper is the first tuto-
rial on IRS-aided wireless communications, with an emphasis
on addressing its main technical challenges from a com-
munication standpoint. In addition to reviewing the state-of-
the-art results on IRS, this paper aims to provide in-depth
technical discussion to facilitate and inspire future research in
modelling, analysis, design, optimization, and implementation
of IRS-aided wireless networks. To this end, this tutorial paper
offers a new and systematic treatment on how to address three
key design issues in them, namely, IRS passive reflection
optimization, IRS channel estimation, and IRS deployment
from various communication perspectives.
The rest of this paper is organized as follows. Section II
introduces the fundamentals of IRS-aided wireless commu-
nication, including its signal and channel models, hard-
ware architecture as well as practical constraints. Section III
addresses the passive reflection optimization for IRS-aided
wireless communication under various system setups, namely,
from single-user to multi-user, single-antenna to multi-
antenna, narrow-band to broadband, as well as from single-
cell to multi-cell. In Section IV, we present promising methods
for IRS channel estimation under different IRS configurations
and communication setups. In Section V, we address the
optimal deployment of IRSs at both the link and network
levels. In Section VI, we discuss other relevant topics on IRS
for broadening its scope. Finally, we conclude this paper in
Section VII.
Notations: In this paper, scalars are denoted by italic letters,
vectors and matrices are denoted by bold-face lower-case and
upper-case letters, respectively. Cx×y
denotes the space of
x × y complex-valued matrices. For a complex-valued vector
x, x denotes its Euclidean norm and diag(x) denotes a
diagonal matrix with the elements in x on its main diagonal.
The distribution of a circularly symmetric complex Gaussian
(CSCG) random vector with mean vector x and covariance
matrix Σ is denoted by CN(x, Σ); and ∼ stands for “dis-
tributed as”. For a square matrix S, tr(S) and S−1
denote
its trace and inverse, respectively, while S 0 means that S
is positive semi-definite, where 0 is a zero matrix of proper
size. For a general matrix A, A∗
, AH
, rank(A), and [A]i,j
denote its conjugate, conjugate transpose, rank, and the (i, j)th
entry, respectively. j denotes the imaginary unit, i.e., j2
= −1.
E(·) denotes the statistical expectation. Re{·} denotes the real
part of a complex number. The operator mod(a, b) returns the
remainder after division of a by b, where a is the dividend
and b is the divisor; denotes the convolution operation; x
and x return the nearest integer less than or equal to, and
greater than or equal to the real number x, respectively.
II. IRS FUNDAMENTALS
In this section, we present the basics pertinent to
IRS-aided wireless communication, where the fundamen-
tal IRS signal and channel models are first introduced,
followed by its hardware architecture and practical con-
straints as well as their induced issues worthy of future
investigation.
Authorized licensed use limited to: City#44; University of London. Downloaded on May 17,2021 at 20:50:28 UTC from IEEE Xplore. Restrictions apply.

TABLE II
LIST OF REPRESENTATIVE SURVEY/OVERVIEW PAPERS RELATED TO IRS
A. IRS Signal and Channel Model
For the purpose of exposition, we consider the basic point-
to-point communication system where an IRS comprising N
passive reflecting elements on a planar surface is deployed to
assist in the communication from a transmitter to its intended
receiver. For ease of illustration, we assume a single antenna
at both the transmitter and receiver and the communication
system is of narrow band at a given carrier frequency, while
the more general multi-antenna and/or broadband systems will
be considered later in this paper. The carrier frequency and
the system bandwidth are denoted by fc and B in hertz (Hz),
respectively, with B fc.
Let x(t) denote the equivalent complex-valued baseband
transmit signal. Without loss of generality, we first focus on
the signal propagation from the transmitter to the receiver via
one particular reflecting element of the IRS, denoted by n,
with n ∈ {1, . . . , N}. Denote by α1,ne−jξ1,n
the equivalent
baseband complex channel coefficient from the transmitter to
IRS element n in which α1,n and ξ1,n represent the amplitude
attenuation and phase shift of the frequency-flat channel of the
narrow-band system, respectively. Then the passband signal
impinging on IRS element n is given by
yin,n(t) = Re

α1,ne−jξ1,n
x(t)ej2πfct

. (1)
Denote the amplitude attenuation and time delay induced by
IRS element n by βn ∈ [0, 1]2
and tn ∈ [0, 1/fc], respectively.
2This is because IRS reflecting elements are passive without signal ampli-
fication.
By ignoring the hardware imperfections such as circuit non-
linearity and phase noise, the reflected signal by IRS element
n is expressed as
yout,n(t) = βnyin,n(t − tn)
= Re

βnα1,ne−jξ1,n
x(t − tn)ej2πfc(t−tn)

≈ Re

βne−jθ
n α1,ne−jξ1,n
x(t)

ej2πfct

, (2)
where we have assumed x(t − tn) ≈ x(t) due to the fact
that tn ≤ 1/fc 1/B, and −γ
n −2πfctn ∈ [−2π, 0]
is the phase shift induced by element n.3
Let sin,n(t)
α1,ne−jξ1,n
x(t) and sout,n(t) βne−jθ
n α1,ne−jξ1,n
x(t),
which denote the equivalent baseband signals of yin,n(t) and
yout,n(t), respectively. The IRS signal reflection model in the
baseband is thus given by
sout,n(t) = βne−jθ
n sin,n(t)
(a)
= βnejθn
sin,n(t), (3)
where γn ∈ [0, 2π] and (a) is due to the fact that the phase
shift is periodic with respect to 2π. Thus, we consider the
IRS phase shift in [0, 2π) for convenience in the sequel of
this paper. From (3), it is observed that in the baseband signal
model, the output/reflected signal of IRS element n is given
by multiplying the corresponding input/impinging signal by a
complex reflection coefficient, βnejθn
.
3Note that to model the narrow-band end-to-end system with IRS, the phys-
ical delay induced by the IRS element is desired to be much less than the
symbol period.

From IRS element n to the receiver, the reflected signal
undergoes a similar equivalent narrow-band frequency-flat
channel given by α2,ne−jξ2,n
. Then the passband signal arriv-
ing at the receiver via IRS element n’s reflection is expressed
as
yr,n(t) = Re

α1,ne−jξ1,n
βnejθn
α2,ne−jξ2,n
x(t) ej2πfct

.
(4)
Thus, the cascaded channel from the transmitter to the receiver
via IRS element n has been modeled. Let h∗
r,n α1,ne−jξ1,n
and gn α2,ne−jξ2,n
. The corresponding baseband signal
model of (4) is given by
yn(t) = βnejθn
h∗
r,ngnx(t). (5)
From (5), it is observed that the IRS reflected channel is a
multiplication of three terms, namely, transmitter-to-element
n channel, IRS reflection, and element n-to-receiver channel.
For simplicity, we assume that there is no signal coupling
in the reflection by neighbouring IRS elements, i.e., all IRS
elements reflect the incident signals independently. Due to the
substantial path loss, we only consider signals reflected by
the IRS for the first time and ignore those reflected by it
two or more times. As such, the received signal from all IRS
elements can be modeled as a superposition of their respective
reflected signals; thus, the baseband signal model accounting
for all the N IRS elements is given by
y(t) =
N
n=1
βnejθn
h∗
r,ngn x(t) = hH
r Θgx(t), (6)
where hH
r = [h∗
r,1, · · · , h∗
r,N ], g = [g1, · · · , gN ]T
, and
Θ = diag(β1 ejθ1
, · · · , βN ejθN
). Note that the IRS with
N elements essentially performs a linear mapping from the
incident (input) signal vector to a reflected (output) signal
vector by an N × N diagonal complex reflecting matrix
Θ, which is diagonal because each IRS element reflects the
signal independently and there is no signal coupling or joint
processing over the IRS elements [8]–[10], [43].
Note that the channel coefficients in hH
r and g generally
depend on distance-related path loss, large-scale shadowing,
and small-scale multipath fading. In particular, the path loss
of IRS-reflected channel captures its average power and is
thus essential to the link budget analysis and performance
evaluation of IRS-aided communications. Without loss of
generality, consider IRS element n, which is assumed to be
located sufficiently far from both the transmitter and receiver,
with the distances from them given by d1,n and d2,n, respec-
tively, as shown in Fig. 4 (a). Under the far-field prorogation
condition, we can assume that d1,n = d1 and d2,n = d2, ∀n.
Then, it follows that E(|hr,n|2
) ∝ c1(d1
d0
)−a1
and E(|gn|2
) ∝
c2(d2
d0
)−a2
, ∀n, where c1(c2) denotes the corresponding path
loss at the reference distance d0, while a1(a2) denotes the
corresponding path loss exponent with typical values from 2
(in free-space propagation) to 6 [7]. From (5), it then follows
that the average received signal power via the reflection by
IRS element n, denoted by Pr,n, is inversely proportional to
da1
1 da2
2 , i.e.,
Pr,n ∝
1
da1
1 da2
2
. (7)
In other words, the IRS reflected channel via element n suffers
from double path loss, which is thus referred to as the product-
distance path loss model. As such, a large number of IRS
reflecting elements are needed in practice to compensate for
the severe power loss due to double attenuation, by jointly
designing their reflection amplitudes and/or phases to achieve
high passive beamforming gains, as will be detailed later in
Section III.
Remark 1: In Fig. 4 (b), the IRS is replaced by an infinitely
large perfect electric conductor (PEC) (or metallic plate).
Assuming free-space propagation and applying the image the-
ory [44], it can be shown that the signal power received at the
receiver via the PEC’s reflection, denoted by Pr, is inversely
proportional to the square of the sum distance of the two-hop
links, i.e.,
Pr ∝
1
(d1 + d2)2
. (8)
This model is usually referred to as the sum-distance path
loss model. Intuitively, due to the reflection of the infinitely
large PEC, the received signal at the receiver were as if from
an equivalent transmitter located at the image point of the
original transmitter as shown in Fig. 1 (b), with the same link
distance d1 + d2, which is also known as specular reflection.
Note that this model is valid for the free-space propagation
with an infinitely large PEC, but in general inapplicable to
the IRS-reflected channel modeled from the element level as
given in (5)–(7). In particular, it is inappropriate to apply
the sum-distance model to the scenario with one or more
finite-size tunable PECs and conclude that the received signal
power scales with the number of PECs by exploiting their
multiplicative passive beamforming gains and at the same time
following the more favorable (as compared to the product-
distance model) sum-distance based path loss, even under the
free-space propagation. More theoretical and/or experimental
studies on this issue can be found in [45]–[49].
B. IRS Architecture, Hardware, and Practical Constraints
The highly controllable reflection of IRS can be practi-
cally achieved by leveraging the existing digitally reconfig-
urable/programmable metasurface [50]. Specifically, metasur-
face is a planar array composed of massive properly designed
reflecting elements/meta-atoms whose electrical thickness is
typically in the order of subwavelength of the signal of interest.
By designing their geometry shape (e.g., square or split-ring),
size/dimension, orientation, arrangement, and so on, desired
signal response (e.g., reflection amplitude and/or phase shift)
of each element/atom can be realized. However, in wireless
communication, the channel is generally time-varying due to
the mobility of the transmitter/receiver as well as the surround-
ing objects, thus calling for real-time tunable response of IRS
based on the channel variation. To this end, IRS elements need
to be manufactured with dynamically adjustable reflection
coefficients and IRS is required to connect to the wireless

Fig. 4. Illustration for the path loss models of IRS-reflected channel.
network to learn the exterior communication environment to
enable its real-time adaptive reflection.
In Fig. 5, we illustrate one typical architecture of IRS, which
consists of three layers and a smart controller. The first/outside
layer is composed of a large number of tunable/reconfigurable
metallic patches printed on a dielectric substrate to directly
manipulate incident signals. In the second/intermediate layer,
a copper plate is usually employed to minimize the signal
energy leakage during IRS’s reflection. It is followed by the
third/inside layer that is a control circuit board responsible
for exciting the reflecting elements as well as tuning their
reflection amplitudes and/or phase-shifts in real time. More-
over, the reflection adaptation is triggered and determined
by a smart controller attached to each IRS, which can be
implemented via field-programmable gate array (FPGA). The
IRS controller also acts as a gateway to communicate with
other network components (e.g., BSs/APs and user terminals)
through wired or wireless backhaul/control links. In practice,
to enhance IRS’s environmental learning capability, dedicated
sensors can also be deployed in the first layer, e.g., interlaced
with the reflecting elements of the IRS, for sensing the
surrounding radio signals of interest to facilitate the smart
controller in designing the reflection coefficients, as will be
discussed in Section IV with more details.
To reconfigure IRS elements for highly controllable
reflection, there are three main approaches proposed in the
literature, namely, 1) mechanical actuation (via, e.g., mechan-
ical rotation and translation), 2) functional materials (e.g.,
liquid crystal and graphene), and 3) electronic devices (e.g.,
positive-intrinsic-negative (PIN) diodes, field-effect transistors
(FETs), or micro-electromechanical system (MEMS) switches)
[51]. In particular, the third approach has been widely adopted
in practical implementation due its fast response time, low
reflection loss as well as relatively low energy consumption
and hardware cost. Fig. 6 shows an example design of the
reflecting element and its equivalent circuit based on the PIN
diode that is loaded in the center of the element. By applying
different biasing voltages to the PIN diode via a direct-current
(DC) feeding line, the PIN diode can be switched to either
“ON” or “OFF” state, which enables the element to result in
a phase-shift difference of π in the incident signal. According
to [52], the PIN diode switching frequency can be up to
5 megahertz (MHz), which corresponds to the switching time
of 0.2 microsecond (μs). This is much smaller than the typical
channel coherence time that is on the order of millisecond (ms)
and thus well suited for mobile applications with time-varying
channels. Besides tuning the phase shift, additional control of
the reflection amplitude of each IRS element provides more
flexibility in reshaping the reflected signal to achieve various
communication objectives effectively. This also offers a flexi-
ble way to trade-off between the hardware cost and reflection
performance in practice, as amplitude control is generally of
lower cost to implement as compared to phase control. There
are various ways to achieve amplitude adjustment for IRS. One
common way is by adjusting the load resistance/impedance in
each element [53]. For example, by changing the resistance of
each element, a certain portion of the incident signal energy
is dissipated as heat, thus achieving a dynamic range of the
reflection amplitude in [0, 1]. This is quite similar to the
operation of a passive radio frequency identification (RFID)
tag that controls the strength of the reflected signal power by
varying its load impedance for data modulation. In practice,
it is desirable to have independent control of the amplitude and
phase shift of each IRS element for optimizing the reflection
design, which, however, requires more sophisticated hardware
designs (e.g., the multilayer surface design [51]) than the
above mentioned for their separate control only.
Ideally, the IRS reflection amplitude and phase shift per
element can be independently and continuously tuned, thus
yielding their following respective feasible sets:
Fβ =

β β ∈ [0, 1]

, (9)
Fθ =

γ γ ∈ [0, 2π)

. (10)
Alternatively, by defining ρ βejθ
as the reflection coeffi-
cient per element, we obtain the following equivalent feasible
set as
F =

ρ |ρ| ≤ 1

. (11)
Note that the above ideal feasible sets for IRS reflecting
elements usually lead to theoretical performance limits of
IRS-aided wireless communication systems under practical
reflection models, which are introduced as follows.
1) Discrete Reflection Amplitude and Phase Shift: While
tuning the reflection coefficient continuously is beneficial for

Fig. 5. Architecture of IRS.
Fig. 6. An example of the tunable reflecting element based on PIN diode.
optimizing the communication performance, it is practically
difficult to implement since higher-resolution reflecting ele-
ments require not only increased cost but also more complex
hardware design. For example, to enable 8 levels of phase
shifts per IRS element, at least log2 8 = 3 PIN diodes are
required. This renders the element design more challenging
due to its limited size, and also requires more controlling
pins at the IRS controller to control the required PIN diodes.
Although a single varactor diode can be used to achieve
multi-level phase shifts, it requires a wide range of biasing
voltages, and thus is more costly to implement. As such, for
practical IRSs that usually have a large number of reflecting
elements, it is more cost-effective to implement only discrete
and finite amplitude/phase-shift levels that require only a
small number of control bits for each element, e.g., two-level
(reflecting or absorbing) amplitude control, and/or two-level
(0 or π) phase-shift control [50], [54].
In general, let bβ and bθ denote the number of bits for
controlling the corresponding number of reflection amplitude
and phase-shift levels denoted by Kβ and Kθ, respectively,
where Kβ = 2bβ
and Kθ = 2bθ
. Then the sets of discrete
reflection amplitudes and phase shifts at each element of IRS
can be respectively expressed as
F
β = {β̄1, · · · , β̄Kβ
}, (12)
F
θ = {γ̄1, · · · , γ̄Kθ
}, (13)
where 0 ≤ β̄m β̄m ≤ 1 for 1 ≤ m m
≤ Kβ and
0 ≤ γ̄l γ̄l 2π for 1 ≤ l l
≤ Kθ. For example,
by assuming that the discrete amplitude and phase-shift values
are obtained by uniformly quantizing the interval [0, 1] and
[0, 2π), respectively, we have
F
β = {0, Δβ, · · · , (Kβ − 1)Δβ}, (14)
F
θ = {0, Δγ, · · · , (Kθ − 1)Δγ}, (15)
where Δβ = 1/(Kβ −1) and Δγ = 2π/Kθ. Compared to the
ideal continuous reflection amplitude/phase-shift models in 9
and (10), their quantized versions in (14) and (15) inevitably
result in coarser reflected signal amplitude/phase control and
thus degraded communication performance. Furthermore, they
also bring new challenges in optimizing the IRS reflecting
elements’ amplitudes/phase-shifts in practice due to the resul-
tant discrete optimization variables that are in general more
difficult to handle than their continuous counterparts, as will
be discussed in Section III with more details.
In practice, to further reduce the hardware cost and design
complexity, only discrete phase-shift control or discrete ampli-
tude control may be implemented, thus leading to the follow-
ing two special cases of the above discrete models, namely,
• IRS with discrete phase-shift control only, where for each
reflecting element, only the phase shift can be tuned while
the reflection amplitude is set to its maximum value of
one, i.e., F
β =

β β = 1

and F
θ = {0, Δγ, · · · , (Kθ−
1)Δγ};
• IRS with discrete amplitude control only, where for
each reflecting element, only the reflection amplitude
can be tuned while the phase shift is set to be a con-
stant (say, zero without loss of generality), i.e., F
β =
{0, Δβ, · · · , (Kβ − 1)Δβ} and F
θ =

γ γ = 0

.
Generally speaking, phase-shift control (or phase beamform-
ing) is of higher cost to implement as compared to amplitude

Fig. 7. Reflection amplitude versus phase shift for the practical IRS reflecting
element [55], [56].
control (or amplitude beamforming) for IRS, while the former
can achieve better passive beamforming performance than the
latter given the same number of control bits/discrete levels
per reflecting element. A detailed study on their performance
comparison will be given in Section III.
2) Coupled Reflection Amplitude and Phase Shift: Although
independent control of the reflection amplitude and phase
shift simultaneously provides the maximum design flexibil-
ity, it imposes challenges in the element design. Recently,
a practical reflection model for IRS was proposed in [55],
[56] by modelling each reflecting element as a resonant
circuit with certain inductance, capacitance, and resistance.
Based on this model, it was revealed that the amplitude
response of the reflecting element is in general non-linearly
coupled with its phase shift, which thus are not independently
adjustable. Specifically, as shown in Fig. 7, the reflection
amplitude typically attains its minimum value at the zero
phase shift, but monotonically increases and asymptotically
approaches the maximum value of one as the phase shift tends
to π or −π. This is due to the fact that in each reflecting
element, any phase shift is achieved by tuning its effective
capacitance/resistance, which inevitably changes the reflection
amplitude. To be specific, when the phase shift approaches
zero, the reflective currents (also referred to as image currents)
are in-phase with the element currents, and thus the electric
field as well as the current flow in the element are both
enhanced, which results in maximum energy dissipation and
the lowest reflection amplitude. In contrast, when the phase
shift is around π or −π, the reflective currents are out-of-
phase with the element currents, and thus the electric field as
well as the current flow in the element both diminish, thus
leading to minimum energy loss and the highest reflection
amplitude. This analytical reflection model of IRS is also
consistent with the experimental results reported in e.g., [57].
The reflection amplitude and phase-shift coupling has a great
impact on the optimal reflection coefficient design of IRS
reflecting elements, as it needs to strike an optimal balance
between the signal amplitude and phase reflected by each
element such that the reflected signals by all IRS elements are
combined at the receiver with maximum power or achieving
maximum signal-to-noise ratio (SNR).
In addition to the ideal versus practical IRS reflection
models discussed in the above, IRS is theoretically passive
but consumes energy in practice as well. Although IRS only
reflects signal without power amplification, it still needs a
power supply to sustain the operation of reconfiguring the
reflecting elements as well as its smart controller. For example,
if PIN diodes are used to tune the phase shifts of IRS
reflecting elements, when the PIN diode is OFF, the element
does not consume any energy, whereas when the PIN diode
is ON, it consumes a certain amount of power, e.g., about
0.33 milliwatt (mW) in [46] and 50 μW in [58]. Nevertheless,
compared to the power consumption of active antennas (e.g.,
massive MIMO and multi-antenna relay), such power con-
sumption is much lower and thus can be practically ignored for
comparison. On the other hand, the power consumption of the
smart controller will depend on the controller’s circuit imple-
mentation (e.g., FPGA) and communication module used. For
example, the power consumption of an IRS controller with
256 reflecting elements is only about 0.72 W [46], which is
significantly lower than that of the active BS/relay in practice.
Therefore, IRS generally consumes substantially lower energy
consumption than the existing active relays in the wireless
network. Interested readers may refer to [46], [51], [58] for
more details on IRS energy consumption modelling as well
as the impact of energy consumption on its performance
comparison with conventional active relays.
C. Other Related Work and Future Direction
IRS reflection modelling and practical hardware design are
critical to its reflection optimization and achievable perfor-
mance gains in IRS-aided wireless systems. However, their
research is still in an early stage and needs to address many
interesting and important problems that are open and thus
worth further investigating. In the following, we list some of
these promising topics to motivate future work.
One crucial assumption for the linear channel model of IRS
given in (6) is the ignorance of any reflected signal coupling
among neighbouring IRS elements. In practice, increasing
the number of reflecting elements given the same IRS size
is generally helpful in achieving more fine-grained passive
beamforming and thus enhanced performance. This, however,
will reduce the element spacing and may render the mutual
coupling more severe and thus no more negligible, as nearby
reflecting elements will interact with each other through
their circuit coupling and thus result in coupled reflection
coefficients. As such, the linear channel model of IRS given
in (6) will become inaccurate and more complex non-linear
modelling may be needed to characterize the mutual coupling
effect, which is an interesting problem to address in future
work. Furthermore, it is also worth developing efficient decou-
pling/isolation techniques to minimize the effect of mutual
coupling by e.g., using defected ground structures, parasitic
scatterers, and neutralization lines [51], [59].

Another assumption adopted in the IRS channel model given
in (6) is that the reflection coefficient is insensitive to the
incident angle of the signal impinging on IRS. However, recent
experimental results in [46] revealed that the IRS reflection
coefficient, particularly the phase shift, can be highly sensitive
to the incident angle. Such an angle-dependent reflection
model poses a new challenge in IRS reflection optimization,
especially in a multi-path propagation environment since the
reflection response of IRS depends on the angle-of-arrival
(AoA) of each signal path to the IRS. More importantly,
the widely adopted assumption of channel reciprocity in
typical time-division duplex (TDD) based wireless systems
may become no longer valid when IRS is involved, which
implies that the uplink training based channel estimation is
not applicable to the donwlink communication. Thus, further
studies on this effect are worth pursuing.
Last but not least, the IRS channel model in (6) has assumed
that the phase shift of each reflecting element is constant over
the signal bandwidth, B. Although this assumption is valid
for the narrow-band system with B fc, it may become
inaccurate for the broadband system when B is much larger
and thus becomes comparable with fc. As shown in (2),
the phase shift by IRS is resulted by delaying the input signal
with a certain amount of time, which will cause a linear phase
drift that increases with the signal frequency deviating from
fc. Such a non-uniform phase shift over frequency can cause
undesired phase errors for multi-carrier (such as orthogonal
frequency-division multiplexing (OFDM)) modulated signal,
and thus should be compensated by either hardware or signal
processing, which needs further investigation. In addition,
the effects of other IRS hardware imperfections such as
frequency/time offset, phase noise, etc. on its reflection design
and achievable performance remain largely unknown, which
call for further investigation in future work.
III. IRS REFLECTION OPTIMIZATION
In this section, we study the passive reflection optimization
for IRS-aided wireless communication under various system
setups, namely, from single-user to multi-user, from single-
antenna to multi-antenna, from narrow-band to broadband,
as well as from single-cell to multi-cell communications. For
the purpose of exposition, we assume perfect knowledge of
all channels considered in this section, while the issue of IRS
channel estimation will be addressed in Section IV. In Fig. 8,
we show a single-cell IRS-aided multi-user system, where an
IRS comprising N reflecting elements is deployed to assist
in the communications from the AP (or BS) to a set of K
users in the downlink. The users are located arbitrarily in the
cell, so some of them may not be in the vicinity of the IRS
in general. It is assumed that the AP is equipped with Mt
antennas and each user is equipped with Mr antennas without
loss of generality.
A. IRS-Aided SISO System: Passive Beamforming Basics and
Power Scaling Order
First, we consider the single-user single-input-single-output
(SISO) case, i.e., K = 1 and Mt = Mr = 1, and the
Fig. 8. IRS-aided multi-user communication system.
narrow-band system with flat-fading channels. The baseband
equivalent channels from the AP to the IRS, from the IRS
to the user, and from the AP to the user are denoted by
g ∈ CN×1
, hH
r ∈ C1×N
, and h∗
d ∈ C, respectively. Based
on the IRS reflection model in (6), the signal received at the
user is expressed as
y = (hH
r Θg + h∗
d)

Ptx + z, (16)
where x is the information signal modeled by an independent
and identically distributed (i.i.d.) random variable with zero
mean and unit variance, Pt is the transmit power at the AP,
and z denotes additive white Gaussian noise (AWGN) at the
user receiver modeled as CSCG with zero mean and variance
σ2
. Accordingly, the user receive SNR is written as
γ =
Pt|hH
r Θg + h∗
d|2
σ2
=
Pt|
N
n=1 h∗
r,nβnejθn
gn + h∗
d|2
σ2
. (17)
Thus, the maximum achievable rate in bits per second per
Hertz (bps/Hz) of the considered IRS-aided point-to-point
SISO link is given by r = log2(1 + γ).
We aim to maximize the achievable rate r (or SNR γ equiv-
alently) by optimizing the passive reflect beamforming at the
IRS. By ignoring the constant terms and assuming continuous
reflection amplitude and phase shift, the optimization problem
is formulated as
(P1): max
θ,β
N
n=1
h∗
r,ngnβnejθn
+ h∗
d
2
(18)
s.t. 0 ≤ γn 2π, n = 1, · · · , N, (19)
0 ≤ βn ≤ 1, n = 1, · · · , N, (20)
where θ = [γ1, · · · , γN ]T
and β = [β1, · · · , βN ]T
. For given
β ≥ 0, the optimal phase-shift solutions to (P1) are given by
[8], [9]
γ
n = mod[ζ − (φn + ρn), 2π], n = 1, · · · , N, (21)
where φn, ρn, and ζ are the phases of h∗
r,n, gn, and h∗
d,
respectively. Note that the solutions in (21) do not depend on
the values in β, which implies that they are indeed optimal to
(P1). This is expected since the optimal phase shifts should
align all the signals reflected by the IRS regardless of their

strength with the signal coming directly from the AP to
achieve coherent combining and thus maximize the received
signal power at the user. Besides, if the AP-user direct link
is negligible compared to the IRS-reflected link (e.g., when
the former is severely blocked) and thus can be ignored,
i.e., h∗
d = 0, the optimal solutions in (21) can be multiplied
with any arbitrary (common) phase shift without changing the
optimal value of (P1). Thus, we can set ζ equal to zero in (21)
without loss of optimality. In general, by substituting (21) into
(P1), this problem is reduced to
(P1.1): max
β
N
n=1
|hr,n||gn|βn + |hd|
2
(22)
s.t. 0 ≤ βn ≤ 1, n = 1, · · · , N. (23)
It follows easily from (22) that the optimal reflection amplitude
solutions are given by β
n = 1, ∀n, since maximizing the
reflection amplitudes helps achieve the maximum user receive
power due to coherent combining. Another interesting observa-
tion from (P1) is that the optimal IRS reflection design depends
on the cascaded channels via the IRS only, i.e., diag(hH
r )g,
without the need of knowing the individual channels, hH
r and
g. It is worth mentioning that this result usually holds for
IRS-aided communication systems and can be leveraged to
greatly simplify the IRS channel estimation design as will be
discussed in Section IV.
1) Receive Power Scaling With N: A fundamental ques-
tion regarding IRS passive beamforming performance is how
the maximum receive SNR grows with the number of IRS
reflecting elements, N, when N becomes asymptotically large.
To address this problem, we focus on the IRS-reflected link
only by assuming that the AP-user direct channel h∗
d equals to
zero, since the former becomes more dominant than the latter
as N → ∞. Based on (21), the user receive power, denoted
by Pr, is given by Pr = Pt
N
n=1 |hr,n||gn|
2
. Assume i.i.d.
Rayleigh fading with average power 2
h and 2
g for each entry
in hH
r and g, respectively. It was shown in [9] that as N → ∞,
the asymptotic receive power is approximately given by
Pr ≈ N2
Ptπ2 2
h
2
g
16
. (24)
This result suggests that with sufficiently large N, the user
receive power increases quadratically with N, i.e., in the order
of O(N2
) [9]. Alternatively, we can scale down the transmit
power at the AP by a factor of 1/N2
without compromising
the receive SNR. This is because the IRS not only achieves
the reflect beamforming gain of O(N) in the IRS-user link,
but also captures an additional aperture gain of O(N) in the
AP-IRS link. Furthermore, it is not difficult to show that such
a receive power scaling law also holds under LoS channel (see
(53)) as well as Rician fading channel.
To illustrate the receive power scaling order, we consider
a setup shown in Fig. 9. The path loss model for both AP-
IRS and IRS-user links is set as c0d−a
0 where d0 is the
corresponding link distance in meter (m) and a denotes the
path loss exponent. Other parameters are set as c0 = −30 dB,
Pt = 50 mW and σ2
= −90 dBm. In Fig. 10, we plot the
achievable rate versus the number of IRS reflecting elements,
Fig. 9. Simulation setup of the IRS-aided single-user system.
Fig. 10. Achievable rate versus N for IRS under different channel models.
N, with d = d1 = 50 m. Specifically, we consider three
different channel models, namely, free-space LoS channel,
Rician fading channel with a Rician factor of 3 and i.i.d.
Rayleigh fading channel (a special case of the Rician fading
channel without the LoS component or with zero Rician
factor). The path loss exponents of the three channel models
are set as 2, 2.4 and 2.8, respectively. From Fig. 10, it is
observed that for all considered channel models, the achievable
rate of the IRS-aided SISO system increases about 2 bps/Hz by
doubling N when N is sufficiently large, e.g., from N = 200
to N = 400, which validates the asymptotic power scaling
order in (24).
In practice, it is useful to estimate the coverage range of IRS
for a given average receive power (or SNR) requirement of the
user with a fixed transmit power at the AP. By substituting
2
h = c0d−a
2 (with d2 shown in Fig. 9) into (24), it follows
that
N ≈ d
a
2
2 c1, (25)
where c1 =

Pr
Ptc0
4
g
. From (25), it follows that as the desired
IRS coverage range increases, the number of its reflecting
elements also needs to be increased to keep the same average
receive power without inuring more transmit power at the AP,
e.g., by increasing N linearly with d2 if a = 2.
2) Comparison With M-MIMO and MIMO Relay: It is
worth comparing the performance of IRS with N passive
reflecting elements to that of M-MIMO or MIMO relay with
the same number (N) of active antennas. For M-MIMO,

it is well known that a transmit/receive beamforming gain of
order O(N) can be achieved [60], [61]. While it was shown
in [9] that even with perfect self-interference cancellation,
the receive SNR by using the FD AF-based MIMO relay also
increases linearly with N when N is asymptotically large. This
is due to the processing noise effect at the AF relay. To be
specific, although the user receive signal power in the FD AF
MIMO relay system scales in the same order of N2
as that of
the IRS-aided system, the forwarded relay noise power at the
receiver scales linearly with N in contrast to the noise-free
signal reflection by the IRS, thus resulting in a lower receive
SNR gain order of O(N). Last, it is worth mentioning that
for the HD AF MIMO relay system, its receive SNR scaling
order with N can be shown to be identical to that of its FD
counterpart.
In Fig. 11, we compare the achievable rates of the aforemen-
tioned three technologies versus N. Rayleigh fading channel
with the path loss exponent of 2.4 is assumed for both AP-
IRS and IRS-user links (with the IRS replaced by a multi-
antenna relay in the MIMO relay case), and other parameters
are set the same as for Fig. 10. It is observed that when N
is small, IRS performs the worst due to the severe product-
distance path loss (see (7) in Section II) but insufficient passive
beamforming gain. However, with the increasing number of
reflecting elements, the rate gaps between the IRS-aided sys-
tem and the M-MIMO and FD/HD MIMO relay-aided systems
become smaller and IRS even outperforms the HD MIMO
relay thanks to its larger beamforming gain (i.e., O(N2
) versus
O(N)) when N is large as well as its more spectral-efficient
FD operation. Furthermore, since IRS’s reflecting elements
are passive and do not need any transmit RF chains, their
cost is much lower as compared to that of active antennas
for the M-MIMO/MIMO relay. Therefore, it may be unfair
to compare their achievable rates with the same number of
passive/active elements as in Fig. 11. Instead, with a given
hardware cost, IRS can potentially offer significantly more
reflecting elements as compared to affordable antennas in
the case of M-MIMO/MIMO relay and thus achieve more
comparable or even superior performance against them in
practice.
B. IRS-Aided MISO System: Joint Active and Passive
Beamforming
Next, we consider the single-user system but with mul-
tiple antennas at the AP, i.e., Mt 1, where the down-
link/uplink communication becomes the multiple-input-single-
output (MISO) and single-input-multiple-output (SIMO) sys-
tem, respectively. In this case, the active AP transmit/receive
beamforming needs to be optimized jointly with the IRS
passive beamforming. For brevity, we consider only the MISO
case in this subsection while the results given are applicable
to the SIMO case in the reverse user-AP link as well. Let G
and hH
d denote the channel matrix and channel vector from
the AP to IRS and from the AP to user, respectively. Similar
to (P1), the rate/SNR maximization problem in the MISO case
Fig. 11. Performance comparison of M-MIMO, IRS and FD/HD AF-based
MIMO relay.
is equivalently formulated as4
(P2): max
w,θ
|(hH
r ΘG + hH
d )w|2
(26)
s.t. w2
≤ Pt, (27)
0 ≤ γn 2π, n = 1, · · · , N, (28)
where w denotes the transmit beamforming vector at the AP
and Pt is the maximum transmit power. Intuitively, if the
channel of the AP-user link, hH
d , is much stronger than that of
the AP-IRS link, G, it is preferable for the AP to beam toward
the user directly, while in the opposite case, especially when
the AP-user link is severely blocked, the AP should steer its
beamforming direction toward the IRS to leverage its reflected
signal to serve the user. In general, the transmit beamforming
at the AP needs to be jointly designed with the phase shifts
at the IRS based on all the AP-IRS, IRS-user, and AP-user
channels in order to maximize their cooperative gain.
Unfortunately, (P2) is a non-convex optimization problem
due to the non-concavity of its objective function with respect
to w and θ. In fact, the coupling between the active beam-
forming and passive beamforming optimization variables is
one major difficulty in their joint optimization. Nevertheless,
one can observe that if we fix the transmit beamforming vector
w, (P2) is reduced to its SISO counterpart, i.e., (P1), which
thus motivates the alternating optimization (AO) approach
to solve (P2) sub-optimally [9]. Specifically, with fixed w,
we obtain the optimal phase shifts by solving the resulted (P1);
while with fixed θ, the optimal transmit beamforming solution
to (P2) is given by the maximum-ratio transmission (MRT),
i.e., wMRT =
√
Pt
(hH
r ΘG+hH
d )H
hH
r ΘG+hH
d
. The above procedure is
repeated until convergence is reached, which guarantees attain-
ing at least a locally optimal solution to (P2) [9].
An alternative approach to solve (P2) is by first reducing it
into an optimization problem with phase shifts only. Specif-
4Note that the reflection amplitudes of all IRS elements are set to one or their
maximum value since with the optimal transmit beamforming vector w,
the objective function of (P2) is reduced to that of (P1), thus β
n = 1, n =
1, · · · , N, are optimal as shown in Section III-A.

ically, by substituting wMRT to (26), (P2) is reduced to the
following problem with θ only,
(P3): max
θ
hH
r ΘG + hH
d 2
(29)
s.t. 0 ≤ γn ≤ 2π, n = 1, · · · , N. (30)
In contrast to (P1) for the SISO case, the IRS phase shifts in
(P3) for the MISO case need to balance the channel powers
from different transmit antennas, thus are more difficult to
solve in general. In fact, it is a non-deterministic polynomial-
time hard (NP-hard) problem with respect to N in general
and thus it is impossible to obtain the optimal solution to (P3)
with polynomial complexity of N.5
To show this, let v =
[v1, · · · , vN ]H
where vn = ejθn
, ∀n. Then, the constraints in
(30) are equivalent to the unit-modulus constraints: |vn|2
=
1, ∀n. By applying the change of variables hH
r ΘG = vH
Φ
where Φ = diag(hH
r )G ∈ CN×Mt
, (P3) is equivalent to
(P4): max
v
vH
ΦΦH
v + vH
Φhd + hH
d ΦH
v + hH
d 2
(31)
s.t. |vn|2
= 1, n = 1, · · · , N. (32)
Note that (P4) is a non-convex quadratically constrained
quadratic program (QCQP), which is generally NP-hard to
solve [63]. In particular, the unit-modulus constraints are non-
convex and difficult to handle, which thus impose another
challenge in the optimization algorithm design. As such,
various methods have been proposed in the literature to attain
high-quality suboptimal solutions for (P4) including, e.g., 1)
semidefinite relaxation (SDR) with Guassian randomization
[9], and 2) AO where each of the phase shifts is optimized
in closed-form as in [64] with the others being fixed in an
iterative manner, which guarantees a locally optimal solution.
In addition, the SDR approach with matrix rank relaxation
also provides a performance upper bound and thus is usually
adopted for benchmarking the performance of suboptimal
solutions [9]. Interested readers may refer to [63], [65] for
details of the above two methods. Moreover, the above meth-
ods can be jointly applied, e.g., the solution obtained by
scheme 1) can be further improved by adopting it as an
initialization for scheme 2). Other approaches to tackle (P4)
may include penalty-based method [66], successive convex
approximation (SCA) [67], [68], manifold optimization [69],
deep learning based method [70], alternating direction method
of multipliers (ADMM), etc.
In Fig. 12, we show the achievable rate of the IRS-aided
MISO system versus the AP-user horizontal distance, d, for
the same setup as in Fig. 9 with Mt = 4 and N = 40. Rayleigh
fading channels are assumed for both the AP-user and IRS-
user links, while the free-space LoS channel is adopted for the
AP-IRS link. Four schemes are considered, namely, 1) the joint
transmit and passive beamforming design by AO [64], 2) AP-
user MRT by setting w =
√
Pt
hd
hd to achieve MRT based
on the AP-user direct channel, 3) AP-IRS MRT by setting
w =
√
Pt
g
g to achieve MRT based on the AP-IRS rank-
one channel, with gH
denoting any row in G, and 4) MRT
5If there exist special channel structures such as rank-one G, the optimal
phase shifts for (P3) can be obtained with a polynomial time complexity [62].
Fig. 12. Achievable rate of IRS-aided MISO system versus AP-user
horizontal distance, d.
without IRS by setting w =
√
Pt
hd
hd . It is observed from
Fig. 12 that the AP-user MRT scheme performs close to the
AO scheme when the user is nearer to the AP, while it results
in considerable rate loss when the user is nearer the IRS. This
is expected since in the former case, the user received signal
is dominated by the AP-user direct link whereas the IRS-user
link is dominant in the latter case. Moreover, it is observed
that the AP-IRS MRT scheme behaves in the opposite way
as compared to the AP-user MRT counterpart. Moreover,
Fig. 12 shows that if the transmit beamforming is not designed
properly based on all the channels, the achievable rate by using
the IRS may be even worse than the conventional MRT without
the IRS, e.g., in the case of the AP-IRS MRT scheme for
d ≤ 40 m. This demonstrates that the joint active and passive
beamforming design is essential to strike an optimal balance
between the transmission toward the user directly and that via
the IRS reflection, to maximize the received signal power at
the user.
C. IRS-Aided MIMO and OFDM Systems
In the above, we have considered narrow-band communi-
cation systems under frequency-flat fading channels, where
the AP and/or user are equipped with a single antenna (i.e.,
Mt = 1 and/or Mr = 1). However, for MIMO systems with
multiple antennas at both the AP and user (i.e., Mt 1
and Mr 1) and/or broadband OFDM systems subject to
frequency-selective fading channels, the passive IRS reflection
optimization problems need to cater to multi-antenna channels
and/or multi-path channels with different delays, thus are
more complicated as well as challenging to solve. In this
subsection, we present the models for IRS-aided MIMO and
OFDM systems, and the new methods used to optimize the
IRS reflection for them.
1) IRS-Aided MIMO System: Different from
SISO/MISO/SIMO systems that can support the transmission
of one single data stream only, the capacity of MIMO systems
is generally achieved by parallel transmissions of multiple
data streams at the same time, or spatial multiplexing.

Therefore, to achieve the capacity of an IRS-aided MIMO
system, the IRS reflection needs to be jointly optimized with
the transmit covariance matrix at the AP [71]. Specifically,
let Hd ∈ CMr×Mt
denote the direct channel from the AP to
the user, G ∈ CN×Mt
denote the channel from the AP to the
IRS, and Hr ∈ CMr×N
denote the channel from the IRS to
the user. The overall IRS-aided MIMO channel can thus be
expressed as H̃ = Hd + HrΘG. Let Q ∈ CMt×Mt
denote
the transmit covariance matrix. By considering continuous
phase shifts and maximum reflection amplitudes at all IRS
elements (i.e., βn = 1, ∀n),6
the capacity optimization
problem is formulated as
(P5): max
Θ,Q
log2 det

IMr +
1
σ2
H̃QH̃
H

(33)
s.t. 0 ≤ γn 2π, n = 1, · · · , N, (34)
βn = 1, n = 1, · · · , N, (35)
tr(Q) ≤ Pt, (36)
Q 0. (37)
Note that (P5) is a non-convex optimization problem due
to the non-concavity of the MIMO channel capacity with
respect to Θ and Q jointly; moreover, (P5) is more difficult to
solve as compared to (P1) and (P2) for the SISO and MISO
cases as the MIMO channel capacity is a more complicated
log-determinant function. Nevertheless, an efficient AO-based
algorithm was proposed in [71] to solve (P5), by iteratively
optimizing one IRS element’s phase shift or the transmit
covariance matrix at each time, with the other variables being
fixed. Specifically, with given transmit covariance matrix and
any N − 1 IRS phase shifts, the optimal solution of the
remaining phase shift to (P5) can be obtained in closed-form
[71]; on the other hand, with given IRS reflection matrix Θ
and consequently the overall MIMO channel H̃, the optimal
solution of the transmit covariance matrix is given by the well-
known eigenmode transmission with water-filling based spatial
power allocation [6], [7]. The above AO-based algorithm is
guaranteed to converge to at least a locally optimal solution
to (P5), with only polynomial complexity in terms of N,
Mt, or Mr [71].
It is worth noting that by properly designing the IRS reflec-
tion coefficients, various key parameters of the IRS-enhanced
MIMO channel can be significantly improved, including the
channel total power, condition number, rank, etc [71]. Particu-
larly, for practical scenarios where the direct channel between
the AP and user is of low rank or even rank-one (e.g., LoS-
dominant channel, key-hole channel), the AO-based algorithm
results in a higher-rank MIMO channel with larger spatial
multiplexing gain as compared to the MIMO channel without
IRS, thus substantially boosting the channel capacity in the
high-SNR regime. For illustration, we show in Fig. 13 the
achievable rate versus transmit power Pt for an IRS-aided
MIMO system with Mt = Mr = 4 and N = 30. The system
6Note that the unit-amplitude assumption here is mainly motivated by the
practical difficulty to jointly tune the amplitude and phase shift of each IRS
reflecting element at the same time, while considering flexible amplitude
design within [0, 1] may lead to improved capacity for IRS-aided MIMO
systems in general [71].
Fig. 13. Achievable rate versus transmit power for IRS-aided MIMO system.
setup is similar to that in Fig. 9, but with d1 = d = 4000 m and
d2 = 1 m. We assume that the AP-user direct channel follows
the free-space LoS model with path loss exponent 2 (thus
being rank-one); the AP-IRS channel follows the Rician fading
model with Rician factor 0.5 and path loss exponent 2.1;
and the user-IRS channel follows the Rayleigh fading model
with path loss exponent 2.1. It is observed from Fig. 13 that
the achievable rate of the AO-based algorithm is significantly
larger than the MIMO channel capacity without IRS, as well as
the achievable rate of a benchmark scheme with the phase shift
of each IRS reflecting element independently and uniformly
generated in [0, 2π) and the corresponding optimal transmit
covariance matrix.
2) IRS-Aided OFDM System: Next, we extend the IRS
reflection design to OFDM systems under frequency-selective
fading channels. Despite the broadband system, we assume
that the system bandwidth is practically much smaller than
the carrier frequency, i.e., B fc for simplicity. To illustrate
the new signal and channel models in this case, we first
consider that both the AP and user are equipped with a
single antenna, i.e., Mt = Mr = 1. Let L0 denote the
number of delayed taps in the AP-user direct channel; L1
and L2 denote that in the AP-IRS and IRS-user channels,
respectively. Let hd
= [hd
0, · · · , hd
L0−1]T
∈ CL0×1
denote
the baseband equivalent time-domain channel for the AP-
user direct link; hr1
n = [hr1
n0, · · · , hr1
n(L1−1)]T
∈ CL1×1
and
hr2
n = [hr2
n0, · · · , hr2
n(L2−1)]T
∈ CL2×1
denote that for the AP-
IRS and IRS-user links associated with IRS reflecting element
n, respectively. Similar to Section II, let x(t) denote the
baseband equivalent transmit signal at the AP. The passband
signal impinging on IRS element n is thus given by
yin,n(t) = Re
L1−1
l=0
hr1
nlx(t − τl(t))

ej2πfct

, (38)
where τl(t) = l/B denotes the time delay of the lth tap. The
reflected signal by IRS element n is then expressed as
yout,n(t)
= βnyin,n(t − tn)

= Re

βn
L1−1
l=0
hr1
nlx(t − tn − τl(t))

ej2πfc(t−tn)

≈ Re

βnejθn
L1−1
l=0
hr1
nlx(t − τl(t))

ej2πfct

, (39)
where tn ∈ [0, 1/fc] denotes the time delay at IRS element
n; x(t − tn − τl(t)) ≈ x(t − τl(t)) holds since tn ≤ 1/fc
1/B; and γn = 2π − 2πfctn ∈ [0, 2π) denotes the phase
shift at IRS element n. After undergoing the IRS-user channel,
the passband signal at the user receiver via IRS element n’s
reflection is given by
yr,n(t) = Re
L2−1
l=0
hr2
nl βnejθn
×
L1−1
l=0
hr1
nlx(t − τl(t) − τl (t))

ej2πfct

. (40)
Accordingly, the baseband signal model of (40) can be
expressed as
yn(t) = (hr2
n βnejθn
hr1
n )x(t)
Δ
= hr
nx(t), (41)
where hr
n = hr2
n βnejθn
hr1
n = βnejθn
(hr2
n hr1
n ) ∈ CLr
×1
denotes the cascaded AP-IRS-user channel via IRS element n,
with Lr
= L1 + L2 − 1. Note that the cascaded channel is the
convolution of the AP-IRS multi-path channel, the (single-tap)
IRS reflection coefficient, and the IRS-user multi-path channel.
We consider an OFDM system where the total bandwidth is
equally divided into Q ≥ 1 orthogonal sub-bands each with a
different central sub-carrier frequency. For ease of exposition,
let g̃r
n = [(hr2
n hr1
n )T
, 0, · · · , 0]T
∈ CQ×1
denote the zero-
padded convolved AP-IRS and IRS-user time-domain channel
via IRS element n, and define G̃
r
= [g̃r
1, · · · , g̃r
N ] ∈ CQ×N
.
The cascaded AP-IRS-user channel can thus be expressed
as h̃
r
= G̃
r
θ, where θ = [β1ejθ1
, · · · , βN ejθN
]T
. Let
h̃
d
= [hdT
, 0, · · · , 0]T
∈ CQ×1
denote the zero-padded
time-domain AP-user direct channel. The superposed effective
channel impulse response (CIR) is thus given by
h̃ = h̃
d
+ h̃
r
= h̃
d
+ G̃
r
θ. (42)
Note that the number of delayed taps in the effective AP-user
channel (or non-zero entries in h̃) is L = max(L0, L1 +L2 −
1). By further assuming that the cyclic prefix (CP) length of
each OFDM symbol is no smaller than L (so that the inter-
symbol interference can be eliminated), the channel frequency
response (CFR) at each sub-carrier q can be expressed as
cq = fH
q h̃
d
+ fH
q G̃
r
θ, q = 1, · · · , Q, (43)
where fH
q denotes the qth row of the Q × Q discrete Fourier
transform (DFT) matrix. It is worth noting from (43) that
the IRS phase-shift values in θ impact the CFR at each
OFDM sub-carrier identically, i.e., without frequency selec-
tivity. By ignoring the rate loss due to the CP insertion,
the achievable rate of the IRS-aided OFDM system in bps/Hz
is given by
r =
1
Q
Q
q=1
log2
⎛
⎝1 +
pq|fH
q h̃
d
+ fH
q G̃
r
θ|2
σ̄2
⎞
⎠, (44)
Fig. 14. Achievable rate versus transmit power for IRS-aided OFDM system.
where pq denotes the transmit power allocated to sub-carrier
q with
Q
q=1 pq ≤ Pt, and σ̄2
denotes the average receiver
noise power at each sub-carrier.
To maximize the achievable rate in (44), the IRS phase
shifts in θ need to cater to the frequency-varying channels at
different sub-carriers, or equivalently, the time-domain chan-
nels at different delayed taps. Moreover, θ needs to be jointly
optimized with the transmit power allocations over the Q sub-
carriers, {pq}Q
q=1, thus rendering the resultant optimization
problem more difficult to solve as compared to (P1) in the
narrow-band case. To tackle this problem, an efficient SCA
based algorithm was proposed in [68] by approximating the
non-concave rate function in (44) using its concave lower
bound based on the first-order Taylor expansion. The SCA-
based algorithm is guaranteed to converge to a stationary
point of the joint IRS reflection and transmit power optimiza-
tion problem, and requires only polynomial complexity over
N or Q [68]. To further lower the complexity, [72] proposed a
simplified algorithm where the IRS phase shifts are designed
to only align with the time-domain channel with strongest
path power, thus termed as the “strongest CIR maximization”.
In Fig. 14, we consider an IRS-aided OFDM system with
Mt = Mr = 1, N = 30, Q = 64, and show the achievable
rate versus transmit power Pt for various schemes. The system
setup is similar to that in Fig. 9, but with d1 = 200 m,
d2 =
√
26 m, and d = 199 m. All the channels involved
are assumed to follow the Rician fading model. The Rician
factors for the direct AP-user channel, AP-IRS channel, and
IRS-user channel are set as 0, 3 dB, and −20 dB, respectively;
while the corresponding path loss exponents for the three
channels are set as 3.5, 2.8, and 2.2, respectively. The number
of delayed taps in the AP-user direct channel is set as L0 = 16,
while those in the AP-IRS channel and IRS-user channel
are set as L1 = 4 and L2 = 13, respectively. We also
set σ̄2
= −108 dBm. It is observed from Fig. 14 that the
SCA-based algorithm proposed in [68] achieves significantly
higher rate as compared to the OFDM system without IRS
and that with random IRS phase shifts and the corresponding
optimal transmit power allocation. Moreover, the strongest

CIR maximization algorithm achieves close performance to the
SCA-based algorithm, thus being an efficient alternative with
lower complexity. Furthermore, we show an achievable rate
upper bound by assuming that (ideally) different IRS reflection
coefficients can be designed for different sub-carriers, thus
making the IRS reflection design “frequency-selective”. It is
observed that this rate upper bound outperforms the SCA-
based solution with the practical frequency-flat (non-selective)
IRS reflection quite substantially, and the rate gap increases
with the number of sub-carriers. This thus reveals that a
fundamental limitation of IRS-aided OFDM systems lies in
the lack of frequency-selective IRS reflection due to its passive
operation.
Finally, it is worth noting that for the more general IRS-
aided MIMO-OFDM systems where the AP and/or user are
equipped with multiple antennas, the IRS reflection design
for rate maximization is more involved, due to the need
of catering to more channels in both space and frequency;
moreover, the IRS reflection needs to be jointly optimized with
multiple transmit covariance matrices at different sub-carriers.
To resolve this problem, [71] proposed an efficient AO-based
algorithm by extending that in the narrow-band MIMO case
and leveraging the convex relaxation technique. The results
in [71] showed that despite the lack of frequency selectivity,
IRS is still effective in improving the rate of MIMO-OFDM
systems with properly designed IRS reflection coefficients over
the conventional system without IRS.
D. IRS-Aided Multi-User System
Next, we consider the general case with multiple users in
the IRS-aided system (see, e.g., Fig. 8). For the purpose of
exposition, we focus on the following two narrow-band multi-
user systems under the SISO and MISO setups, respectively.
In particular, for the SISO case, we compare the performance
of orthogonal multiple access (OMA) such as time division
multiple access (TDMA) and frequency division multiple
access (FDMA) with that of non-orthogonal multiple access
(NOMA); while for the MISO case, we consider the spatial
division multiple access (SDMA) where the multi-antenna
AP serves multiple single-antenna users simultaneously in the
same frequency band. Our main objective is to highlight the
main differences in system design and performance optimiza-
tion with IRS versus traditional systems without IRS.
1) OMA Versus NOMA: In TDMA, the users are served in
orthogonal time slots, thus different IRS reflection coefficients
can be applied over time to maximize the rate of each
user, by exploiting the “time selectivity” of IRS. In con-
trast, in FDMA and NOMA where the users are simulta-
neously served in orthogonal or common frequency bands,
the IRS reflection design needs to cater to the channels of
all users, thus is more challenging compared to the TDMA
case. To investigate which multiple access scheme is more
favorable for IRS-aided systems, [73] considered a two-user
downlink communication system with K = 2, and compared
the average transmit power at the AP required by TDMA,
FDMA, and NOMA for achieving the same rate targets of
the two users. Under this setup, it was shown that TDMA
Fig. 15. Minimum transmit power versus user 1’s rate target for an IRS-aided
two-user system with the sum-rate fixed as 4 bps/Hz.
outperforms FDMA due to the lack of frequency-selective
IRS reflection in the latter case; while, surprisingly, although
NOMA outperforms OMA (i.e., both TDMA and FDMA) due
to successive interference cancellation in traditional multi-user
systems without the IRS, TDMA may perform better than
NOMA for near-IRS users with symmetric rate targets, thanks
to the IRS adaptive reflections based on users’ individual
channels in TDMA. For illustration, we consider an IRS-aided
two-user system where the locations of AP, IRS, and two
users are set as [0, 0] m, [50, 0] m, [48, 1] m, and [48, −1] m,
respectively, in a two-dimensional plane, and show in Fig. 15
the minimum average transmit power required versus the rate
requirement of user 1 with the sum-rate of the two users
being fixed as 4 bps/Hz. Similar to [73], the number of IRS
reflecting elements is set as N = 100, which are divided
into 5 sub-surfaces; the reflection coefficient at each element
is assumed to have unit amplitude and discrete phase shift
with 8 levels; all the channels are modeled by i.i.d. Rayleigh
fading with the path loss exponents of the AP-user, AP-IRS,
and IRS-user links being 3.2, 2.5, and 2.6, respectively. The
average noise power at the user receivers is set as σ2
=
−80 dBm. It is observed from Fig. 15 that both TDMA and
NOMA outperform FDMA in terms of the minimum transmit
power required. Moreover, when the user rates are symmetric,
NOMA requires even larger transmit power than TDMA; while
the performance of NOMA is more robust (less sensitive) to
the users’ rate disparity as compared to that of TDMA.
For the more general multi-user case with K 2, the IRS
reflection designs for FDMA and NOMA become more diffi-
cult, due to the increased number of user channels that need
to be catered to at the same time, which also limits the
passive beamforming gain of IRS for each user. To overcome
this difficulty, [74] proposed a novel dynamic passive beam-
forming scheme for an IRS-aided orthogonal frequency divi-
sion multiple access (OFDMA) system, where each channel
coherence interval is divided into multiple time slots each for
serving only a subset of selected users, and the IRS passive
beamforming is dynamically adjusted over different time slots

to induce artificial (yet properly-tuned) channel fading for
exploiting the IRS’s time-selective reflection as well as multi-
user channel diversity in resource allocation optimization.
By jointly designing the IRS passive beamforming over time
with the OFDMA time-frequency resource allocation, it was
shown in [74] that improved performance can be achieved
over the benchmark scheme with optimized but fixed IRS
reflection coefficients for all users in each channel coherence
interval.
2) SDMA: In SDMA, the multi-antenna AP employs dif-
ferent linear precoding/combining vectors to serve different
users simultaneously in the same frequency band for the
downlink/uplink transmissions, thus significantly improving
the spectral efficiency over the single-antenna AP. However,
conventional SDMA does not perform well for users close
to each other (e.g., in a hot-spot scenario), especially when
they are located at the cell edge. This is because users in this
case are likely to lie in similar directions from their serving
AP, which induces high correlation among their channels
that is detrimental to the achievable spatial multiplexing gain
due to the more severe multi-user co-channel interference.
As such, conventional beamforming techniques such as zero-
forcing (ZF) to null/suppress their mutual interference become
inefficient.
However, by properly deploying IRS in such scenarios,
the above issue can be efficiently solved by leveraging its
spatial interference nulling/cancellation capability [9]. Specif-
ically, IRS can effectively reduce the undesired channel
correlation among these users by optimizing its reflection
coefficients to provide additional controllable signal paths.
Besides, the users near the IRS are expected to be able
to tolerate more interference from the AP as compared to
those farther away from the IRS. This thus provides more
flexibility in designing the transmit beamforming at the AP
for serving other users outside the coverage of any IRS.
As a result, the signal-to-interference-plus-noise ratio (SINR)
performance of all users in the network can be significantly
improved, regardless of whether they are aided directly by any
IRS or not [9].
Despite the above benefits, the active and passive beam-
forming designs in the multi-user case are closely coupled
and usually lead to more complicated optimization problems
as compared to that in the single-user case. As such, AO
has been widely adopted to obtain high-quality suboptimal
solutions, by iteratively optimizing one of the transmit and
reflect beamforming vectors/coefficients with the other being
fixed, until the convergence is reached [9], [75], [76]. However,
such AO-based algorithms may become inefficient as the
number of QoS constraints increases since they are prone to
getting trapped at undesired suboptimal solutions due to the
more stringent coupling among the variables. To address this
issue, a new and more efficient penalty-based algorithm was
proposed in [66] for a more general system setup, which guar-
antees attaining at least a locally optimal solution. Specifically,
the QoS constraints are first decoupled by introducing a set
of auxiliary variables and the reformulated problem is then
solved by jointly applying the penalty-based method and block
coordinate descent (BCD) method [66].
E. Passive Beamforming With Discrete Reflection Amplitude
and Phase Shift
In the preceding subsections, we have mainly considered
continuous phase shifts of IRS with the maximum reflection
amplitude, while as discussed in Section II-B, practical IRSs
have discrete phase shifts and/or discrete reflection ampli-
tudes, which may result in signal misalignment at designated
receivers and thus degraded communication performance as
compared to the ideal case of continuous phase shifts with
flexible reflection amplitudes [54], [64], [76]–[81]. Besides,
such constraints also complicate the IRS reflection design and
render the corresponding optimization problems more difficult
to solve as compared to their continuous counterparts. For
example, exhaustive search over all possible discrete phase-
shift values is needed to solve the design problem optimally,
which, however, incurs prohibitively high complexity in prac-
tice with large values of N [54]. Although some optimization
techniques such as branch-and-bound (BB) method can be
used to obtain the optimal solution to such problems with
reduced complexity on average, the computational complexity
in the worst case is still exponential over N and thus the same
as that of the exhaustive search. In practice, one heuristic
approach is to firstly relax such constraints and solve the
problem with continuous amplitude/phase-shift values, then
quantize the obtained solutions to their nearest values in the
corresponding discrete sets. While this approach is generally
able to reduce the computational complexity significantly to
polynomial orders of N, it may suffer various losses in
performance as compared to the continuous solution due to
quantization/round-off errors, depending on the number of
discrete levels as well as N, and is also generally suboptimal
for the original discrete optimization problem. To further
improve the performance of the above approach, AO can be
applied to iteratively optimize the discrete amplitude/phase-
shift of each element by fixing those of the others at each
time [54]. Other methods that are possible to handle discrete
reflection constraints include penalty-based method [66], [76],
ADMM, etc.
To draw useful insight into the performance loss of employ-
ing discrete phase shifts for IRS as compared to continuous
phase shifts, we consider a modified problem of (P1) by
ignoring the AP-user direct link and replacing the phase-shift
constraints in (19) by their discrete counterparts. As a result,
the user receive power (or achievable rate) maximization
problem with discrete phase shifts can be formulated as
(P1-DP): max
θ
N
n=1
h∗
r,ngnejθn
2
(45)
s.t. γn ∈ F
θ, n = 1, · · · , N, (46)
where F
θ is defined in (15). To facilitate our analysis, we adopt
the quantization approach to solve (P1-DP) by first obtaining
the optimal continuous phase shifts in (21) and then quantizing
them independently to the nearest values in F
θ. The corre-
sponding objective value is denoted by Pr(bθ). In general,
Pr(bθ) serves as a lower bound of the maximum objective
value of (P1), whereas when bθ → ∞, Pr(bθ) converges
to its maximum value in the continuous phase-shift case.

Fig. 16. Illustration of 1-bit phase shift control versus 1-bit reflection amplitude control in the complex signal plane, where the former utilizes both
−
→
S1 and
−
→
S2 while the latter only uses
−
→
S1.
As such, the performance loss of IRS with bθ-bit phase shifters
(or 2bθ
uniform phase-shift levels) per reflecting element as
compared to the ideal case with continuous phase shifts at all
reflecting elements can be characterized by a ratio given by
η(bθ) Pr(bθ)/Pr(∞). Under the same channel assumption
for deriving the asymptotic result given in (24), it was shown
in [64] that
η(bθ) =
2bθ
π
sin
π
2bθ
2
. (47)
It is observed from (47) that when N is sufficiently large,
the power ratio η(bθ) depends only on the number of discrete
phase-shift levels, 2bθ
, but is regardless of N. In other words,
using a practical IRS even with discrete phase shifts can still
achieve the same asymptotic receive power scaling order of
O(N2
) given in (24) as in the case of continuous phase shifts.
As such, the design of IRS hardware and control module can
be greatly simplified by using discrete phase shifters, without
compromising the performance significantly in the large-N
regime. Since η(1) = −3.9 dB while η(2) = −0.9 dB, using
2-bit phase shifters is practically sufficient to achieve close-
to-optimal performance within 1 dB.
In [24], a practical binary/1-bit reflection amplitude control
was considered for IRS where the phase shifts of all the
elements are set to zero and only the reflection amplitude of
each IRS element is optimized as either 0 or 1. Similar to (P1-
DP), the receive power maximization problem in this case can
be formulated as
(P1-DA): max
β
N
n=1
h∗
r,ngnβn
2
(48)
s.t. βn ∈ {0, 1}, n = 1, · · · , N, (49)
where β = [β1, · · · , βN ]T
. To solve (P1-DA), a heuristic
method was proposed in [24] where βn = 1 if the phase of
h∗
r,ngn lies in [−π/2, π/2]; otherwise, βn = 0, n = 1, · · · , N.
It was shown that the user receive power with this scheme is
reduced by a factor of π2
, i.e., incurring about 10 dB loss,
as compared to the case with continuous phase shifts [24].
Fig. 17. Receive SNR versus N for different discrete amplitude/phase-shift
designs of IRS.
To compare the two schemes intuitively, we illustrate them
in the complex signal plane in Fig. 16. Suppose that
−
→
S0
indicates the desired signal to be aligned with, while
−
→
S1 and
−
→
S2 denote the reflected signals by any two IRS elements
without applying any phase shift/amplitude control yet. For
the 1-bit phase shift control shown in Fig. 16 (a) [54], [64],
it is intuitive that
−
→
S1 can help improve the user receive power
by simply setting γ1 = 0 since it has an acute angle with
−
→
S0,
i.e., in the right halfplane. On the other hand, setting γ2 = π
will rotate
−
→
S2 into the right halfplane, which helps improve the
combined signal amplitude as well. In contrast, for the 1-bit
amplitude control shown in Fig. 16 (b) [24], since
−
→
S1 and
−
→
S2
cannot be rotated, the optimal strategy is to turn ON and OFF
their corresponding reflecting elements, by setting β1 = 1 and
β2 = 0. By comparing Figs. 16 (a) and (b), it is observed that
both
−
→
S1 and
−
→
S2 are utilized in 1-bit phase shift control whereas
only
−
→
S1 is used in 1-bit amplitude control, which implies that
the former generally outperforms the latter or equivalently the

Fig. 18. Downlink transmission in IRS-aided multi-cell wireless network.
latter needs to deploy more reflecting elements (i.e., larger
IRS) to achieve the same performance as the former.
Considering the same simulation setup as in Fig. 9, we plot
the maximum receive SNR versus N under different discrete
amplitude/phase-shift cases of IRS in Fig. 17 with Mt = 1,
Pt = 5 mW, and σ2
= −100 dBm. It is observed that
when N is sufficiently large, the SNR loss of using IRS
with 1-bit or 2-bit phase shift control approaches a constant,
i.e., 3.9 dB or 0.9 dB, which is consistent with the theoretical
result in (47). Besides, one can observe that compared with
1-bit phase shift control, 1-bit amplitude control suffers sub-
stantial power/SNR loss since the IRS reflected signals cannot
be fully utilized with the simple ON/OFF control.
F. Other Related Work and Future Direction
In the last subsection, we overview other related works on
IRS reflection design and point out promising directions for
future work.
Although this section focuses on the single-cell system with
one IRS, it is worthy of extending the results to more general
setups with multiple APs and/or multiple IRSs in an IRS-aided
wireless network as shown in Fig. 18. As such, the reflection
coefficients of IRSs need to be jointly designed with the
transmissions of multiple APs to not only improve the desired
signal power but also mitigate the intra-cell interference as
well as inter-cell interference (ICI) [82]–[84]. For the purpose
of illustration, we consider an IRS-aided two-cell system with
one user at the edge of both cells, as shown in Fig. 19, where
the user is served by AP1 but suffers from the inter-cell/co-
channel interference from AP2. Similar to Section III-A,
we consider the narrow-band system with flat-fading channels.
Then, based on the IRS reflection model in (6), the signal
received at the user can be expressed as
y=(hH
r Θg1 +h∗
d,1)

P1x1 +(hH
r Θg2 +h∗
d,2)

P2x2 +z,
(50)
where x1 and x2 are modeled by i.i.d. random variables with
zero mean and unit variance which denote the information
signals sent by AP1 and AP2, respectively, P1 and P2 are
the corresponding transmit powers, and z denotes the AWGN
Fig. 19. Simulation setup for IRS-aided interference suppression.
Fig. 20. Achievable rate versus transmit power at the AP1 and AP2.
at the user receiver modeled as CSCG with zero mean and
variance σ2
. Accordingly, the user receive SINR is expressed
as
γ =
P1|hH
r Θg1 + h∗
d,1|2
P2|hH
r Θg2 + h∗
d,2|2 + σ2
=
P1|vH
h̄r,1 + h∗
d,1|2
P2|vHh̄r,2 + h∗
d,2|2 + σ2
, (51)
where v = [ejθ1
, · · · , ejθN
]H
, h̄r,1 = diag(hH
r )g1 and h̄r,2 =
diag(hH
r )g2.
Our objective is to maximize the user achievable rate (or γ
in (51) equivalently) via optimizing the phase shifts at the IRS.
As shown in Fig. 19, the IRS and the user are located on the
perpendicular bisector of the line segment connecting AP1 and
AP2 with d1 = 100 m, d2 = 5 m, and d3 = 50 m. Both the
AP-IRS and AP-user links are assumed to follow Rayleigh
fading with the path loss exponent of 3.6, whereas the IRS-
user link is assumed to follow Rician fading with Rician
factor of 3 dB and path loss exponent of 2.2. For comparison,
we consider three schemes: 1) Rate maximization where the
phase shifts are optimized to maximize the achievable rate by
applying the fractional programming, followed by Gaussian
Randomization [93]; 2) Signal power maximization where
the phase shifts are optimized to maximize the information
signal power at the user based on (21), i.e., the numerator
in (51); and 3) Without IRS. It is worth noting that the rate
maximization scheme requires the knowledge of the CSI for
the desired signal links, i.e., AP1-user and AP1-IRS-user links,

TABLE III
A SUMMARY OF REPRESENTATIVE WORKS ON IRS REFLECTION OPTIMIZATION
as well as the interference links, i.e., AP2-user and AP2-IRS-
user links; whereas the signal power maximization scheme
needs only the knowledge of the former CSI. In Fig. 20,
we plot the user achievable rate versus the transmit power
at the AP1(AP2) with P1 = P2 and N = 100. It is observed
from Fig. 20 that deploying the IRS at the cell-edge of two
neighbouring APs is able to significantly improve the user
achievable rate. Furthermore, the performance gain of the
rate maximization scheme over the signal power maximization
scheme becomes larger as the transmit power increases. This is
expected since with increasing transmit power, the interference
from AP2 becomes more dominant as compared to the receiver
noise and thus it needs to be effectively suppressed to sustain
the user’s rate increase.
In Table III, we summarize the representative works on
IRS reflection optimization based on their considered system
setups, adopted optimization techniques, etc. Furthermore,
besides discrete reflection amplitudes/phase shifts, the perfor-
mance of IRS-aided systems under other hardware constraints,
such as coupled reflection amplitude and phase shift shown
in Section II-B, is also worth investigating. In particular,
it was shown in [56] that the asymptotic power scaling
order, i.e., O(N2
) unveiled in [9] under the ideal phase
shift model, still holds for the practical case with phase-
shift dependent non-uniform IRS reflection amplitude. On
the other hand, from the optimization perspective, it is also
important to develop more advanced and computationally
efficient algorithms such as machine learning based methods
for IRS reflection design, especially for practically large IRSs
[69], [70], [94], [95].
Besides SINR/rate maximization, another line of research
has aimed to study other performance metrics of IRS-aided
systems such as outage probability and average bit error rate
(BER) [96], [97]. In [62], random matrix theory was leveraged
to study the asymptotic max-min SINR in the single-cell MISO
downlink system, which characterizes the effect of channel
large-scale fading parameters on the SINR performance. More-
over, IRS reflection design and spatial throughput analysis for
large hybrid active and passive wireless networks is also an
interesting topic to pursue [12], [98].
Last but not the least, while significant performance gains
offered by IRS have been shown in this section, they are based
on the assumption of perfect CSI for all channels considered.
In practice, the acquisition of accurate CSI of the IRS-reflected
links is crucial, which, however, is a difficult task due to
the passive operation of IRS as well as its large number
of reflecting elements. Thus, how to efficiently estimate the
channels in IRS-aided wireless systems and optimize the IRS
reflection with imperfect/partial CSI (see, e.g., [75], [76], [89],
[99]) is practically important, as will be addressed in the next
section.
IV. IRS CHANNEL ESTIMATION
A. Problem Description and Challenges
To fully achieve various performance gains brought by IRS,
the acquisition of accurate CSI is crucial, which, however,
is practically challenging. To be specific, considering the IRS-
aided uplink multi-user MIMO communication in a narrow-
band system over flat-fading channels, the received signal at

Fig. 21. Two practical IRS configurations and their respective transmission protocols.
the MB-antenna BS from K users (each of which is equipped
with Mu antennas) can be expressed as
y =
K
k=1

GT
ΘHr,k + Hd,k

xk + z, (52)
where G ∈ CN×MB
, Hr,k ∈ CN×Mu
and Hd,k ∈
CMB×Mu
denote the IRS-BS, user k-IRS, and user k-BS
direct channels, respectively, with k = 1, . . . , K; Θ =
diag

ejθ1
, ejθ2
, . . . , ejθN

represents the diagonal phase-shift
matrix of one or more IRSs comprising N reflecting elements
in total, with the reflection amplitude of each element set to
one or its maximum value for simplicity; xk ∈ CMu×1
is
the transmit signal of user k; and z ∈ CMB×1
is the AWGN
vector at the BS. Accordingly, the uplink CSI includes G,
{Hr,k}
K
k=1, and {Hd,k}
K
k=1, and thus the total number of
uplink channel coefficients consists of two parts:
• The number of channel coefficients (equal to K×NMu+
MBN) for the links to/from the (equivalent single) IRS
(i.e., {Hr,k}
K
k=1 and G), which are newly introduced due
to the employment of IRS;
• The number of channel coefficients (equal to K ×
MBMu) for the direct links (i.e., {Hd,k}
K
k=1), which
exist in conventional communication systems without
IRS.
Note that the total number of channel coefficients may be
different for TDD and frequency-division duplexing (FDD)
systems. In particular, the FDD system requires to estimate
twice the number of channel coefficients in (52) due to the
generally non-symmetric uplink and downlink channels; while
in contrast, the TDD system may only need to acquire either
the uplink or downlink channel coefficients by exploiting the
uplink-downlink channel reciprocity. Furthermore, in broad-
band communication systems over frequency-selective fading
channels, more channel coefficients are induced for both user-
BS direct channels as well as user-IRS-BS reflected channels
due to the multi-path delay spread and the resultant convolu-
tion of time-domain impulse responses of user-IRS and IRS-
BS multi-path channels, which makes the channel acquisition
problem even more challenging [68], [72], [100].
Besides the substantially more IRS-induced channel coeffi-
cients as compared to the conventional system without IRS,
another challenge in IRS channel estimation arises from its
low-cost reflecting elements that do not possess any active
RF chains and thus cannot transmit pilot/training signals to
facilitate channel estimation, which is in sharp contrast to the
active BSs/user terminals in conventional wireless systems.
In the existing literature, there are two main approaches for
IRS channel estimation based on two different IRS config-
urations, depending on whether it is mounted with sensing
devices (receive RF chains) or not, termed as semi-passive
IRS and (fully) passive IRS, respectively, as shown in Fig. 21.
In the following, we present these two IRS configurations
in detail, discuss the state-of-the-art results on IRS channel
estimation based on them, respectively, and finally highlight
the remaining important issues that need to be tackled in future
work.
B. Semi-Passive IRS Channel Estimation
To endow the IRS with sensing capability for channel
estimation, additional sensing devices (such as low-power
sensors) need to be integrated into IRS, e.g., interlaced with
IRS reflecting elements, as shown in Fig. 21 (a), each equipped
with a low-cost receive RF chain (e.g., low-resolution analog-
to-digital converter (ADC)) for processing the sensed signal.
As such, the semi-passive IRS generally operates in one of the
following two modes alternately over time:
• Channel sensing mode: With all the reflecting elements
turned OFF, the sensors are activated to receive the pilot
signals from the BS/users in the downlink/uplink for
estimating their respective channels to IRS;

• Reflection mode: With the sensors deactivated, the IRS
reflecting elements are turned ON to reflect the data sig-
nals from the BS/users for enhancing the downlink/uplink
communication, respectively.
Accordingly, a general transmission protocol for semi-
passive IRS is illustrated in Fig. 21 (a), where each channel
coherence interval is divided into three phases. In the first
phase, the BS/users send their pilot signals in the down-
link/uplink to estimate the BS-user direct channels as in the
conventional wireless system without IRS, while the IRS
operates in the channel sensing mode to estimate the CSI
from the BS/users based on the signals received by its sensors.
After that, in the second phase, the CSI is exchanged between
the IRS controller and BS, based on which the active and
passive beamforming coefficients are jointly designed at the
IRS controller or BS and then sent to the other via the separate
wired/wireless backhaul link between them. Finally, in the
third phase, the IRS switches to the reflection mode to assist
data transmission between the BS and users with the designed
active/passive beamforming coefficients set at the BS/IRS. It
is worth pointing out that for semi-passive IRS, only the
downlink/uplink CSI of BS/users→IRS links can be estimated
by the IRS sensors; whereas that of their corresponding reverse
links can be obtained only in the TDD system by leveraging
the channel reciprocity. However, such CSI is unavailable in
the FDD system, thus making channel estimation in FDD
systems infeasible for semi-passive IRS.
As shown in Fig. 21 (a), Ḡ ∈ CNs×MB
and H̄r,k ∈
CNs×Mu
denote the channels from the BS and user k to the
Ns IRS sensors, respectively, with Ns N in general for
reducing the IRS cost and energy consumption. It should be
noted that the channels from the BS/users to the IRS sensors
are not identical to those from them to the IRS reflecting
elements given in (52), while they are usually correlated due
to close proximity. As such, the essential challenge for semi-
passive IRS channel estimation is how to construct the high-
dimensional channels G and Hr,k in (52) from the estimated
CSI on the low-dimensional channels Ḡ and H̄r,k. To resolve
this problem, advanced signal processing tools, such as com-
pressed sensing, data interpolation, and machine learning, can
be applied to construct the CSI of BS/users→IRS links from
the estimated CSI via the IRS sensors by exploiting their inher-
ent spatial correlation. Moreover, it is worth noting that the
channel estimation accuracy for semi-passive IRS is generally
limited by the number of available sensors, their finite (e.g., 1-
bit) ADC resolution, and the channel sensing (downlink/uplink
training) time. Intuitively, installing more sensors provides
more channel-sensing measurements for reducing the IRS CSI
construction error in general, applying higher-resolution ADCs
can reduce the quantization error, and increasing the channel
sensing time can help average out the sensing noise more
effectively. Finally, it is worth pointing out that since the AP
and IRS are at fixed locations once deployed, the AP-IRS link
typically has a much longer channel coherence time than the
IRS-user link (due to user mobility) and thus can be considered
as quasi-static. Moreover, as the AP and IRS are typically at
high altitudes with few scatterers between them, the AP-IRS
link can be efficiently estimated by exploiting some channel
properties such as low-rank, sparsity, spatial correlation, etc.
As a result, it is less challenging to acquire the CSI of the
AP-IRS link, as compared to that of the IRS-user link.
However, a systematic study on the fundamental limits,
practical algorithms, and their cost-performance trade-offs for
semi-passive IRS channel estimation is still lacking in the lit-
erature, although a handful of preliminary works [101]–[105]
on addressing some of these aspects have recently appeared.
Specifically, an AO approach using random spatial sampling
and analog combining techniques was proposed in [101] for
semi-passive IRS channel estimation under the narrow-band
beamspace-based channel model, where its mean-squared error
(MSE) performance was evaluated with respect to the chan-
nel sensing time. In [102]–[104], semi-passive IRS channel
estimation algorithms based on compressed sensing and deep
learning techniques were proposed for the IRS-aided SISO
system, where the impact of both the number of IRS sensors
and channel sensing time on the achievable rate was evaluated
by simulations. In [105], the direction-of-angle estimation
problem was investigated in the IRS-aided M-MIMO system,
for which a deep neural network-based method was proposed
to effectively reduce the quantization error induced by the low-
resolution ADCs.
C. Passive IRS Channel Estimation
When there are no sensors mounted on the IRS for low-
cost implementation, IRS becomes fully passive and thus it is
generally infeasible to acquire the CSI between the IRS and
BS/users directly. In this (perhaps more practical yet challeng-
ing) case, an alternative approach is to estimate the cascaded
user-IRS-BS channels at the BS/users in the uplink/downlink,
respectively. Note that this approach, unlike the case of semi-
passive IRS, applies to both TDD and FDD systems, although
in the TDD case, only the uplink or downlink cascaded
channels need to be estimated thanks to the uplink-downlink
channel reciprocity. Without loss of generality, we consider
the IRS-aided uplink multi-user MIMO system given in (52),
where the cascaded user-IRS-BS channel is the transposed
Khatri-Rao product of the user-IRS and IRS-BS channels, with
each channel coefficient given by [G]n,mB
[Hr,k]n,mu
, ∀n =
1, . . . , N, mB = 1, . . . , MB, mu = 1, . . . , Mu, and k =
1, . . . , K. As such, the total number of channel coefficients
for the cascaded user-IRS-BS links is K × NMBMu, which
is in general (much) larger than that of channel coefficients
for the separate user-IRS and IRS-BS links (equal to K ×
NMu + MBN). This indicates that there is an inherent
redundancy in the cascaded channels, which is due to the fact
that all users’ cascaded channels involve the common IRS-
BS channel GT
, which, however, is difficult to be resolved
from any user’s estimated cascaded channel. On the other
hand, it is worth noting that from the perspective of designing
the joint passive/active beamforming, the CSI on the cascaded
links is generally sufficient and without loss of optimality as
compared to that on their corresponding links, as shown in
Section III. Moreover, it is worth pointing out that different
from the semi-passive IRS case where the IRS channels need

to be constructed from the estimated CSI via IRS sensors,
the cascaded user-IRS-BS channels in the case of passive IRS
can be directly estimated at the BS/users without the need of
sophisticated channel reconstruction.
In Fig. 21 (b), we illustrate the transmission protocol for the
uplink with the passive IRS, where each channel coherence
interval is divided into three consecutive phases. First, users
transmit orthogonal pilots to the BS and meanwhile the IRS
varies its reflection coefficients according to a pre-designed
reflection pattern, based on which the BS estimates both the
user-BS direct channels and the cascaded user-IRS-BS chan-
nels. Second, based on the estimated CSI, the IRS reflection
coefficients for data transmission are designed at the BS jointly
with its receive beamforming and then sent to the IRS con-
troller through the backhaul link. Third, the IRS controller sets
the reflection coefficients accordingly for assisting independent
data transmissions from the users to BS. Note that for TDD
system, the estimated uplink CSI can also be used to design
IRS reflection for data transmissions in the downlink from the
BS to users; while for FDD system, the transmission protocol
in Fig. 21 (b) is still applicable for the downlink, with the only
modification that the roles of the BS and users are swapped in
the first phase, i.e., the direct/cascaded channels are estimated
at the users based on the pilot signals sent by the BS, then
the users need to feed back their estimated CSI to the BS
for the joint optimization of IRS reflection and BS transmit
beamforming.
For the (uplink) passive IRS channel estimation, the key
problem is the joint design of the pilot sequences, IRS reflec-
tion pattern, and signal processing algorithm at the receiver to
accurately estimate both the direct user-BS and cascaded user-
IRS-BS channels with minimum training overhead (number of
pilot symbols). This problem has not been addressed before in
channel estimation for conventional wireless systems without
IRS and thus is new and also non-trivial to solve in general.
For the purpose of exposition, we consider first the simple
IRS-aided single-user system (i.e., K = 1) with flat-fading
channels, where both the BS and user are equipped with one
single antenna (i.e., the SISO case with MB = Mu = 1). In
this case, one practical method for IRS channel estimation is
by employing an ON/OFF-based IRS reflection pattern [106],
i.e., each one of the IRS elements is turned ON sequentially
with the others set OFF at each time, thereby the user-BS
direct channel and the cascaded channels associated with
different IRS elements are estimated separately. Note that this
method requires at least N +1 pilot symbols for estimating the
total N + 1 channel coefficients in this system. Albeit being
simple to implement, the ON/OFF-based IRS reflection pattern
incurs substantial reflection power loss as only one element is
switched ON at each time and thereby the reflected signal is
rather weak. To overcome this power loss issue and improve
the channel estimation accuracy, the all-ON IRS reflection
pattern can be employed with orthogonal reflection coefficients
over time, such as those drawn from different columns of an
(N + 1) × (N + 1) DFT matrix [72], [107], whereby all the
N + 1 channel coefficients can be estimated over N + 1 pilot
symbol durations. Moreover, to reduce the training overhead
for IRS with practically large number of reflecting elements as
well as simplify IRS reflection design for data transmission,
an efficient approach is to group adjacent IRS elements (over
which the IRS channels are usually spatially correlated) into
a sub-surface [68], [72], referred to as IRS element grouping;
as a result, only the effective cascaded user-IRS-BS channel
associated with each sub-surface needs to be estimated, thus
greatly reducing the training overhead. In addition, based on
the estimated effective cascaded channels for the sub-surfaces,
their reflection coefficients (each set identical for all the IRS
elements in the same sub-surface) can be optimized more
efficiently for data transmission. Thus, the element grouping
strategy provides a flexible trade-off between training/design
overhead/complexity and IRS passive beamforming gain in
practice [68], [72]. Moreover, to further improve the training
efficiency for the cascaded user-IRS-BS channel, a fast channel
estimation scheme based on the novel concept of intra-symbol
IRS reflection variation was proposed in [108].
For the passive IRS-aided single-user MIMO/MISO system
(i.e., MB 1 and Mu ≥ 1), it is required to estimate more
channel coefficients due to the increased channel dimensions
and the matrix multiplication of the user-IRS and IRS-BS
channels in the cascaded user-IRS-BS channel. By adopting
the orthogonal pilots over different transmit antennas at the
BS/user for the downlink/uplink training, the total training
overhead is increased by MB or Mu times as compared to the
single-user SISO case, which can be practically prohibitive if
the number of transmit antennas MB or Mu is large. To tackle
this difficulty, certain IRS channel properties (such as low-
rank, sparsity, and spatial correlation) can be exploited to facil-
itate the cascaded channel decomposition as well as reduce the
training overhead [109]–[113].7
Furthermore, to accelerate the
training process, deep learning and hierarchical searching algo-
rithms have also been developed for the channel estimation in
passive IRS-aided MIMO/MISO systems [119]–[123].
For channel estimation with passive IRS serving multiple
users (i.e., K 1), a straightforward method is by adopting
the single-user channel estimation design to estimate the chan-
nels of K users separately over consecutive time [124], which,
however, increases the total training overhead by K times
as compared to the single-user case and thus is practically
prohibitive if K is very large. Recall that all the users in
fact share the same (common) IRS-BS channel GT
in (52) in
their respective cascaded user-IRS-BS channels. By exploiting
this fact, the training overhead for IRS channel estimation in
the multi-user case can be significantly reduced. For exam-
ple, a user can be selected as the reference user of which
the cascaded channel is first estimated. Then, based on this
reference CSI, the cascaded channels of the remaining K − 1
users can be efficiently estimated by exploiting the fact that
these cascaded channels are scaled versions of the reference
user’s cascaded channel and thus only the low-dimensional
scaling factors, rather than the whole high-dimensional cas-
caded channels, need to be estimated. In particular, for the
7The cascaded channel decomposition exploiting certain IRS channel
properties can also be readily applied to the multi-user case by treating
multiple users as one equivalent user with KMu transmit antennas in the
uplink or estimating the channels of different users in parallel in the downlink
[114]–[118].

IRS-aided multi-user MISO system (i.e., Mu = 1), it was
shown in [125] that the minimum uplink training overhead is
K + N + max

K − 1,

(K−1)N
MB

, which decreases with an
increasing MB. Note that this result exploits the redundancy
of receive antennas at the BS for training overhead reduction
and thus is in sharp contrast to the conventional multi-
user channel estimation without IRS for which the minimum
training overhead is independent of the number of receive
antennas at the BS. In [115], [126], the static BS-IRS channel
which is common to all users was estimated first to reduce the
training time for their cascaded channel estimation. Moreover,
it was shown in [126] that the sign uncertainty in the BS-
IRS channel estimation does not affect the uniqueness of the
cascaded channel recovery.
For broadband systems with frequency-selective fading
channels, it is necessary to estimate more channel coeffi-
cients due to the multi-path delay spread and the resultant
convolution of the user-IRS and IRS-BS multi-path channels
in each cascaded user-IRS-BS channel. Moreover, although
the channels are frequency-selective, the IRS reflection coef-
ficients are frequency-flat (as shown in Section III), which
thus cannot be flexibly designed for different frequencies
(e.g., different sub-carriers in OFDM-based systems). Due to
the above reasons, the IRS channel estimation for broadband
frequency-selective fading channels is even more challenging
as compared to that for narrow-band flat-fading channels,
under all the SISO/MISO/MIMO as well as single-user/multi-
user setups. Fortunately, since the number of OFDM sub-
carriers is typically much larger than the maximum num-
ber of delayed paths in practical systems, there exists great
redundancy for channel estimation, which can be exploited
for designing OFDM-based pilot symbols to efficiently esti-
mate the channels of multiple users at the same time [100].
In addition, the previously discussed techniques for narrow-
band IRS channel estimation such as DFT-based IRS reflection
pattern, IRS element grouping, compressed sensing, IRS-BS
common channel exploitation, etc., can be jointly applied with
customized OFDM-based pilots to achieve efficient broadband
IRS channel estimation. For example, by exploiting the com-
mon IRS-BS channel and LoS dominant user-IRS channels,
it was shown in [100] that only N + 1 OFDM symbols
are required for estimating the cascaded channels of up to
!
(N+1)(Q−L)
N+L

+ 1 users at the same time, with Q and L
denoting the number of OFDM sub-carriers and the maximum
number of delayed paths, respectively. Note that different
from [125] that exploits the redundancy of receive antennas
at the BS for training overhead reduction, [100] exploits the
redundancy of OFDM sub-carriers to support more users for
concurrent channel estimation and thus improve the training
efficiency, which is applicable to all the SISO/MISO/MIMO
setups with training overhead independent of the number of
receive antennas at the BS. Therefore, it is also possible to
fully exploit the redundancy of both receive antennas and
OFDM sub-carriers to further improve the training efficiency
for the general IRS-aided broadband multi-user MIMO system.
Most of the existing works (e.g., [107], [109]–[113]) on
passive IRS channel estimation have assumed continuous
Fig. 22. Achievable rate versus IRS element grouping ratio ρ = N̄/N with
N = 12 × 12 = 144, SNR = 5 dB, and channel coherence time Tc = 150
channel uses.
phase shifts of IRS reflecting elements. However, as shown
in Section II, the number of phase-shift levels in practice
needs to be finite considering the hardware implementation.
This thus gives rise to another issue on how to design
orthogonal/near-orthogonal IRS reflection coefficients over
time with discrete phase-shift levels, which is not an issue
in the case of continuous phase shifts with the DFT-based
orthogonal training reflection design. One efficient way for
tackling this problem is to construct near-orthogonal reflection
coefficients using proper quantization techniques [127], [128].
However, the resultant channel estimation errors are generally
higher than the case with orthogonal reflection coefficients,
and also correlated over the reflecting elements. As a result,
their induced interference is correlated in general during data
transmission, which needs to be properly taken into account
in the design of IRS passive/reflect beamforming for data
transmission [127], [128]. Moreover, existing IRS channel
estimation works have mostly focused on the design of “all-
at-once channel estimation, i.e., all the cascaded channel
coefficients of all IRS reflecting elements are estimated at
once. However, this requires long training time which may
not be practically available in some scenarios such as short-
packet transmissions with insufficient pilot length and/or fast
fading channels with short coherence time. Thus, it is desirable
to develop flexible training protocols with adjustable pilot
durations to adapt to different transmission requirement and
channel coherence time. In [127], a novel hierarchical training
reflection pattern was proposed to progressively resolve the
cascaded CSI of the IRS reflecting elements based on the
element grouping concept, which also leads to successively
refined passive beamforming designs for data transmission that
effectively enhance the system rate performance in a block-
by-block manner.
It is worth noting that for IRS channel estimation under
different setups, there exists a general trade-off between
the training overhead for channel estimation and passive

TABLE IV
A SUMMARY OF REPRESENTATIVE WORKS ON IRS CHANNEL ESTIMATION
beamforming gain for data transmission. Specifically, too little
training leads to inaccurate/insufficient CSI which results in
inefficient reflection design and thus degrades the passive
beamforming gain for data transmission, while too much
training renders less time for data transmission, both causing
reduced achievable rate. To show this fundamental trade-off in
IRS channel estimation, Fig. 22 plots the achievable rate versus
the IRS element grouping ratio ρ = N̄/N with N̄ denoting
the number of IRS sub-surfaces (which is proportional to the
training overhead) for a passive IRS-aided single-user narrow-
band system with different channel estimation schemes. It is
observed that given a channel coherence time, the achievable
rate with each scheme first increases with the element grouping
ratio, due to the enhanced passive beamforming gain as a
result of more accurate CSI, while it then decreases as the
element grouping ratio exceeds a certain value, due to the
reduced time for data transmission. Moreover, compared to
the optimal DFT-based orthogonal IRS reflection pattern with
continuously adjustable phase shift at each reflecting element,
some performance loss is observed for the IRS reflection
pattern design under the practical constraint of discrete phase
shifts with Kθ uniformly quantized levels in [0, 2π) [127],
[128]. On the other hand, both DFT-based reflection patterns
(with continuous/discrete phase shifts) achieve much larger
rates than the ON/OFF-based reflection pattern that does
not fully utilize the large aperture of IRS as well as the
heuristic random-phase-shift reflection pattern where the IRS
independently generates multiple sets of reflection coefficients
and the set that achieves the largest effective channel gain is
selected for data transmission.
D. Other Related Work and Future Direction
In Table IV, we summarize the representative works on IRS
channel estimation according to their considered IRS config-
urations and system setups. It is noted that there has been
very limited work on semi-passive IRS channel estimation,
and it is still unclear how this approach is compared with
the passive IRS channel estimation in terms of performance
and cost. In addition, most of the existing works consider
narrow-band channel estimation, while channel estimation for
the more prevalent broadband communication requires more
investigation. Furthermore, most of the works assume TDD
systems and/or continuous phase shifts, while channel estima-
tion for FDD systems (e.g., downlink channel training with
passive IRS) and with discrete phase shifts is also practically
important as well as challenging to design, which deserves
further studies in the future. With the CSI of IRS channels
estimated explicitly, there are generally two approaches for
designing the joint active and passive beamforming for data
transmission: one is by optimizing the joint beamforming
based on the (estimated) CSI directly using the techniques
reviewed in Section III; and the other is via the robust
beamforming design with the channel estimation error taken
into account, which has been recently studied in e.g., [128],
[130]–[133] based on various CSI error models and deserves
further investigation.
Besides the CSI-based (robust) beamforming design, there
are other alternative beamforming design methods that do not
require the CSI explicitly. Among others, beam training is
an efficient codebook-based design that directly searches for
the best beam in the spatial domain that yields the optimum

performance [121], [134]. However, applying the conventional
single-beam training method to IRS will incur prohibitively
high training overhead in the IRS-aided communication system
due to the practically large number of IRS reflecting elements.
To address this issue, a novel multi-beam beam training
method was recently proposed in [135] to substantially reduce
the training overhead by properly designing the multi-beam
directions over time together with efficient beam identification
at the users. This fast beam training method can be combined
with the IRS element grouping [68] to further reduce the
training overhead. In addition, machine learning techniques
can be used for predicting the optimal passive beamforming
for users without the need of their explicit CSI [123], [136]–
[138]. For instance, with a fingerprinting database at the
IRS that includes several indoor reference positions and their
corresponding optimal passive beamforming, a deep neural
network (DNN) based method was proposed in [136] to predict
the optimal passive beamforming for indoor users based on
their respective position information only. The study on the
above beam training and machine learning based methods for
the IRS-aided communication system is still at an initial stage
and thus deserves more investigation in the future.
V. IRS DEPLOYMENT
In the preceding sections, we have shown the effectiveness
of IRS in serving as a local hub to improve the communication
performance of nearby users via optimized signal reflection,
where the IRS is assumed to be deployed at a given location.
However, with a total number of reflecting elements, there are
various IRS deployment strategies by placing these elements
at different locations, e.g., near the BS/AP/users, or dividing
them into smaller-size IRSs that are deployed in the network
in a distributed manner. Note that the deployment strategy
for IRS reflecting elements has a significant impact on the
realizations/distributions of all IRS-reflected channels in an
IRS-aided system and hence its fundamental performance
limit. On the other hand, from the implementation perspective,
IRS deployment also needs to take into account various
practical factors such as deployment/operational cost, user
demand/distribution, space constraint, as well as the propa-
gation environment.
It is worth noting that the deployment strategy for IRSs is
generally different from that for active communication nodes
such as BSs/APs or relays, which has been thoroughly inves-
tigated in the literature (see, e.g., [139]–[142]), due to the fol-
lowing reasons. Firstly, since IRS-reflected channels suffer the
severe product-distance/double path-loss due to IRS’s passive
reflection without signal amplification/regeneration, IRS needs
to be placed in close vicinity to the transmitter or receiver to
minimize the path-loss. This is in sharp contrast to the deploy-
ment of active BSs/APs that need to be geographically well
separated for coverage maximization [139], [140], or active
relays with signal processing and amplification capabilities
that are usually deployed in the middle of the transmitter and
receiver for balancing the SNRs of the two-hop links [141],
[142]. Secondly, thanks to the significantly lower cost of IRSs
as compared to active BSs/APs/relays, they can be much more
Fig. 23. IRS deployment in a point-to-point communication system.
densely deployed in the network so as to effectively alter the
signal propagation in the network. However, this results in
a much larger-scale deployment optimization problem with
drastically increased complexity to solve. Thirdly, one chal-
lenging issue in deploying active communication nodes arises
from their mutual interference, which greatly complicates the
deployment optimization problem. In contrast, since IRSs are
passive, their reflected signals decay in power rapidly over dis-
tance; as a result, as long as IRSs are deployed sufficiently far
apart from each other, their mutual interference is practically
negligible, which greatly simplifies their deployment design.
Motived by the above, in this section, we study the new IRS
deployment problem to draw essential and useful insights for
practical design. For the purpose of exposition, we consider
the basic single-/multi-user communication system setups and
focus on characterizing their maximum achievable rates (or
capacity/capacity region) by optimizing their corresponding
IRS deployment. In particular, we divide our discussion
into two subsections, which address the link-level/single-
user design and network-level/multi-user design, respectively.
Moreover, we discuss related works on IRS deployment under
other/more general setups, as well as promising directions for
future work.
A. IRS Deployment Optimization at the Link Level
First, we consider a point-to-point communication link
aided by IRS and discuss three key aspects of IRS deploy-
ment for enhancing its performance at the link level in this
subsection. For the purpose of exposition, we consider the
deployment for a total of N IRS reflecting elements to assist
the communication from a single user to an AP, as illustrated
in Fig. 23.8
8Note that the discussions in this subsection are also applicable to the
communication from the AP to user, as well as that between the AP and
a group of users located close to each other, where the IRS deployment
influences the channels of all users in the same cluster similarly.

1) Optimal Deployment of Single IRS: To start with,
we consider the simplest case where all the N reflecting
elements form one single IRS, as illustrated in Fig. 23 (a).
Moreover, we assume that both the AP and user are equipped
with a single antenna, thus the achievable rate is solely deter-
mined by the receive SNR at the user. To focus our analysis
on the effect of IRS location, we consider the scenario where
the direct user-AP link is blocked, and the other two links
to/from the IRS follow the free-space LoS channel model.
Note that despite the O(N2
) passive beamforming gain
brought by a single IRS as shown in Section III, the received
signal power at the user suffers the double path-loss propor-
tional to the product of the distances of the user-IRS and IRS-
AP links. To illustrate this effect, we consider a simplified 2D
system setup shown in Fig. 23 (a) where the AP and user are
located on a line with horizontal distance D m, and the IRS
can be flexibly deployed on a line above the AP and user by
H m with H D. In this case, the receive SNR with the
optimal IRS passive beamforming (see Section III-A) is given
by
ρS =
Pβ2
0 N2
(d2 + H2)((D − d)2 + H2)σ2
, (53)
where P denotes the transmit power; β0 denotes the path-
loss at reference distance of 1 m; d denotes the horizon-
tal distance between the IRS and user; and σ2
denotes
the average receiver noise power. Note that the user-IRS
distance
√
d2 + H2 increases while the IRS-AP distance

(D − d)2 + H2 decreases as d increases from 0 to D, which
suggests that when H is much smaller than D, the IRS should
be placed in the vicinity of the AP or user to maximize the
SNR. Specifically, we have the following proposition.
Proposition 1: The optimal IRS-user horizontal distance
for maximizing the receive SNR in (53) is given by d
1 =
D−
√
D2−4H2
2 and d
2 = D+
√
D2−4H2
2 if H D/2, and
d
0 = D/2 if H ≥ D/2.
Proof: Note that maximizing ρS is equivalent to min-
imizing f(d)
Δ
= (H2
+ d2
)(H2
+ (D − d)2
), which is a
quartic function over d. The first-order derivative of f(d)
is ∂f(d)
∂d = 2(2d3
− 3Dd2
+ (2H2
+ D2
)d − H2
D) =
4(d − D/2)(d2
− Dd + H2
). If H ≥ D/2, there is only
one solution for ∂f(d)
∂d = 0, i.e., d0 = D/2, which minimizes
f(d). If H D/2, the three solutions for ∂f(d)
∂d = 0 are
d0 = D/2, which maximizes f(d); and d1 = D−
√
D2−4H2
2 ,
d2 = D+
√
D2−4H2
2 , which minimize f(d). This completes the
proof of Proposition 1.
Note that in practice, the AP-IRS and IRS-user vertical
distance H is usually very small (much smaller than D/2)
so as to minimize the path loss of the reflected channel.
Particularly, by denoting α = H/D, we have lim
α→0
d
1 = 0 and
lim
α→0
d
2 = D; namely, as α → 0, or H D, d = 0 and d = D
are both asymptotically optimal, which suggests that the IRS
should be placed right above the user (d = 0) or the AP (d =
D). The asymptotically maximum receive SNR with H D
is thus given by lim
α→0
ρ
S = lim
α→0
P β2
0 N2
H2(D2+H2)σ2 =
P β2
0 N2
H2D2σ2 .
For illustration, we show the receive SNR ρS versus the IRS-
user horizontal distance d in Fig. 24, under a setup with N =
Fig. 24. Receive SNR versus IRS-user horizontal distance.
300, D = 500 m, H = 1 m, P = 30 dBm, σ2
= −90 dBm,
and β0 = −30 dB. It is observed that placing the IRS near
the user/AP yields the largest SNR; while placing it around
the middle between the user and AP (usually optimal in the
case of an active relay instead of IRS) leads to the smallest
SNR, which is consistent with the results in Proposition 1.
2) Single IRS Versus Multiple Cooperative IRSs: With a
given number of IRS reflecting elements, forming them as
one single IRS as discussed above is not the only strategy.
Generally speaking, the IRS reflecting elements can form
multiple smaller-size IRSs (e.g., as illustrated in Fig. 23 (b)
for the case of two IRSs), which, however, has both pros
and cons as compared to the single-IRS deployment. On one
hand, the increased number of IRSs in the user-AP link results
in more inter-IRS reflections, thus causing even higher path-
loss; while on the other hand, multiple IRSs open up the
opportunity of harvesting larger multiplicative beamforming
gains by performing cooperative passive beamforming over
them. Hence, it is not straightforward to conclude whether
splitting the reflecting elements into multiple cooperative
IRSs is advantageous over combining them as a single large
IRS or not.
To answer this question, [143] made the initial attempt
by considering the scenario where two IRSs each with N/2
reflecting elements are deployed right above the user and the
AP, respectively, which is illustrated in Fig. 23 (b), under the
same AP/user setup as in Fig. 23 (a); moreover, all the links
are assumed to be blocked except for the user-IRS 1-IRS 2-
AP link, with all the involved channels following the free-
space LoS model. It was shown in [143] that as long as the
distance between IRS 1 and IRS 2 satisfies a far-field condition
such that the inter-IRS channel is of rank one, a passive
beamforming gain of order O((N/2)4
) can be achieved by
properly aligning the passive beamforming directions of the
two IRSs. As a result, the receive SNR at the AP is given by
[143]
ρD =
Pβ3
0N4
16H4D2σ2
. (54)
Based on (54), it can be shown that the receive SNR with two
cooperative IRSs is larger than that with a single IRS (i.e.,
ρD ρ
S) if the total number of reflecting elements is larger

than a threshold given by
N
4H
√
β0
. (55)
This result is expected since the beamforming gain by two
cooperative IRSs increases with N more significantly as
compared to that of one single IRS (i.e., in the order of
O((N/2)4
) versus O(N2
)), while the additional path-loss in
the two-IRS case is fixed with given H. For comparison,
we also show in Fig. 24 the receive SNR with two cooperative
IRSs deployed as in Fig. 23 (b). It is observed that the two-
IRS case achieves substantial SNR gain as compared to the
maximum SNR in the single-IRS case. Moreover, according
to (54), this gain is anticipated to be more pronounced as N
increases, which makes it more appealing for practical systems
with large N. Thus, it is worth exploiting the partition of the
available reflecting elements into multiple cooperative IRSs
for practical deployment.
The work of [143] was later extended to more practical
and general scenarios of Rician fading channel and multi-
antenna/user systems in [131] and [144], respectively. In [131],
it was shown that with the channel estimation error and train-
ing overhead taken into account, deploying two cooperative
IRSs can still achieve rate improvement over the conventional
case of single IRS deployment under the singe-antenna/user
setup. Furthermore, under the setup with multiple antennas
at the AP and a single antenna at the user(s), the supe-
rior performance of the double-IRS cooperative system to
the conventional single-IRS system was analytically shown
under arbitrary channels in terms of the maximum SNR
and multi-user effective channel rank for single- and multi-
user systems, respectively [144]. It is worth noting that the
more promising passive beamforming gain by (cooperative)
double IRSs is generally achieved at the expense of more
challenging channel estimation in practice, due to a large
number of additional channel coefficients introduced by the
inter-IRS links. To achieve practically low training overhead,
two efficient channel estimation schemes with ON/OFF reflec-
tion and always-ON reflection were recently proposed for
the double-IRS aided multi-user MISO system in [145] and
[146], respectively. In particular, by exploiting the intrinsic
relationship between the single- and double-reflection chan-
nels as well as that among different users, both double-IRS
channel estimation schemes in [145] and [146] were able to
achieve low training overhead and yet high channel estimation
accuracy.
3) LoS Versus Non-LoS (NLoS): As shown in the above,
one main design objective for IRS deployment with single-
antenna AP and user is to minimize their end-to-end path-
loss via IRS reflection(s) and thus achieve the maximum
receive SNR. Besides the link product-distance, the path-loss
is also critically dependent on the path-loss exponent, which
characterizes the average channel attenuation over distance.
Therefore, it is practically desirable to place IRS in LoS con-
ditions with AP/user as well as other cooperative IRSs to attain
the smallest (free-space) path-loss exponent (i.e., 2) that leads
to the minimum path-loss under a given link distance. This is
particularly important in mmWave/THz communications since
the NLoS paths are usually much weaker than the LoS path
due to severe signal blockage and penetration losses [20].
However, on the other hand, LoS channels are typically
of low-rank, which limits the achievable spatial multiplex-
ing gains of IRS-aided multi-antenna/multi-user systems with
multi-stream or SDMA transmissions and thus results in their
low capacity at the high-SNR regime [9], [71]. While deploy-
ing IRSs in locations with comparable NLoS (random) and
LoS (deterministic) channel components with AP/users/other
cooperating IRSs can be beneficial to improve the end-to-end
MIMO channel’s spatial power distribution (in terms of rank,
condition number, etc.) and hence its capacity. Therefore, there
is in general an interesting trade-off in IRS deployment for
choosing IRS locations to achieve balanced LoS versus NLoS
propagation as well as passive beamforming versus spatial
multiplexing gains, so as to maximize the system capacity.
B. IRS Deployment at the Network Level:
Distributed or Centralized?
Next, we address the IRS deployment for the more general
multi-user network with one AP communicating with K 1
users (or K groups of nearby users) that are located suffi-
ciently far apart from each other. In this case, there are two
different strategies to deploy N IRS reflecting elements in the
network: distributed deployment where the reflecting elements
form multiple distributed IRSs each located near one user,
as illustrated in Fig. 25 (a); or centralized deployment where
all the reflecting elements form one large IRS located in the
vicinity of the AP, as illustrated in Fig. 25 (b) [90]. Note that
for the single-user case with K = 1, the above two deployment
strategies are equivalent since they both achieve the maximum
received signal power at the AP as shown in Fig. 24. However,
for the multi-user case with K 1, the two deployment
strategies lead to distinct channels between the users and
AP in general. Specifically, under centralized deployment, all
users can be served by N reflecting elements; while under
distributed deployment, each user is served by its nearest
IRS with only a fraction of the N reflecting elements, since
the signals reflected by other (farther-away) IRSs are too
weak due to the much larger path loss. In the following,
we compare the achievable rates (or rate/capacity region) of
the K-user multiple access channel (MAC) in the uplink com-
munication9
with the above two IRS deployment strategies,
under different multiple access schemes. For the purpose of
exposition, we consider that the AP and users are all equipped
with a single antenna, and similar to Fig. 23, we ignore the
direct links between the users and AP. In addition, for fair
comparison of the two deployment strategies, we assume that
their respective user-IRS-AP channels are akin to each other
(the so-called “twin” channels in [90], [91]),10
as illustrated
in Fig. 25.
First, we consider TDMA where the K users communicate
with the AP in orthogonal time slots. For ease of exposition,
9The discussions for the downlink communication in the broadcast channel
(BC) are similar and can be found in [91].
10Specifically, for each IRS reflecting element, its channels with AP and
assigned user k under distributed deployment are equal to those with user k
and AP under centralized deployment, respectively.

Fig. 25. IRS deployment in a multi-user communication network.
we assume that N reflecting elements are equally divided
into K IRSs, each located near a different user under the
distributed deployment, which yields an O((N/K)2
) passive
beamforming gain per user. In contrast, under the centralized
deployment, each user can enjoy a passive beamforming gain
of O(N2
) by utilizing all the N reflecting elements. Thus,
it can be shown that the achievable rate region with centralized
deployment contains that with distributed deployment under
the twin channel condition [91]. In particular, by assuming
equal time allocations over the users and free-space LoS model
for all channels involved, it can be shown that the achievable
sum-rates of the K users under centralized/distributed deploy-
ment are given by Rcen = 1
K
K
k=1 log2

1 +
P̄ β2
0 N2
d2D2
kσ̄2

and
Rdis = 1
K
K
k=1 log2

1 +
P̄ β2
0 (N/K)2
d2D2
kσ̄2

, respectively, where
P̄ and σ̄2
denote the transmit power at each user and noise
power at the AP receiver, respectively. Thus, when N → ∞,
it can be shown that their asymptotic rate difference is given
by
Rcen − Rdis → 2 log2 K. (56)
Therefore, centralized deployment has more pronounced rate
gains over distributed deployment with increasing K, which is
intuitive since each user is assigned with a smaller number of
IRS reflecting elements and thus reduced passive beamfoming
gain in the latter case.
Next, we consider NOMA and FDMA where the users
communicate simultaneously with the AP in the same fre-
quency band and over orthogonal frequency bands, respec-
tively. Unlike TDMA, for centralized IRS deployment, the N
reflecting elements need to cater to the channels of all users at
the same time in both NOMA and FDMA, which thus results
in smaller passive beamforming gain per user than O(N2
).
In contrast, for distributed IRS deployment, the reflecting
elements of each distributed IRS are designed similarly as in
Fig. 26. Performance comparison of different IRS deployment strategies and
multiple access schemes in a two-user MAC.
the TDMA case, i.e., to maximize the passive beamforming
gain of its nearby user. Nevertheless, it was proved in [90], [91]
that under the twin channel condition, centralized deployment
outperforms distributed deployment in terms of multi-user
capacity region/achievable rate region for the case of NOMA
and FDMA, respectively. This is because the larger number
of reflecting elements at the centralized IRS provides more
flexibility to trade-off users’ individual achievable rates as
compared to distributed IRSs each helping one user only.
In Fig. 26, we compare the achievable rates of TDMA,
NOMA, and FDMA under the two deployment strategies in a
two-user MAC with d = 1 m, D1 = 1000 m, D2 = 200
m, P̄ = 15 dBm, σ̄2
= −90 dBm, and β0 = −30 dB.
All channels involved are assumed to follow the free-space
LoS model for the purpose of exposition. It is observed that
the capacity region/achievable rate regions under centralized
IRS deployment constitute their counterparts under distributed
IRS deployment in each case of TDMA, NOMA, and FDMA.
Furthermore, the gains are observed to be more pronounced
when the rates of the two users are asymmetric. It is also
observed that compared to distributed deployment, centralized
deployment achieves substantial rate improvement for the user
farther away from the AP, which thus helps deal with the
“near-far” issue and achieve more balanced rates for the two
users.
Despite that centralized IRS deployment is generally more
favorable than its distributed counterpart in terms of achievable
rates, it is worth noting that other practical factors may
also need to be taken into account for their implementation.
Firstly, distributed deployment requires more IRSs and thus
more backhaul links between the AP and IRS controllers
for exchanging information, which thus results in more over-
head for the network. Secondly, it may not be always fea-
sible to deploy a large centralized IRS near the AP due
to site/space constraint, while deploying multiple distributed
IRSs at the user side is generally more flexible. Thirdly,
the performance gains of centralized against distributed IRS
deployment is under the assumption of twin channel condition
[90], [91], while in practice, the channel statistics under the
two deployment strategies may vary drastically, in terms of

LoS probability, NLoS fading distribution, channel correlation,
etc., which may lead to distinct comparison results under
different scenarios (e.g., MIMO, multiple users served by each
distributed IRS).
C. Other Related Work and Future Direction
Besides the above discussions which have unveiled some
essential trade-offs/considerations for IRS deployment in prac-
tical systems, there are other related works on this topic in
the literature under various different setups, which are briefly
discussed in this subsection, together with some promising
directions for future work.
First, it is worthwhile to extend the results in [90], [91],
[143] discussed above to more general system setups, such
as multiple cooperative IRSs, other multi-user/MIMO channel
models, hybrid distributed and centralized IRS deployment,
etc. Furthermore, it is also important to study IRS deploy-
ment in the general multi-cell network with the inter-cell
interference. In [83], the authors considered a simplified setup
with one IRS serving two adjacent cells and showed that the
IRS should be deployed at the intersection of their edges to
maximally benefit both cells. However, if multiple IRSs are
available for serving multiple cells, it is still unknown of the
optimal IRS deployment/association strategy in general.
Moreover, for large-size wireless network with massively
deployed IRSs, how to optimally deploy them is a computa-
tionally formidable task. Therefore, new mathematical tools
such as stochastic geometry [12], [98], [147] and machine
learning [148] need to be invoked to perform system-level
analysis and efficient deployment design. It is expected that
these new tools will play an important role in the design
of IRS deployment in future wireless networks with hybrid
active/passive components as well as heterogeneous user QoS
requirements.
VI. EXTENSIONS
In this section, we overview other relevant topics on IRS
which are important and deserve further investigation.
A. Physical-Layer Security
Besides cryptographic algorithms and secure communica-
tion protocol-layer designs, physical-layer (PHY-layer) secu-
rity techniques can enhance the wireless network security by
altering the distributions of random noise, interference, and/or
fading channels in the legitimate and eavesdropping links,
together with the use of Wyner’s wiretap coding [149]. As
a new means to smartly reconfigure the wireless propagation,
IRS is highly promising to further boost the secrecy rate of
traditional wireless networks (see, e.g., [93], [94], [132], [132],
[150]–[160]), even for the challenging scenario when the con-
ventional PHY-layer security techniques are ineffective [10],
[93]. Specifically, if the eavesdropping channel is stronger than
the legitimate channel and/or they are highly correlated (e.g.,
when their channels are along the same direction from the
transmitter), the legitimate receiver can hardly achieve a posi-
tive secrecy rate using conventional techniques (such as trans-
mit beamforming and artificial noise (AN)/jamming). How-
ever, this issue can be efficiently addressed by deploying IRSs
in the vicinity of the legitimate/eavesdropping user and prop-
erly designing IRS passive beamforming to increase/reduce
the achievable rate of the legitimate/eavesdropping user, thus
significantly enhancing the achievable secrecy rate [93]. More-
over, AN/jamming by the transmitter can be jointly designed
with IRS passive beamforming to further reduce the eaves-
dropper’s rate without compromising the rate performance of
the legitimate user [155], [156].
Despite the above advantages, several practical issues in
the IRS-aided secrecy communication system need to be
addressed. First, exploiting IRS passive beamforming for
reducing the eavesdroppers’ rates requires CSI of the links
between the AP and eavesdroppers as well as between the
IRS and eavesdroppers, which is practically difficult to obtain
since 1) IRS usually has a massive number of passive reflecting
elements, and 2) the eavesdroppers may intentionally remain
covert or their CSI may be coarse/outdated if their signal
leakage is utilized for channel estimation. This thus calls
for new channel estimation designs as well as robust/secure
IRS passive beamforming by taking into account the incom-
plete/imperfect CSI of the eavesdroppers [132]. Moreover,
in a large-scale secrecy communication network with many
legitimate users and eavesdroppers as well as densely deployed
IRSs, IRS deployment is a key to maximize the network
secrecy throughput, which deserves further studies.
B. Wireless Power Transfer
RF-transmission enabled wireless power transfer (WPT) is
envisioned as a promising technology to prolong the battery
lives of Internet-of-Things (IoT) devices in future wireless
networks [161]. To overcome the high power loss over long
distance, various techniques have been proposed to be imple-
mented at the energy transmitter and/or receiver for improving
the WPT efficiency, such as waveform design, energy trans-
mission and scheduling, energy beamforming, rectifiers with
high energy conversion efficiency, and so on [161], [162].
However, the limited size of IoT devices results in small
receiving antenna aperture, which fundamentally constrains
their received power level in WPT systems. Fortunately, this
issue can be efficiently alleviated by deploying IRSs in the
proximity of IoT devices to create a highly efficient energy
charging zone for them (see, e.g., [66], [67], [163]–[165]).
By properly deploying IRSs in LoS with the energy transmit-
ters and receivers as well as exploiting their large aperture
and high passive beamforming gains, the received power of
nearby IoT devices can be substantially enhanced. Moreover,
the improvement of wireless charging efficiency in turn helps
reduce the transmit power as well as provide more flexibility in
designing the transmit beamforming for information receivers
and thus improves both the rate and energy performance in
SWIPT systems [66], [67], [163]–[165].
In practice, the benefits of IRS passive beamforming in
WPT/SWIPT critically depend on the available CSI at the
energy transmitter/AP, which needs to be acquired at the
cost of time and energy resources. Specifically, given a lim-
ited channel coherence time, the uplink pilot-assisted IRS
channel estimation (see Section IV) needs to balance the

trade-off between channel estimation and downlink power
transfer, since too little channel training will result in inac-
curate/insufficient CSI which in turn reduces the energy
beamforming gain, while too much training will consume
substantial energy at the energy receiver and also leave less
time for energy harvesting. In addition, IRS deployment (e.g.,
single versus multiple cooperative IRSs, centralized versus
distributed IRS as discussed in Section V) also greatly impacts
the WPT/SWIPT system performance, which is thus worthy
of further investigation.
C. UAV Communications
Most of the existing research on IRS has assumed terrestrial
IRSs at fixed locations, which, however, has the intrinsic
limitation that the IRS can only serve nearby users usually
located in half of the space, i.e., both the transmitter and
receiver must lie on the same side of the IRS. Recently,
unmanned aerial vehicle (UAV) has been exploited as a more
flexible communication platform to assist terrestrial wireless
communications [166]–[168]. To enlarge the service region,
a new network architecture called aerial IRS (AIRS) was
proposed in [169], [170], where the IRS is mounted on an
UAV to realize over-the-air intelligent reflection. Compared
to terrestrial IRS, AIRS is more likely to establish strong LoS
links with the ground nodes due to the relatively higher altitude
of UAV, thus reducing the blockage probability with them.
Besides, instead of the half-space reflection only by terrestrial
IRS, AIRS can realize 360◦
panoramic full-range reflection,
i.e., one AIRS can assist in the communication between any
pair of ground nodes within its coverage area, thus greatly
increasing the number of serviceable users as compared to
terrestrial IRS. Moreover, the coverage area of AIRS can be
further expanded by exploiting the high mobility of UAV
to move closer to multiple geographically separated users
sequentially to improve their communication performance by
exploiting short-range LoS channels as well as minimizing
the product distance of the IRS-reflected link. On the other
hand, terrestrial IRS can also be employed to help enhance
the UAV-ground communication performance via optimized
signal reflections [15], [171]–[176].
Compared to terrestrial IRS, AIRS brings new challenges.
First, for the general case of employing multiple AIRSs to aid
the communications among multiple ground nodes, it is crucial
to design the three dimensional (3D) placement/trajectories
of AIRSs jointly with their passive beamforming and user
association for maximizing the average rates of all users
under the practical UAV-ground channel models [167] (such
as the elevation-angle dependent Rician fading channel [177]
and the probabilistic LoS channel [178]). In addition, prac-
tical low-complexity channel estimation/tracking methods are
required for acquiring the AIRS-users channels over their 3D
trajectories, which is more challenging as compared to the
case of terrestrial IRSs that are at fixed locations. Moreover,
the difficulty in channel tracking makes the compensation of
the Doppler and delay spread a more critical issue due to the
high mobility of AIRS, which deserves further investigation
in the future [14], [15].
D. MmWave Communications
Due to the enormous bandwidth available in mmWave
frequencies (i.e., 30–300 GHz), mmWave communications are
promising to support gigabit-per-second data rates for enabling
rate-demanding wireless applications (augmented/virtual real-
ity (AR/VR), online gaming, etc.) in the future; whereas they
also incur significantly higher energy consumption and hard-
ware cost due to the increasing number of active antennas/RF
chains required operating at much higher frequencies as com-
pared to sub-6 GHz wireless systems. Moreover, the mmWave
communication channels are more susceptible to blockage and
subjected to higher propagation loss in general. These critical
issues can be efficiently resolved by properly deploying IRSs
in mmWave systems to create virtual LoS channels between
the BS/AP and users through smart reflections of the IRSs,
which can bypass the main obstructors between them, thus
achieving both high spectral and energy efficiencies [20], [77],
[179]–[183].
To fully reap the IRS passive beamforming gain in the
IRS-aided mmWave communication system, it is necessary to
conduct initial beam training/alignment at the BS/AP as well
as at the IRS to establish high-SNR links before executing
channel estimation and data transmission, due to the lim-
ited scattering of mmWave channels. Compared to traditional
mmWave systems, the beam training/alignment in the IRS-
aided mmWave system is much more involved due to the
following reasons. First, the large number of IRS reflecting
elements generate pencil-like IRS beam with narrow beam-
width, thus incurring higher training overhead for scanning
the beam space. This issue becomes more severe when the
IRS needs to serve multiple users, since the overhead of the
straightforward user-by-user successive beam training linearly
scales with the number of users. Second, the 3D IRS passive
beam training should be jointly designed with BS/AP’s active
beam training to adapt to the time-varying BS/AP-user and
IRS-user channels due to user mobility. Although some initial
attempts have been made to design the IRS beam training
(e.g., [122]), the training overhead is still overwhelming, thus
calling for new approaches for designing more efficient IRS
beam training for mmWave systems [131].
E. Mobile Edge Computing
Recent years have witnessed the computing paradigm shift
from centralized cloud computing to distributed edge com-
puting, which pushes abundant computational resources to
the edges of networks (e.g., BS and AP) to help offload
the computation tasks from mobile devices for reducing their
computation latency and energy consumption [184], [185].
Such computing performance gains are expected to be further
enhanced if IRS is deployed for mobile edge computing
(MEC) systems [88], [186]–[189]. Specifically, for a single
user, as IRS passive beamforming can improve its wireless
link capacity with the edge server, it is able to offload
computation-intensive tasks to the edge server for reducing
its computation latency without incurring high transmission
energy consumption. Furthermore, for a large-size MEC net-
work with a massive number of users and edge servers,

Fig. 27. Smart and reconfigurable radio environment with massive IRSs and multi-hop signal reflections.
IRS passive beamforming can have a great impact on the
offloading decisions and policies (i.e., amount of offloaded
data bits and radio-and-computational resources) of all users
by properly manipulating their offloading channels to different
edge servers, thus reshaping the distribution of offloaded
computation loads over the entire MEC network for achieving
better computational resource utilization and computation-
latency reduction. For example, instead of congesting the same
nearby server that would cause high computation latency,
some users can offload their computation tasks to the servers
farther away that are of light computation burdens, if their
corresponding links can be greatly improved by the IRS
passive beamforming.
As different users usually have asynchronous computation
task arrivals and completion deadlines, it is necessary to
jointly design the time-varying IRS passive beamforming for
serving different users with the offloading scheduling order
and multi-user time allocation based on their arrival-and-
deadline information as well as channel conditions, which is
worthy of further investigation in the future.
F. Other Miscellaneous Topics
In addition to the above mentioned ones, there are also
other interesting and promising topics on IRS that deserve
further studies, including IRS-aided cognitive radio network
[190]–[192], IRS-aided full-duplex communication [192]–
[194], IRS-aided sensing and localization [195]–[197], IRS-
aided THz communication [198]–[200], IRS-based spatial
modulation for data communication [201]–[204], etc., which
help significantly broaden the IRS research scope as well
as application horizon. For example, in the cognitive radio
network, properly deploying IRS and designing its passive
beamforming can not only enhance the received signal at the
secondary receiver but also suppress the interference at the
primary receiver [190]. Moreover, as IRS naturally operates
at the full-duplex and noise-free mode without the need of
costly self-interference cancellation, it can be integrated into
the conventional full-duplex wireless communication system
to enhance its spectral efficiency [192]. In addition, for the
challenging scenario of wireless localization where the direct
link between an anchor BS and a targeted agent is blocked,
an efficient positioning method is by properly deploying an
IRS and estimating both the direction and distance from the
agent to IRS via beam training [197]. Last but not least,
by leveraging the concept of index modulation, the IRS-based
spatial modulation encodes the information in the indices of
activated IRS reflecting elements selected, while its major
challenge lies in the limited IRS (PIN diode) switching speed,
which thus fundamentally constraints the system bandwidth
and hence the communication rate [201], [202].
VII. CONCLUSION
In this paper, we have provided a comprehensive tutorial
on the new IRS technology as a promising enabler for smart
and reconfigurable wireless communication environment. It
is shown that IRS-aided wireless communication has led
to a fundamental paradigm shift in the design of wireless
system/network from the traditional one with active com-
ponents only to a new hybrid architecture comprising both
active and passive components that co-work in an intelli-
gent way. Despite that the research on IRS-aided wireless
communication is still in its infancy, we have elaborated
the new challenges and potential directions for the research
into future IRS-aided wireless networks, spanning from the
theoretical IRS signal and channel modelling to practical
IRS beamforming design, channel estimation and deployment.
In addition, we have also highlighted the research trends
for several important IRS applications, including physical
layer security, wireless power transfer, UAV communication,
mobile edge computing, etc. Finally, most existing works

on IRS consider a limited number of IRSs with single-
/double-reflections per IRS only, whereas the ultimate goal of
achieving smart and reconfigurable radio environment requires
the dense deployment of IRSs by exploiting their multi-hop
reflections with cooperatively designed passive beamforming
and data routing [205], [206], as shown in Fig. 27. This thus
calls for significant new research effort to devise innovative
solutions for multi-IRS beamforming, channel estimation, and
deployment. As such, it is hoped that this paper will serve
as a useful and inspiring resource for future research on
IRS to unlock its full potential in future-generation (B5G/6G)
wireless communications.
REFERENCES
[1] N. Rajatheva et al., “Scoring the Terabit/s goal:Broadband con-
nectivity in 6G,” 2020, arXiv:2008.07220. [Online]. Available:
http://arxiv.org/abs/2008.07220
[2] F. Boccardi, R. W. Heath, A. Lozano, T. L. Marzetta, and P. Popovski,
“Five disruptive technology directions for 5G,” IEEE Commun. Mag.,
vol. 52, no. 2, pp. 74–80, Feb. 2014.
[3] M. Shafi et al., “5G: A tutorial overview of standards, trials, challenges,
deployment, and practice,” IEEE J. Sel. Areas Commun., vol. 35, no. 6,
pp. 1201–1221, Jun. 2017.
[4] S. Zhang, Q. Wu, S. Xu, and G. Y. Li, “Fundamental green tradeoffs:
Progresses, challenges, and impacts on 5G networks,” IEEE Commun.
Surveys Tuts., vol. 19, no. 1, pp. 33–56, 1st Quart., 2017.
[5] Q. Wu, G. Y. Li, W. Chen, D. W. K. Ng, and R. Schober, “An overview
of sustainable green 5G networks,” IEEE Wireless Commun., vol. 24,
no. 4, pp. 72–80, Aug. 2017.
[6] D. Tse and P. Viswanath, Fundamentals of Wireless Communication.
Cambridge, U.K.: Cambridge Univ. Press, 2005.
[7] A. Goldsmith, Wireless Communications. Cambridge, U.K.: Cambridge
Univ. Press, 2005.
[8] Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless
network: Joint active and passive beamforming design,” in Proc. IEEE
Global Commun. Conf. (GLOBECOM), Dec. 2018, pp. 1–6.
[9] Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless
network via joint active and passive beamforming,” IEEE Trans.
Wireless Commun., vol. 18, no. 11, pp. 5394–5409, Nov. 2019.
[10] Q. Wu and R. Zhang, “Towards smart and reconfigurable environment:
Intelligent reflecting surface aided wireless network,” IEEE Commun.
Mag., vol. 58, no. 1, pp. 106–112, Jan. 2020.
[11] S. Hu, F. Rusek, and O. Edfors, “Beyond massive MIMO: The potential
of data transmission with large intelligent surfaces,” IEEE Trans. Signal
Process., vol. 66, no. 10, pp. 2746–2758, May 2018.
[12] J. Lyu and R. Zhang, “Hybrid active/passive wireless network
aided by intelligent reflecting surface: System modeling and per-
formance analysis,” 2020, arXiv:2004.13318. [Online]. Available:
[13] S. Xia and Y. Shi, “Intelligent reflecting surface for massive device
connectivity: Joint activity detection and channel estimation,” in Proc.
IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP), May 2020,
pp. 5175–5179.
[14] E. Basar, “Reconfigurable intelligent surfaces for Doppler effect and
multipath fading mitigation,” 2019, arXiv:1912.04080. [Online]. Avail-
able: http://arxiv.org/abs/1912.04080
[15] B. Matthiesen, E. Bjornson, E. De Carvalho, and P. Popovski, “Intel-
ligent reflecting surface operation under predictable receiver mobility:
A continuous time propagation model,” IEEE Wireless Commun. Lett.,
early access, Sep. 18, 2020, doi: 10.1109/LWC.2020.3024781.
[16] Z. Huang, B. Zheng, and R. Zhang, “Transforming fading channel
from fast to slow: IRS-assisted high-mobility communication,” 2020,
arXiv:2011.03147. [Online]. Available: http://arxiv.org/abs/2011.03147
[17] M. Latva-Aho and K. Leppänen, “Key drivers and research challenges
for 6G ubiquitous wireless intelligence,” 6G Flagship, Univ. Oulu,
Oulu, Finland, Tech. Rep., 2019.
[18] N. Rajatheva et al., “White paper on broadband connec-
tivity in 6G,” 2020, arXiv:2004.14247. [Online]. Available:
[19] L. Subrt and P. Pechac, “Intelligent walls as autonomous parts of smart
indoor environments,” IET Commun., vol. 6, no. 8, pp. 1004–1010,
May 2012.
[20] X. Tan, Z. Sun, D. Koutsonikolas, and J. M. Jornet, “Enabling
indoor mobile millimeter-wave networks based on smart reflect-arrays,”
in Proc. IEEE Conf. Comput. Commun. (INFOCOM), Apr. 2018,
pp. 270–278.
[21] M. Di Renzo et al., “Smart radio environments empowered by recon-
figurable AI meta-surfaces: An idea whose time has come,” EURASIP
J. Wireless Commun. Netw., vol. 2019, no. 1, pp. 1–20, May 2019.
[22] E. Basar, M. Di Renzo, J. De Rosny, M. Debbah, M.-S. Alouini, and
R. Zhang, “Wireless communications through reconfigurable intelligent
surfaces,” IEEE Access, vol. 7, pp. 116753–116773, Aug. 2019.
[23] Y.-C. Liang, R. Long, Q. Zhang, J. Chen, H. V. Cheng, and H. Guo,
“Large intelligent surface/antennas (LISA): Making reflective radios
smart,” J. Commun. Inf. Netw., vol. 4, no. 2, pp. 40–50, Jun. 2019.
[24] V. Arun and H. Balakrishnan, “RFocus: Practical beamforming
for small devices,” 2019, arXiv:1905.05130. [Online]. Available:
[25] NTT DOCOMO and Metawave Announce Successful Demonstration
of 28 GHz-Band 5G Using World’s First Meta-Structure Technology.
Accessed: Jul. 2020. [Online]. Available: https://www.businesswire.
com/news/home/20181204005253/en/NTT-DOCOMO-Metawave-
Announce-Successful-Demonstration-28GHz-Band
[26] TowerJazz and Lumotive Demonstrate Solid-State Beam Steer-
ing for LiDAR. Accessed: Jul. 2020. [Online]. Available: https://
techtime.news/2019/06/26/lidar/
[27] Pivotal Commware Achieves Gigabit Connectivity in Live 5G
mmWave Demo at Mobile World Congress Los Angeles 2019.
Accessed: Jul. 2020. [Online]. Available: https://pivotalcommware.
com/2019/11/04/pivotal-commware-achieves-gigabit-connectivity-in-
live-5g-mmwave-demo-at-mobile-world-congress-los-angeles-2019/
[28] DOCOMO Conducts World’s First Successful Trial of Transparent
Dynamic Metasurface. Accessed: Jul. 2020. [Online]. Available: https://
www.nttdocomo.co.jp/english/info/media_center/pr/2020/0117_00.html
[29] A Technological Platform to Simplify all EM Infrastructures: Replacing
Hardware Complexity by Algorithms. Accessed: Jul. 2020. [Online].
Available: http://greenerwave.com/our-technology/
[30] A Hardware Platform for Software-Driven Functional Metasurfaces.
Accessed: Jul. 2020. [Online]. Available: http://www.visorsurf.eu
[31] Artificial Intelligence Aided D-Band Network for 5G Long Term
Evolution. Accessed: Jul. 2020. [Online]. Available: https://www.ict-
ariadne.eu
[32] Harnessing Multipath Propagation in Wireless Networks: A Meta-
Surface Transformation of Wireless Networks Into Smart Reconfig-
urable Radio Environments. Accessed: Jul. 2020. [Online]. Available:
https://cordis.europa.eu/project/id/891030
[33] Q. He, S. Sun, and L. Zhou, “Tunable/reconfigurable metasur-
faces: Physics and applications,” Research, vol. 2019, Jul. 2019,
Art. no. 1849272.
[34] S. V. Hum and J. Perruisseau-Carrier, “Reconfigurable reflectarrays and
array lenses for dynamic antenna beam control: A review,” IEEE Trans.
Antennas Propag., vol. 62, no. 1, pp. 183–198, Jan. 2014.
[35] C. Liaskos, S. Nie, A. Tsioliaridou, A. Pitsillides, S. Ioannidis,
and I. Akyildiz, “A new wireless communication paradigm through
software-controlled metasurfaces,” IEEE Commun. Mag., vol. 56, no. 9,
pp. 162–169, Sep. 2018.
[36] M. Di Renzo et al., “Reconfigurable intelligent surfaces vs. Relaying:
Differences, similarities, and performance comparison,” IEEE Open J.
Commun. Soc., vol. 1, pp. 798–807, Jun. 2020.
[37] M. A. ElMossallamy, H. Zhang, L. Song, K. G. Seddik, Z. Han,
and G. Ye Li, “Reconfigurable intelligent surfaces for wireless
communications: Principles, challenges, and opportunities,” 2020,
[38] C. Huang et al., “Holographic MIMO surfaces for 6G wireless net-
works: Opportunities, challenges, and trends,” IEEE Wireless Commun.,
vol. 27, no. 5, pp. 118–125, Oct. 2020.
[39] M. Di Renzo et al., “Smart radio environments empowered by reconfig-
urable intelligent surfaces: How it works, state of research, and the road
ahead,” IEEE J. Sel. Areas Commun., vol. 38, no. 11, pp. 2450–2525,
Nov. 2020.
[40] X. Yuan, Y.-J. Angela Zhang, Y. Shi, W. Yan, and H. Liu,
“Reconfigurable-intelligent-surface empowered wireless communi-
cations: Challenges and opportunities,” 2020, arXiv:2001.00364.
[Online]. Available: http://arxiv.org/abs/2001.00364
[41] S. Gong et al., “Towards smart radio environment for wireless commu-
nications via intelligent reflecting surfaces: A comprehensive survey,”
IEEE Commun. Surveys Tuts., to be published.

[42] E. Björnson, Ö. Özdogan, and E. G. Larsson, “Reconfigurable
intelligent surfaces: Three myths and two critical questions,” 2020,
[43] C. Huang, A. Zappone, M. Debbah, and C. Yuen, “Achievable rate
maximization by passive intelligent mirrors,” in Proc. IEEE Int. Conf.
Acoust., Speech Signal Process. (ICASSP), Apr. 2018, pp. 3714–3718.
[44] B. B. Baker and E. T. Copson, The Mathematical Theory of Huygens’
Principle, vol. 329. Providence, RI, USA: American Mathematical
Society, 2003.
[45] J. C. B. Garcia, A. Sibille, and M. Kamoun, “Reconfigurable intelligent
surfaces: Bridging the gap between scattering and reflection,” IEEE J.
Sel. Areas Commun., vol. 38, no. 11, pp. 2538–2547, Nov. 2020.
[46] W. Tang et al., “Wireless communications with reconfigurable intelli-
gent surface: Path loss modeling and experimental measurement,” 2019,
[47] O. Ozdogan, E. Bjornson, and E. G. Larsson, “Intelligent reflecting
surfaces: Physics, propagation, and pathloss modeling,” IEEE Wireless
Commun. Lett., vol. 9, no. 5, pp. 581–585, May 2020.
[48] S. W. Ellingson, “Path loss in reconfigurable intelligent surface-
enabled channels,” 2019, arXiv:1912.06759. [Online]. Available:
[49] M. Di Renzo, F. Habibi Danufane, X. Xi, J. de Rosny, and S. Tretyakov,
“Analytical modeling of the path-loss for reconfigurable intelligent
surfaces–anomalous mirror or scatterer ?” 2020, arXiv:2001.10862.
[50] T. J. Cui, M. Q. Qi, X. Wan, J. Zhao, and Q. Cheng, “Coding
metamaterials, digital metamaterials and programmable metamaterials,”
Light: Sci. Appl., vol. 3, no. 10, p. e218, Oct. 2014.
[51] P. Nayeri, F. Yang, and A. Z. Elsherbeni, Reflectarray Antennas:
Theory, Designs, and Applications. Hoboken, NJ, USA: Wiley, 2018.
[52] L. Zhang et al., “Space-time-coding digital metasurfaces,” Nature
Commun., vol. 9, no. 1, p. 4334, 2018.
[53] H. Yang et al., “Design of resistor-loaded reflectarray elements for both
amplitude and phase control,” IEEE Antennas Wireless Propag. Lett.,
vol. 16, pp. 1159–1162, Nov. 2017.
[54] Q. Wu and R. Zhang, “Beamforming optimization for intelligent
reflecting surface with discrete phase shifts,” in Proc. IEEE Int. Conf.
Acoust., Speech Signal Process. (ICASSP), May 2019, pp. 7830–7833.
[55] S. Abeywickrama, R. Zhang, and C. Yuen, “Intelligent reflecting
surface: Practical phase shift model and beamforming optimization,”
in Proc. IEEE Int. Conf. Commun. (ICC), Jun. 2020, pp. 1–6.
[56] S. Abeywickrama, R. Zhang, Q. Wu, and C. Yuen, “Intelligent reflect-
ing surface: Practical phase shift model and beamforming optimiza-
tion,” IEEE Trans. Commun., vol. 68, no. 9, pp. 5849–5863, Sep. 2020.
[57] B. O. Zhu, J. Zhao, and Y. Feng, “Active impedance metasurface with
full 360◦ reflection phase tuning,” Sci. Rep., vol. 3, p. 3059, Oct. 2013.
[58] N. Kaina, M. Dupré, G. Lerosey, and M. Fink, “Shaping complex
microwave fields in reverberating media with binary tunable metasur-
faces,” Sci. Rep., vol. 4, no. 1, p. 6693, Oct. 2014.
[59] S. Zhang, X. Chen, and G. F. Pedersen, “Mutual coupling suppression
with decoupling ground for massive MIMO antenna arrays,” IEEE
Trans. Veh. Technol., vol. 68, no. 8, pp. 7273–7282, Aug. 2019.
[60] H. Quoc Ngo, E. G. Larsson, and T. L. Marzetta, “Energy and
spectral efficiency of very large multiuser MIMO systems,” IEEE
Trans. Commun., vol. 61, no. 4, pp. 1436–1449, Apr. 2013.
[61] J. Hoydis, S. ten Brink, and M. Debbah, “Massive MIMO in the UL/DL
of cellular networks: How many antennas do we need?” IEEE J. Sel.
Areas Commun., vol. 31, no. 2, pp. 160–171, Feb. 2013.
[62] Q.-U.-A. Nadeem, A. Kammoun, A. Chaaban, M. Debbah, and
M.-S. Alouini, “Asymptotic max-min SINR analysis of reconfigurable
intelligent surface assisted MISO systems,” IEEE Trans. Wireless
Commun., vol. 19, no. 12, pp. 7748–7764, Dec. 2020.
[63] A. M.-C. So, J. Zhang, and Y. Ye, “On approximating complex
quadratic optimization problems via semidefinite programming relax-
ations,” Math. Program., vol. 110, no. 1, pp. 93–110, Mar. 2007.
[64] Q. Wu and R. Zhang, “Beamforming optimization for wireless
network aided by intelligent reflecting surface with discrete phase
shifts,” IEEE Trans. Commun., vol. 68, no. 3, pp. 1838–1851,
Mar. 2020.
[65] D. P. Bertsekas, Nonlinear Programming. Cambridge, MA, USA: MIT
Press, 1999.
[66] Q. Wu and R. Zhang, “Joint active and passive beamforming optimiza-
tion for intelligent reflecting surface assisted SWIPT under QoS con-
straints,” IEEE J. Sel. Areas Commun., vol. 38, no. 8, pp. 1735–1748,
Aug. 2020.
[67] Q. Wu and R. Zhang, “Weighted sum power maximization for intelli-
gent reflecting surface aided SWIPT,” IEEE Wireless Commun. Lett.,
vol. 9, no. 5, pp. 586–590, May 2020.
[68] Y. Yang, B. Zheng, S. Zhang, and R. Zhang, “Intelligent reflect-
ing surface meets OFDM: Protocol design and rate maximiza-
tion,” IEEE Trans. Commun., vol. 68, no. 7, pp. 4522–4535,
Jul. 2020.
[69] X. Yu, D. Xu, and R. Schober, “Optimal beamforming for MISO
communications via intelligent reflecting surfaces,” in Proc. IEEE
21st Int. Workshop Signal Process. Adv. Wireless Commun. (SPAWC),
May 2020, pp. 1–5.
[70] K. Feng, Q. Wang, X. Li, and C. Wen, “Deep reinforcement learning
based intelligent reflecting surface optimization for MISO communica-
tion systems,” IEEE Wireless Commun. Lett., vol. 9, no. 5, pp. 745–749,
May 2020.
[71] S. Zhang and R. Zhang, “Capacity characterization for intelligent
reflecting surface aided MIMO communication,” IEEE J. Sel. Areas
Commun., vol. 38, no. 8, pp. 1823–1838, Aug. 2020.
[72] B. Zheng and R. Zhang, “Intelligent reflecting surface-enhanced
OFDM: Channel estimation and reflection optimization,” IEEE Wireless
Commun. Lett., vol. 9, no. 4, pp. 518–522, Apr. 2020.
[73] B. Zheng, Q. Wu, and R. Zhang, “Intelligent reflecting surface-assisted
multiple access with user pairing: NOMA or OMA?” IEEE Commun.
Lett., vol. 24, no. 4, pp. 753–757, Apr. 2020.
[74] Y. Yang, S. Zhang, and R. Zhang, “IRS-enhanced OFDMA: Joint
resource allocation and passive beamforming optimization,” IEEE
Wireless Commun. Lett., vol. 9, no. 6, pp. 760–764, Jun. 2020.
[75] M.-M. Zhao, Q. Wu, M.-J. Zhao, and R. Zhang, “Intelligent reflect-
ing surface enhanced wireless networks: Two-timescale beamforming
optimization,” IEEE Trans. Wireless Commun., vol. 20, no. 1, pp. 2–17,
Jan. 2021.
[76] M.-M. Zhao, Q. Wu, M.-J. Zhao, and R. Zhang, “Exploiting amplitude
control in intelligent reflecting surface aided wireless communication
with imperfect CSI,” 2020, arXiv:2005.07002. [Online]. Available:
[77] B. Di, H. Zhang, L. Song, Y. Li, Z. Han, and H. V. Poor, “Hybrid
beamforming for reconfigurable intelligent surface based multi-user
communications: Achievable rates with limited discrete phase shifts,”
IEEE J. Sel. Areas Commun., vol. 38, no. 8, pp. 1809–1822, Aug. 2020.
[78] H. Zhang, B. Di, L. Song, and Z. Han, “Reconfigurable intelligent
surfaces assisted communications with limited phase shifts: How many
phase shifts are enough?” IEEE Trans. Veh. Technol., vol. 69, no. 4,
pp. 4498–4502, Apr. 2020.
[79] J. Xu, W. Xu, and A. L. Swindlehurst, “Discrete phase shift design
for practical large intelligent surface communication,” in Proc. IEEE
Pacific Rim Conf. Commun., Comput. Signal Process. (PACRIM),
Aug. 2019, pp. 1–5.
[80] C. Huang, G. C. Alexandropoulos, A. Zappone, M. Debbah, and
C. Yuen, “Energy efficient multi-user MISO communication using
low resolution large intelligent surfaces,” in Proc. IEEE Globecom
Workshops (GC Wkshps), Dec. 2018, pp. 1–6.
[81] S. Hu, Z. Wei, Y. Cai, D. Wing Kwan Ng, and J. Yuan, “Sum-
rate maximization for multiuser MISO downlink systems with
self-sustainable IRS,” 2020, arXiv:2005.11663. [Online]. Available:
[82] M. Hua, Q. Wu, D. Wing Kwan Ng, J. Zhao, and L. Yang,
“Intelligent reflecting surface-aided joint processing coordinated mul-
tipoint transmission,” 2020, arXiv:2003.13909. [Online]. Available:
[83] C. Pan et al., “Multicell MIMO communications relying on intelligent
reflecting surfaces,” IEEE Trans. Wireless Commun., vol. 19, no. 8,
pp. 5218–5233, Aug. 2020.
[84] W. Mei and R. Zhang, “Performance analysis and user associa-
tion optimization for wireless network aided by multiple intelli-
gent reflecting surfaces,” 2020, arXiv:2009.02551. [Online]. Available:
[85] Y. Yang, S. Zhang, and R. Zhang, “IRS-enhanced OFDM: Power
allocation and passive array optimization,” in Proc. IEEE Global
Commun. Conf. (GLOBECOM), Dec. 2019, pp. 1–6.
[86] S. Zhang and R. Zhang, “On the capacity of intelligent reflecting
surface aided MIMO communication,” in Proc. IEEE Int. Symp. Inf.
Theory (ISIT), Jun. 2020, pp. 2977–2982.
[87] C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and
C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in
wireless communication,” IEEE Trans. Wireless Commun., vol. 18,
no. 8, pp. 4157–4170, Aug. 2019.

[88] Y. Liu et al., “Intelligent reflecting surface meets mobile edge
computing: Enhancing wireless communications for computation
offloading,” 2020, arXiv:2001.07449. [Online]. Available:
[89] Q. Tao, S. Zhang, C. Zhong, and R. Zhang, “Intelligent reflecting
surface aided multicasting with random passive beamforming,” IEEE
Wireless Commun. Lett., vol. 10, no. 1, pp. 92–96, Jan. 2021.
[90] S. Zhang and R. Zhang, “Intelligent reflecting surface aided multiple
access: Capacity region and deployment strategy,” in Proc. IEEE
21st Int. Workshop Signal Process. Adv. Wireless Commun. (SPAWC),
May 2020, pp. 1–5.
[91] S. Zhang and R. Zhang, “Intelligent reflecting surface aided multi-
user communication: Capacity region and deployment strategy,” 2020,
[92] H. Guo, Y.-C. Liang, J. Chen, and E. G. Larsson, “Weighted
sum-rate optimization for intelligent reflecting surface enhanced
wireless networks,” 2019, arXiv:1905.07920. [Online]. Available:
[93] M. Cui, G. Zhang, and R. Zhang, “Secure wireless communication via
intelligent reflecting surface,” IEEE Wireless Commun. Lett., vol. 8,
no. 5, pp. 1410–1414, Oct. 2019.
[94] H. Yang et al., “Intelligent reflecting surface assisted anti-jamming
communications: A fast reinforcement learning approach,” 2020,
[95] C. Huang, R. Mo, and C. Yuen, “Reconfigurable intelligent surface
assisted multiuser MISO systems exploiting deep reinforcement learn-
ing,” IEEE J. Sel. Areas Commun., vol. 38, no. 8, pp. 1839–1850,
Aug. 2020.
[96] Z. Zhang, Y. Cui, F. Yang, and L. Ding, “Analysis and opti-
mization of outage probability in multi-intelligent reflecting surface-
assisted systems,” 2019, arXiv:1909.02193. [Online]. Available:
[97] L. Yang, F. Meng, Q. Wu, D. B. da Costa, and M.-S. Alouini,
“Accurate closed-form approximations to channel distributions of RIS-
aided wireless systems,” IEEE Wireless Commun. Lett., vol. 9, no. 11,
pp. 1985–1989, Nov. 2020.
[98] J. Lyu and R. Zhang, “Spatial throughput characterization for intelligent
reflecting surface aided multiuser system,” IEEE Wireless Commun.
Lett., vol. 9, no. 6, pp. 834–838, Jun. 2020.
[99] M.-M. Zhao, Q. Wu, M.-J. Zhao, and R. Zhang, “Exploiting amplitude
control in intelligent reflecting surface aided wireless communication
with imperfect CSI,” 2020, arXiv:2005.07002. [Online]. Available:
[100] B. Zheng, C. You, and R. Zhang, “Intelligent reflecting surface
assisted multi-user OFDMA: Channel estimation and training design,”
IEEE Trans. Wireless Commun., vol. 19, no. 12, pp. 8315–8329,
Dec. 2020.
[101] G. C. Alexandropoulos and E. Vlachos, “A hardware architecture for
reconfigurable intelligent surfaces with minimal active elements for
explicit channel estimation,” in Proc. IEEE Int. Conf. Acoust., Speech
Signal Process. (ICASSP), May 2020, pp. 9175–9179.
[102] A. Taha, M. Alrabeiah, and A. Alkhateeb, “Enabling large intelli-
gent surfaces with compressive sensing and deep learning,” 2019,
[103] A. Taha, M. Alrabeiah, and A. Alkhateeb, “Deep learning for large
intelligent surfaces in millimeter wave and massive MIMO systems,”
in Proc. IEEE Global Commun. Conf. (GLOBECOM), Dec. 2019,
pp. 1–6.
[104] A. Taha, Y. Zhang, F. B. Mismar, and A. Alkhateeb, “Deep reinforce-
ment learning for intelligent reflecting surfaces: Towards standalone
operation,” in Proc. IEEE 21st Int. Workshop Signal Process. Adv.
Wireless Commun. (SPAWC), May 2020, pp. 1–5.
[105] W. Han, P. Chen, and Z. Cao, “Deep neural network-based quantized
signal reconstruction for DOA estimation,” 2020, arXiv:2005.01102.
[106] D. Mishra and H. Johansson, “Channel estimation and low-complexity
beamforming design for passive intelligent surface assisted MISO
wireless energy transfer,” in Proc. IEEE Int. Conf. Acoust., Speech
Signal Process. (ICASSP), May 2019, pp. 4659–4663.
[107] T. L. Jensen and E. De Carvalho, “An optimal channel estimation
scheme for intelligent reflecting surfaces based on a minimum variance
unbiased estimator,” in Proc. IEEE Int. Conf. Acoust., Speech Signal
Process. (ICASSP), May 2020, pp. 5000–5004.
[108] B. Zheng, C. You, and R. Zhang, “Fast channel estimation for
IRS-assisted OFDM,” 2020, arXiv:2008.04476. [Online]. Available:
[109] P. Wang, J. Fang, H. Duan, and H. Li, “Compressed channel estimation
for intelligent reflecting surface-assisted millimeter wave systems,”
IEEE Signal Process. Lett., vol. 27, pp. 905–909, May 2020.
[110] Z.-Q. He and X. Yuan, “Cascaded channel estimation for large intel-
ligent metasurface assisted massive MIMO,” IEEE Wireless Commun.
Lett., vol. 9, no. 2, pp. 210–214, Feb. 2020.
[111] G. T. de Araujo and A. L. F. de Almeida, “PARAFAC-based channel
estimation for intelligent reflective surface assisted MIMO system,” in
Proc. IEEE 11th Sensor Array Multichannel Signal Process. Workshop
(SAM), Jun. 2020, pp. 1–5.
[112] J. Mirza and B. Ali, “Channel estimation method and phase shift design
for reconfigurable intelligent surface assisted MIMO networks,” 2019,
[113] J. He, M. Leinonen, H. Wymeersch, and M. Juntti, “Channel estimation
for RIS-aided mmwave MIMO channels,” 2020, arXiv:2002.06453.
[114] J. Chen, Y.-C. Liang, H. Victor Cheng, and W. Yu, “Chan-
nel estimation for reconfigurable intelligent surface aided multi-
user MIMO systems,” 2019, arXiv:1912.03619. [Online]. Available:
[115] C. Hu and L. Dai, “Two-timescale channel estimation for recon-
figurable intelligent surface aided wireless communications,” 2019,
[116] H. Liu, X. Yuan, and Y.-J.-A. Zhang, “Matrix-calibration-based cas-
caded channel estimation for reconfigurable intelligent surface assisted
multiuser MIMO,” IEEE J. Sel. Areas Commun., vol. 38, no. 11,
pp. 2621–2636, Nov. 2020.
[117] L. Wei, C. Huang, G. C. Alexandropoulos, and C. Yuen, “Parallel factor
decomposition channel estimation in RIS-assisted multi-user MISO
communication,” in Proc. IEEE 11th Sensor Array Multichannel Signal
Process. Workshop (SAM), Jun. 2020, pp. 1–5.
[118] Z. Wan, Z. Gao, and M.-S. Alouini, “Broadband channel estimation for
intelligent reflecting surface aided mmwave massive MIMO systems,”
[119] S. Khan, K. S Khan, N. Haider, and S. Young Shin, “Deep-
learning-aided detection for reconfigurable intelligent surfaces,” 2019,
[120] A. M. Elbir, A. Papazafeiropoulos, P. Kourtessis, and S. Chatzinotas,
“Deep channel learning for large intelligent surfaces aided mm-wave
massive MIMO systems,” IEEE Wireless Commun. Lett., vol. 9, no. 9,
pp. 1447–1451, Sep. 2020.
[121] B. Ning, Z. Chen, W. Chen, and Y. Du, “Channel estimation and
transmission for intelligent reflecting surface assisted THz communi-
cations,” in Proc. IEEE Int. Conf. Commun. (ICC), Jun. 2020, pp. 1–7.
[122] B. Ning, Z. Chen, W. Chen, Y. Du, and J. Fang, “Terahertz multi-user
massive MIMO with intelligent reflecting surface: Beam training and
hybrid beamforming,” 2019, arXiv:1912.11662. [Online]. Available:
[123] C. Liu, X. Liu, D. Wing Kwan Ng, and J. Yuan, “Deep residual learning
for channel estimation in intelligent reflecting surface-assisted multi-
user communications,” 2020, arXiv:2009.01423. [Online]. Available:
[124] Q.-U.-U. Nadeem, A. Kammoun, A. Chaaban, M. Debbah, and
M.-S. Alouini, “Intelligent reflecting surface assisted wireless commu-
nication: Modeling and channel estimation,” 2019, arXiv:1906.02360.
[125] Z. Wang, L. Liu, and S. Cui, “Channel estimation for intelligent
reflecting surface assisted multiuser communications: Framework, algo-
rithms, and analysis,” IEEE Trans. Wireless Commun., vol. 19, no. 10,
pp. 6607–6620, Oct. 2020.
[126] X. Guan, Q. Wu, and R. Zhang, “Anchor-assisted intelligent reflecting
surface channel estimation for multiuser communications,” in Proc.
IEEE GLOBECOM, Dec. 2020.
[127] C. You, B. Zheng, and R. Zhang, “Intelligent reflecting surface with
discrete phase shifts: Channel estimation and passive beamforming,”
[128] C. You, B. Zheng, and R. Zhang, “Channel estimation and passive
beamforming for intelligent reflecting surface: Discrete phase shift and
progressive refinement,” IEEE J. Sel. Areas Commun., vol. 38, no. 11,
pp. 2604–2620, Nov. 2020.
[129] L. Wei et al., “Channel estimation for RIS-empowered multi-user
MISO wireless communications,” 2020, arXiv:2008.01459. [Online].
Available: http://arxiv.org/abs/2008.01459
[130] M.-M. Zhao, A. Liu, and R. Zhang, “Outage-constrained robust
beamforming for intelligent reflecting surface aided wireless
communication,” 2020, arXiv:2007.10769. [Online]. Available:

[131] C. You, B. Zheng, and R. Zhang, “Wireless communication via
double IRS: Channel estimation and passive beamforming designs,”
IEEE Wireless Commun. Lett., early access, Oct. 28, 2020, doi:
10.1109/LWC.2020.3034388.
[132] X. Yu, D. Xu, Y. Sun, D. W. K. Ng, and R. Schober, “Robust and secure
wireless communications via intelligent reflecting surfaces,” IEEE
J. Sel. Areas Commun., vol. 38, no. 11, pp. 2637–2652, Nov. 2020.
[133] G. Zhou, C. Pan, H. Ren, K. Wang, M. D. Renzo, and A. Nallanathan,
“Robust beamforming design for intelligent reflecting surface aided
MISO communication systems,” IEEE Wireless Commun. Lett., vol. 9,
no. 10, pp. 1658–1662, Oct. 2020.
[134] W. Wang and W. Zhang, “Joint beam training and position-
ing for intelligent reflecting surfaces assisted millimeter wave
communications,” 2020, arXiv:2009.03536. [Online]. Available:
[135] C. You, B. Zheng, and R. Zhang, “Fast beam training for IRS-assisted
multiuser communications,” IEEE Wireless Commun. Lett., vol. 9,
no. 11, pp. 1845–1849, Nov. 2020.
[136] C. Huang, G. C. Alexandropoulos, C. Yuen, and M. Debbah, “Indoor
signal focusing with deep learning designed reconfigurable intelligent
surfaces,” in Proc. IEEE 20th Int. Workshop Signal Process. Adv.
Wireless Commun. (SPAWC), Jul. 2019, pp. 1–5.
[137] T. Jiang, H. Victor Cheng, and W. Yu, “Learning to reflect
and to beamform for intelligent reflecting surface with implicit
channel estimation,” 2020, arXiv:2009.14404. [Online]. Available:
[138] F. Jiang, L. Yang, D. Benevides da Costa, and Q. Wu, “Chan-
nel estimation via direct calculation and deep learning for RIS-
aided mmWave systems,” 2020, arXiv:2008.04704. [Online]. Available:
[139] J. Pan, L. Cai, Y. T. Hou, Y. Shi, and S. X. Shen, “Optimal base-station
locations in two-tiered wireless sensor networks,” IEEE Trans. Mobile
Comput., vol. 4, no. 5, pp. 458–473, Sep. 2005.
[140] K. Akkaya, M. Younis, and W. Youssef, “Positioning of base stations
in wireless sensor networks,” IEEE Commun. Mag., vol. 45, no. 4,
pp. 96–102, Apr. 2007.
[141] M. Yu, J. Li, and H. Sadjadpour, “Amplify-forward and decode-
forward: The impact of location and capacity contour,” in Proc. IEEE
Mil. Commun. Conf. (MILCOM), Oct. 2005, pp. 1609–1615.
[142] B. Lin, P.-H. Ho, L.-L. Xie, X. Shen, and J. Tapolcai, “Optimal relay
station placement in broadband wireless access networks,” IEEE Trans.
Mobile Comput., vol. 9, no. 2, pp. 259–269, Feb. 2010.
[143] Y. Han, S. Zhang, L. Duan, and R. Zhang, “Cooperative double-
IRS aided communication: Beamforming design and power scal-
ing,” IEEE Wireless Commun. Lett., vol. 9, no. 8, pp. 1206–1210,
Aug. 2020.
[144] B. Zheng, C. You, and R. Zhang, “Double-IRS assisted multi-
user MIMO: Cooperative passive beamforming design,” 2020,
arXiv:2008.13701. [Online]. Available: http://arxiv.org/abs/
2008.13701
[145] B. Zheng, C. You, and R. Zhang, “Uplink channel estimation for
double-IRS assisted multi-user MIMO,” 2020, arXiv:2010.06155.
[146] B. Zheng, C. You, and R. Zhang, “Efficient channel estimation for
double-IRS aided multi-user MIMO system,” 2020, arXiv:2011.00738.
[147] M. A. Kishk and M.-S. Alouini, “Exploiting randomly-located
blockages for large-scale deployment of intelligent surfaces,” IEEE
J. Sel. Areas Commun., early access, Aug. 24, 2020, doi:
10.1109/JSAC.2020.3018808.
[148] X. Liu, Y. Liu, Y. Chen, and H. V. Poor, “RIS enhanced massive
non-orthogonal multiple access networks: Deployment and passive
beamforming design,” IEEE J. Sel. Areas Commun., early access,
Aug. 24, 2020, doi: 10.1109/JSAC.2020.3018823.
[149] S. Leung-Yan-Cheong and M. Hellman, “The Gaussian wire-tap chan-
nel,” IEEE Trans. Inf. Theory, vol. 24, no. 4, pp. 451–456, Jul. 1978.
[150] H. Yang et al., “Intelligent reflecting surface assisted anti-jamming
communications based on reinforcement learning,” in Proc. IEEE
GLOBECOM, Dec. 2020.
[151] X. Zhou, S. Yan, Q. Wu, F. Shu, and D. Wing Kwan Ng, “Intel-
ligent reflecting surface (IRS)-aided covert wireless communication
with delay constraint,” 2020, arXiv:2011.03726. [Online]. Available:
[152] J. Chen, Y.-C. Liang, Y. Pei, and H. Guo, “Intelligent reflecting surface:
A programmable wireless environment for physical layer security,”
IEEE Access, vol. 7, pp. 82599–82612, Jul. 2019.
[153] X. Lu, W. Yang, X. Guan, Q. Wu, and Y. Cai, “Robust and secure
beamforming for intelligent reflecting surface aided mmWave MISO
systems,” IEEE Wireless Commun. Lett., vol. 9, no. 12, pp. 2068–2072,
Dec. 2020.
[154] Z. Zhang, L. Lv, Q. Wu, H. Deng, and J. Chen, “Robust and
secure communications in intelligent reflecting surface assisted
NOMA networks,” 2020, arXiv:2009.00267. [Online]. Available:
[155] X. Guan, Q. Wu, and R. Zhang, “Intelligent reflecting surface assisted
secrecy communication: Is artificial noise helpful or not?” IEEE
Wireless Commun. Lett., vol. 9, no. 6, pp. 778–782, Jun. 2020.
[156] D. Xu, X. Yu, Y. Sun, D. W. K. Ng, and R. Schober, “Resource
allocation for secure IRS-assisted multiuser MISO systems,” in Proc.
IEEE Globecom Workshops (GC Wkshps), Dec. 2019, pp. 1–6.
[157] H. Yang, Z. Xiong, J. Zhao, D. Niyato, L. Xiao, and Q. Wu, “Deep
reinforcement learning-based intelligent reflecting surface for secure
wireless communications,” IEEE Trans. Wireless Commun., vol. 20,
no. 1, pp. 375–388, Jan. 2021.
[158] L. Lv, Q. Wu, Z. Li, N. Al-Dhahir, and J. Chen, “Secure
two-way communications via intelligent reflecting surfaces,” 2020,
[159] Z. Chu, W. Hao, P. Xiao, and J. Shi, “Intelligent reflecting surface
aided multi-antenna secure transmission,” IEEE Wireless Commun.
Lett., vol. 9, no. 1, pp. 108–112, Jan. 2020.
[160] L. Dong and H.-M. Wang, “Secure MIMO transmission via intelligent
reflecting surface,” IEEE Wireless Commun. Lett., vol. 9, no. 6,
pp. 787–790, Jun. 2020.
[161] Y. Zeng, B. Clerckx, and R. Zhang, “Communications and signals
design for wireless power transmission,” IEEE Trans. Commun.,
vol. 65, no. 5, pp. 2264–2290, May 2017.
[162] Q. Wu, W. Chen, D. W. K. Ng, and R. Schober, “Spectral and energy-
efficient wireless powered IoT networks: NOMA or TDMA?” IEEE
Trans. Veh. Technol., vol. 67, no. 7, pp. 6663–6667, Jul. 2018.
[163] C. Pan et al., “Intelligent reflecting surface enhanced MIMO broadcast-
ing for simultaneous wireless information and power transfer,” IEEE
J. Sel. Areas Commun., vol. 38, no. 8, pp. 1719–1734, Aug. 2020.
[164] Y. Tang, G. Ma, H. Xie, J. Xu, and X. Han, “Joint transmit and reflec-
tive beamforming design for IRS-assisted multiuser MISO SWIPT
systems,” in Proc. IEEE Int. Conf. Commun. (ICC), Jun. 2020, pp. 1–6.
[165] W. Shi, X. Zhou, L. Jia, Y. Wu, F. Shu, and J. Wang, “Enhanced
secure wireless information and power transfer via intelligent
reflecting surface,” 2019, arXiv:1911.01001. [Online]. Available:
[166] Q. Wu, Y. Zeng, and R. Zhang, “Joint trajectory and communication
design for multi-UAV enabled wireless networks,” IEEE Trans. Wire-
less Commun., vol. 17, no. 3, pp. 2109–2121, Mar. 2018.
[167] Y. Zeng, Q. Wu, and R. Zhang, “Accessing from the sky: A tutorial
on UAV communications for 5G and beyond,” Proc. IEEE, vol. 107,
no. 12, pp. 2327–2375, Dec. 2019.
[168] Q. Wu et al., “5G-and-Beyond networks with UAVs: From communi-
cations to sensing and intelligence,” 2020, arXiv:2010.09317. [Online].
[169] H. Lu, Y. Zeng, S. Jin, and R. Zhang, “Enabling panoramic full-angle
reflection via aerial intelligent reflecting surface,” in Proc. IEEE Int.
Conf. Commun. Workshops (ICC Workshops), Jun. 2020, pp. 1–6.
[170] H. Lu, Y. Zeng, S. Jin, and R. Zhang, “Aerial intelligent reflect-
ing surface: Joint placement and passive beamforming design with
3D beam flattening,” 2020, arXiv:2007.13295. [Online]. Available:
[171] S. Li, B. Duo, X. Yuan, Y.-C. Liang, and M. Di Renzo, “Reconfigurable
intelligent surface assisted UAV communication: Joint trajectory design
and passive beamforming,” IEEE Wireless Commun. Lett., vol. 9, no. 5,
pp. 716–720, May 2020.
[172] D. Ma, M. Ding, and M. Hassan, “Enhancing cellular communications
for UAVs via intelligent reflective surface,” in Proc. IEEE Wireless
Commun. Netw. Conf. (WCNC), May 2020, pp. 1–6.
[173] M. Hua, L. Yang, Q. Wu, C. Pan, C. Li, and A. Lee
Swindlehurst, “UAV-assisted intelligent reflecting surface symbi-
otic radio system,” 2020, arXiv:2007.14029. [Online]. Available:
[174] L. Ge, P. Dong, H. Zhang, J.-B. Wang, and X. You, “Joint beamforming
and trajectory optimization for intelligent reflecting surfaces-assisted
UAV communications,” IEEE Access, vol. 8, pp. 78702–78712, 2020.
[175] L. Yang, F. Meng, J. Zhang, M. O. Hasna, and M. D. Renzo, “On the
performance of RIS-assisted dual-hop UAV communication systems,”
IEEE Trans. Veh. Technol., vol. 69, no. 9, pp. 10385–10390, Sep. 2020.

[176] Z. Wei et al., “Sum-rate maximization for IRS-assisted UAV OFDMA
communication systems,” 2020, arXiv:2008.09939. [Online]. Available:
[177] C. You and R. Zhang, “3D trajectory optimization in Rician fading for
UAV-enabled data harvesting,” IEEE Trans. Wireless Commun., vol. 18,
no. 6, pp. 3192–3207, Jun. 2019.
[178] C. You and R. Zhang, “Hybrid offline-online design for UAV-enabled
data harvesting in probabilistic LoS channel,” IEEE Trans. Wireless
Commun., vol. 19, no. 6, pp. 3753–3768, Jun. 2020.
[179] P. Wang, J. Fang, X. Yuan, Z. Chen, and H. Li, “Intelligent reflecting
surface-assisted millimeter wave communications: Joint active and
passive precoding design,” IEEE Trans. Veh. Technol., early access,
Oct. 16, 2020, doi: 10.1109/TVT.2020.3031657.
[180] V. Jamali, A. M. Tulino, G. Fischer, R. Müller, and R. Schober,
“Intelligent reflecting and transmitting surface aided millimeter
wave massive MIMO,” 2019, arXiv:1902.07670. [Online]. Available:
[181] E. Basar and I. Yildirim, “SimRIS channel simulator for reconfig-
urable intelligent surface-empowered communication systems,” 2020,
[182] J. He, H. Wymeersch, L. Kong, O. Silvén, and M. Juntti, “Large
intelligent surface for positioning in millimeter wave MIMO systems,”
in Proc. IEEE 91st Veh. Technol. Conf. (VTC-Spring), May 2020,
pp. 1–5.
[183] E. Basar and I. Yildirim, “Indoor and outdoor physical channel mod-
eling and efficient positioning for reconfigurable intelligent surfaces
in mmwave bands,” 2020, arXiv:2006.02240. [Online]. Available:
[184] Y. Mao, C. You, J. Zhang, K. Huang, and K. B. Letaief, “A survey
on mobile edge computing: The communication perspective,” IEEE
Commun. Surveys Tuts., vol. 19, no. 4, pp. 2322–2358, Aug. 2017.
[185] C. You, K. Huang, H. Chae, and B.-H. Kim, “Energy-efficient resource
allocation for mobile-edge computation offloading,” IEEE Trans. Wire-
less Commun., vol. 16, no. 3, pp. 1397–1411, Mar. 2017.
[186] F. Zhou, C. You, and R. Zhang, “Delay-optimal scheduling for IRS-
aided mobile edge computing,” 2020, arXiv:2011.03690. [Online].
[187] Y. Cao and T. Lv, “Intelligent reflecting surface enhanced resilient
design for MEC offloading over millimeter wave links,” 2019,
[188] T. Bai, C. Pan, Y. Deng, M. Elkashlan, A. Nallanathan, and L. Hanzo,
“Latency minimization for intelligent reflecting surface aided mobile
edge computing,” IEEE J. Sel. Areas Commun., vol. 38, no. 11,
pp. 2666–2682, Nov. 2020.
[189] S. Hua, Y. Zhou, K. Yang, and Y. Shi, “Reconfigurable intelligent sur-
face for green edge inference,” in Proc. IEEE GLOBECOM Workshop,
Dec. 2019, pp. 1–6.
[190] X. Guan, Q. Wu, and R. Zhang, “Joint power control and passive
beamforming in irs-assisted spectrum sharing,” IEEE Commun. Lett.,
vol. 24, no. 7, pp. 1153–1157, Jul. 2020.
[191] J. Yuan, Y.-C. Liang, J. Joung, G. Feng, and E. G. Larsson, “Intelli-
gent reflecting surface-assisted cognitive radio system,” IEEE Trans.
Commun., vol. 69, no. 1, pp. 675–687, Jan. 2021.
[192] D. Xu, X. Yu, Y. Sun, D. W. K. Ng, and R. Schober, “Resource
allocation for IRS-assisted full-duplex cognitive radio systems,” IEEE
Trans. Commun., vol. 68, no. 12, pp. 7376–7394, Dec. 2020.
[193] H. Shen, T. Ding, W. Xu, and C. Zhao, “Beamformig design with
fast convergence for IRS-aided full-duplex communication,” IEEE
Commun. Lett., vol. 24, no. 12, pp. 2849–2853, Dec. 2020.
[194] G. Pan, J. Ye, J. An, and M.-S. Alouini, “When full-duplex
transmission meets intelligent reflecting surface: Opportunities
and challenges,” 2020, arXiv:2005.12561. [Online]. Available:
[195] H. Wymeersch, J. He, B. Denis, A. Clemente, and M. Juntti, “Radio
localization and mapping with reconfigurable intelligent surfaces: Chal-
lenges, opportunities, and research directions,” IEEE Veh. Technol.
Mag., vol. 15, no. 4, pp. 52–61, Dec. 2020.
[196] A. Bourdoux et al., “6G white paper on localization and sensing,” 2020,
[197] J. He, H. Wymeersch, T. Sanguanpuak, O. Silven, and M. Juntti,
“Adaptive beamforming design for mmwave RIS-aided joint localiza-
tion and communication,” in Proc. IEEE Wireless Commun. Netw. Conf.
Workshops (WCNCW), Apr. 2020, pp. 1–6.
[198] C. Chaccour, M. N. Soorki, W. Saad, M. Bennis, and P. Popovski,
“Risk-based optimization of virtual reality over terahertz reconfig-
urable intelligent surfaces,” in Proc. IEEE Int. Conf. Commun. (ICC),
Jun. 2020, pp. 1–6.
[199] W. Hao et al., “Robust design for intelligent reflecting surface assisted
MIMO-OFDMA terahertz communications,” 2020, arXiv:2009.05893.
[200] X. Ma et al., “Joint channel estimation and data rate maximization for
intelligent reflecting surface assisted terahertz MIMO communication
systems,” IEEE Access, vol. 8, pp. 99565–99581, May 2020.
[201] E. Basar, “Reconfigurable intelligent surface-based index modulation:
A new beyond MIMO paradigm for 6G,” IEEE Trans. Commun.,
vol. 68, no. 5, pp. 3187–3196, May 2020.
[202] S. Lin, B. Zheng, G. C. Alexandropoulos, M. Wen, M. Di Renzo,
and F. Chen, “Reconfigurable intelligent surfaces with reflection pat-
tern modulation: Beamforming design and performance analysis,”
IEEE Trans. Wireless Commun., early access, Oct. 8, 2020, doi:
10.1109/TWC.2020.3028198.
[203] T. Ma, Y. Xiao, X. Lei, P. Yang, X. Lei, and O. A. Dobre, “Large
intelligent surface assisted wireless communications with spatial mod-
ulation and antenna selection,” IEEE J. Sel. Areas Commun., vol. 38,
no. 11, pp. 2562–2574, Nov. 2020.
[204] Q. Li, M. Wen, and M. Di Renzo, “Single-RF MIMO: From
spatial modulation to metasurface-based modulation,” 2020,
[205] W. Mei and R. Zhang, “Cooperative beam routing for multi-IRS
aided communication,” IEEE Wireless Commun. Lett., early access,
Oct. 28, 2020, doi: 10.1109/LWC.2020.3034370.
[206] W. Mei and R. Zhang, “Cooperative multi-beam routing for multi-IRS
aided massive MIMO,” 2020, arXiv:2011.02354. [Online]. Available:
Qingqing Wu (Member, IEEE) received the B.Eng.
degree in electronic engineering from the South
China University of Technology in 2012 and the
Ph.D. degree in electronic engineering from Shang-
hai Jiao Tong University (SJTU) in 2016.
From 2016 to 2020, he was a Research Fellow
with the Department of Electrical and Computer
Engineering, National University of Singapore. He
is currently an Assistant Professor with the State
Key Laboratory of Internet of Things for Smart City,
University of Macau. His current research interests
include intelligent reflecting surface (IRS), unmanned aerial vehicle (UAV)
communications, and MIMO transceiver design. He is listed as a World’s
Top 2% Scientist by Stanford University in 2020. He was a recipient of
the IEEE WCSP Best Paper Award in 2015, the Outstanding Ph.D. Thesis
Funding in SJTU in 2016, the Outstanding Ph.D. Thesis Award of the China
Institute of Communications in 2017. He was the Exemplary Editor of IEEE
COMMUNICATIONS LETTERS in 2019 and the Exemplary Reviewer of several
IEEE journals. He serves as an Associate Editor for IEEE COMMUNICATIONS
LETTERS, IEEE OPEN JOURNAL OF COMMUNICATIONS SOCIETY, and
IEEE OPEN JOURNAL OF VEHICULAR TECHNOLOGY. He is also the Lead
Guest Editor of IEEE JOURNAL ON SELECTED AREAS IN COMMUNICA-
TIONS on UAV Communications in 5G and Beyond Networks and a Guest
Editor of IEEE OPEN JOURNAL ON VEHICULAR TECHNOLOGY on 6G Intel-
ligent Communications and IEEE OPEN JOURNAL OF COMMUNICATIONS
SOCIETY on Reconfigurable Intelligent Surface-Based Communications for
6G Wireless Networks. He is also the Workshop Co-Chair of IEEE ICC
2019–2021 Workshop on Integrating UAVs into 5G and Beyond and the
Workshop Co-Chair of IEEE GLOBECOM 2020 Workshop on Reconfigurable
Intelligent Surfaces for Wireless Communication for Beyond 5G. He serves
as the Workshops and Symposia Officer for the Reconfigurable Intelligent
Surfaces Emerging Technology Initiative and a Research Blog Officer for the
Aerial Communications Emerging Technology Initiative.

Shuowen Zhang (Member, IEEE) received the
B.Eng. degree in information engineering from the
Chien-Shiung Wu Honors College, Southeast Uni-
versity, Nanjing, China, in June 2013, and the Ph.D.
degree from the NUS Graduate School for Inte-
grative Sciences and Engineering (NGS), National
University of Singapore, in January 2018, under
the NGS scholarship. She was a Research Fellow
Engineering, National University of Singapore, from
August 2017 to July 2020. She is currently a
Research Assistant Professor with the Department of Electronic and Infor-
mation Engineering, The Hong Kong Polytechnic University. Her current
research interests include intelligent reflecting surface aided communications,
unmanned aerial vehicle (UAV) communications, multiuser multiple-input
multiple-output (MIMO) communications, and optimization methods.
Beixiong Zheng (Member, IEEE) received the B.S.
and Ph.D. degrees from the South China University
of Technology, Guangzhou, China, in 2013 and
2018, respectively.
From 2015 to 2016, he was a Visiting Student
Research Collaborator with Columbia University,
New York, NY, USA. He is currently a Research Fel-
low with the Department of Electrical and Computer
Engineering, National University of Singapore. His
recent research interests include intelligent reflecting
surface (IRS), index modulation (IM), and non-
orthogonal multiple access (NOMA). He was a recipient of the Best Paper
Award from the IEEE International Conference on Computing, Networking
and Communications in 2016, the Best Ph.D. Thesis Award from China
Education Society of Electronics in 2018, and the Outstanding Reviewer of
Physical Communication in 2019.
Changsheng You (Member, IEEE) received the
B.Eng. degree in electronic engineering and infor-
mation science from the University of Science and
Technology of China in 2014 and the Ph.D. degree
in electrical and electronic engineering from The
University of Hong Kong in 2018.
He is currently a Research Fellow with the
Department of Electrical and Computer Engineering,
National University of Singapore. His research inter-
ests include intelligent reflecting surface, unmanned
aerial vehicle (UAV) communications, edge learning,
mobile-edge computing, wireless power transfer, and convex optimization. He
received the IEEE Communications Society Asia-Pacific Region Outstanding
Paper Award in 2019 and the Exemplary Reviewer of IEEE TRANSACTIONS
ON COMMUNICATIONS and IEEE TRANSACTIONS ON WIRELESS COMMU-
NICATIONS. He was a recipient of the Best Ph.D. Thesis Award of The Uni-
versity of Hong Kong in 2019. He is an Editor of IEEE COMMUNICATIONS
LETTERS.
Rui Zhang (Fellow, IEEE) received the B.Eng.
(Hons.) and M.Eng. degrees from the National
University of Singapore, Singapore, and the Ph.D.
degree from the Stanford University, Stanford, CA,
USA, all in electrical engineering.
From 2007 to 2010, he worked with the Institute
for Infocomm Research, ASTAR, Singapore. Since
2010, he has been working with the National Univer-
sity of Singapore, where he is currently a Professor
Engineering. He has published over 200 journal
articles and over 180 conference papers. His current research interests include
unmanned aerial vehicle (UAV) /satellite communications, wireless power
transfer, reconfigurable MIMO, and optimization methods.
Dr. Zhang was a recipient of the 6th IEEE Communications Society Asia-
Pacific Region Best Young Researcher Award in 2011, the Young Researcher
Award of National University of Singapore in 2015, and the Wireless Com-
munications Technical Committee Recognition Award in 2020. He was a co-
recipient of the IEEE Marconi Prize Paper Award in Wireless Communications
in 2015 and 2020, the IEEE Communications Society Asia-Pacific Region
Best Paper Award in 2016, the IEEE Signal Processing Society Best Paper
Award in 2016, the IEEE Communications Society Heinrich Hertz Prize Paper
Award in 2017 and 2020, the IEEE Signal Processing Society Donald G. Fink
Overview Paper Award in 2017, and the IEEE Technical Committee on Green
Communications Computing (TCGCC) Best Journal Paper Award in 2017.
His coauthored paper received the IEEE Signal Processing Society Young
Author Best Paper Award in 2017. He has been listed as a Highly Cited
Researcher by Thomson Reuters/Clarivate Analytics since 2015. He served
for over 30 international conferences as the TPC Co-Chair or an Organizing
Committee Member and as the Guest Editor for three special issues on
IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING and IEEE
JOURNAL ON SELECTED AREAS IN COMMUNICATIONS. He was an elected
member of the IEEE Signal Processing Society SPCOM Technical Committee
from 2012 to 2017 and the SAM Technical Committee from 2013 to 2015.
He served as the Vice Chair for the IEEE Communications Society Asia-
Pacific Board Technical Affairs Committee from 2014 to 2015. He served as
an Editor for IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS from
2012 to 2016, IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS:
Green Communications and Networking Series from 2015 to 2016, IEEE
TRANSACTIONS ON SIGNAL PROCESSING from 2013 to 2017, and IEEE
TRANSACTIONS ON GREEN COMMUNICATIONS AND NETWORKING from
2016 to 2020. He is also an Editor of IEEE TRANSACTIONS ON COMMU-
NICATIONS. He serves as a member for the Steering Committee of IEEE
WIRELESS COMMUNICATIONS LETTERS. He is a Distinguished Lecturer of
the IEEE Signal Processing Society and the IEEE Communications Society.

wu2021.pdf

More Related Content

Similar to wu2021.pdf (20)

Recently uploaded (20)

wu2021.pdf