Reconfigurable_Intelligent_Surfaces_Channel_Characterization_and_Modeling

Reconfigurable Intelligent
Surfaces: Channel
Characterization
and Modeling
This article provides an overview of wireless channel models for reconfigurable
intelligent surfaces, and channel measurements and experiments are presented.
By J IE H UANG , Member IEEE, C HENG -X IANG WANG , Fellow IEEE,
Y INGZHUO S UN , R UI F ENG , Member IEEE, J IALING H UANG , B OLUN G UO,
Z HIMENG Z HONG , Senior Member IEEE, AND T IE J UN C UI , Fellow IEEE
ABSTRACT | Reconfigurable intelligent surfaces (RISs) are sixth-generation (6G) wireless communication systems mainly
2-D metasurfaces, which can intelligently manipulate elec- due to its advantages in tuning wireless signals, thus smartly
tromagnetic waves by low-cost near passive reflecting ele- controlling propagation environments. In this article, we aim
ments. RIS is viewed as a potential key technology for the at addressing channel characterization and modeling issues
of RIS-assisted wireless communication systems. First, the
concept, principle, and potential applications of RIS are given.
Manuscript received 30 September 2021; revised 26 January 2022; accepted An overview of RIS-based channel measurements and experi-
17 April 2022. Date of publication 13 July 2022; date of current version ments is presented by classifying frequency bands, scenarios,
19 September 2022. This work was supported in part by the National Key
Research and Development Program of China under Grant 2018YFB1801101, in system configurations, RIS constructions, experiment pur-
part by the National Natural Science Foundation of China (NSFC) under Grant poses, and channel observations. Then, RIS-based channel
61960206006 and Grant 61901109, in part by the Frontiers Science Center for
Mobile Information Communication and Security, in part by the High Level characteristics are studied, including reflection and trans-
Innovation and Entrepreneurial Research Team Program in Jiangsu, in part by the mission, the Doppler effect and multipath fading mitigation,
High Level Innovation and Entrepreneurial Talent Introduction Program in
Jiangsu, in part by the High Level Innovation and Entrepreneurial Doctor channel reciprocity, channel hardening, rank improvement, far
Introduction Program in Jiangsu under Grant JSSCBS20210082, in part by the field, near field, and so on. RIS-based channel modeling works
Fundamental Research Funds for the Central Universities under Grant
2242022R10067, in part by the Research Fund of National Mobile are investigated, including large-scale path loss models and
Communications Research Laboratory, Southeast University, under Grant small-scale multipath fading models. Finally, future research
2021B02, and in part by the EU H2020 RISE TESTBED2 project under Grant
872172. (Corresponding author: Cheng-Xiang Wang.) directions related to RIS-assisted channels are also discussed.
Jie Huang, Cheng-Xiang Wang, Yingzhuo Sun, and Jialing Huang are with
the National Mobile Communications Research Laboratory, School of Information KEYWORDS | Channel characteristics; channel measurements;
Science and Engineering, Southeast University, Nanjing 210096, China, and also
channel modeling; reconfigurable intelligent surface (RIS);
with the Purple Mountain Laboratories, Nanjing 211111, China (e-mail:
j_huang@seu.edu.cn; chxwang@seu.edu.cn; sunyingzhuo@seu.edu.cn; sixth-generation (6G) wireless communications.
jlhuang@seu.edu.cn).
Rui Feng is with the Purple Mountain Laboratories, Nanjing 211111, China, and I. I N T R O D U C T I O N
also with the School of Information Science and Engineering, Southeast
University, Nanjing 210096, China (e-mail: fengxiurui604@163.com). Although the fifth-generation (5G) wireless communica-
Bolun Guo is with the Research Department of Beijing R&D Subdivision, tion systems are being deployed worldwide, the per-
Wireless Network, Huawei Technologies Company Ltd., Beijing 100085, China
(e-mail: guobolun@huawei.com).
formance metrics of 5G are not fully satisfied, and
Zhimeng Zhong is with the Wireless Network RAN Research Department, the three application scenarios, i.e., enhanced mobile
Huawei Technologies Company Ltd., Shanghai 210206, China (e-mail:
zhongzhimeng@huawei.com).
broadband (eMBB), massive machine-type communica-
Tie Jun Cui is with the State Key Laboratory of Millimeter Wave, School of tions (mMTCs), and ultrareliable and low latency com-
Information Science and Engineering, Southeast University, Nanjing 210096,
munication (URLLC), are not well supported. Hence, 5G
China (e-mail: tjcui@seu.edu.cn).
will not meet all requirements of the future in 2030 and
Digital Object Identifier 10.1109/JPROC.2022.3186087 beyond [1]. Research on the sixth-generation (6G) wireless
0018-9219 © 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
1290 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022
Authorized licensed use limited to: Innovation Science & Economic Development Canada. Downloaded on December 28,2022 at 16:20:26 UTC from IEEE Xplore. Restrictions apply.
Huang et al.: Reconfigurable Intelligent Surfaces: Channel Characterization and Modeling
communication key technologies has been launched,

with global coverage, all spectra, full applications, and
strong network security as the vision and paradigm
shifts [1], [2].
In the previous wireless communication systems,
researchers are mainly focused on designing the trans-
mitter (Tx), receiver (Rx), and corresponding algorithms
to adapt to different propagation environments or wire-
less channels. Recently, reconfigurable intelligent surface
(RIS), also known as intelligent reflecting surface (IRS),
has been proposed and viewed as a potential key technol-
ogy for the 6G wireless communication systems [3]–[21]
since it can intelligently manipulate electromagnetic waves
in the desired manner and smartly control propagation
environments by low-cost and low-power near passive Fig. 1. Illustration of the basic RIS-assisted communication
reflecting elements, thus improving the cost efficiency, scenario.
energy efficiency, and spectral efficiency.

In principle, RIS is a 2-D metasurface consisting of
massive elements (unit cells) with the subwavelength dis-
tance, which is beyond the concept of massive multiple- 1) High-frequency band communications: All the spec-
input multiple-output (MIMO). Each element can be tra will be fully explored in the 6G era, especially
made by positive-intrinsic-negative (PIN) diodes, varac- the high-frequency bands. As the frequency band
tor diodes, liquid crystals, or graphene by implementing increases, the wireless channel shows higher path
different fabrication methods. For example, the popular loss, which limits the communication distance. On the
PIN/varactor diode-based RIS is composed of three layers. one hand, the molecular absorption in high-frequency
The outside layer has massive tunable metallic patches bands aggravates the path loss. On the other hand,
printed on a dielectric substrate to directly manipulate the high-frequency signal will easily form an inac-
incident signals. The middle layer uses a copper panel cessible shadow behind the obstacles due to the
to avoid signal energy leakage. The inside layer is a weak diffraction ability and serious penetration loss.
control circuit board employed to tune the reflection RIS can be well combined with high-frequency band
coefficients of RIS elements in real time, which can be communications by creating an alternative reflecting
enabled via a field-programmable gate array (FPGA) [9], path and employing massive elements to enable high
[13], [22]. Other approaches include the nanonetwork- gains and reconstruct the propagation path toward
based method, where the RIS elements are connected the target user to avoid blockage. Initial works on RIS-
to a communication nanonetwork and switched between assisted millimeter wave (mmWave), terahertz (THz)
different states to maximize the power sensed by the [24]–[28], and optical [29] wireless communications
nanonetwork [23]. are promising. Since the optical wireless communica-
While metasurface has been the frontier of physics in tion is mainly limited to LOS propagation and easily
the last two decades, its applications in wireless com- impaired by obstacles and geometric misalignment
munications are just on the beginning [17]. Neverthe- losses, RIS can help to confront non-LOS (NLOS)
less, its advantages have attracted researchers in wireless propagation by creating new links to improve the
communications to fully explore its potential to meet the coverage range and reduce the outage probability.
requirements of 6G. The basic communication scenario is It showed that placing RIS closer to Rx can reduce
that the line-of-sight (LOS) component between the Tx and outage probability when the jitter of Rx is more
the Rx is heavily blocked by objects such as buildings. In violent, while the middle of Tx and Rx is a better
this case, an RIS can be employed to provide strong LOS choice for other cases [30]. In addition, multibranch
links from Tx to RIS and from RIS to Rx, as shown in RIS can further reduce outage probability compared
Fig. 1. An RIS can be easily deployed on internal walls with single-branch RIS [31].
of indoor environments, external facades of buildings, 2) Space–air–ground–sea integrated networks: While 5G
glasses of windows, and so on. As the distance of the is mainly for terrestrial communications, 6G will be
link via RIS is longer than the direct Tx–Rx link, a large space–air–ground–sea integrated networks to enable
number of RIS elements are essential to combat the higher the vision of global coverage. Satellite, unmanned
path loss. aerial vehicle (UAV), and maritime communications
Starting from the basic RIS-assisted communication sce- will be integrated with terrestrial communications.
nario, many new applications are promising and under The merits of RIS can be well exploited in the space–
development, as illustrated in Fig. 2. The potential appli- air–ground–sea integrated networks to achieve a wide
cation scenarios include the following. range of signal coverage. As an example, for UAV
Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1291
Fig. 2. Potential RIS-assisted application scenarios.
communications, the trajectory mobility of UAV can employing the storage at edge servers to tackle the
create favorable propagation conditions, while the computation latency and guarantee the privacy and
RIS can further improve the system performance by security of big data analysis [12]. Because RIS can
being deployed on the ground to reflect signals from control the environment by adjusting the reflection
ground base stations to UAV [32]. coefficients, a beamforming gain can be achieved.
3) Nonorthogonal multiple access (NOMA): RIS-assisted Using this feature, in the edge network, RIS can
NOMA communication systems can improve the increase the success rate of offloading of edge devices,
spectral efficiency and user connectivity by oppor- thereby improving the performance of the entire
tunistically exploring the users’ different channel network and reducing the end-to-end signal trans-
conditions [16]. Compared to conventional MIMO- mission delay. In addition, the adjustment efficiency
NOMA systems, RIS-NOMA systems can overcome of the RIS will further be improved with the util-
the challenges of the random fluctuation of wireless ity of computing power of the edge server when
channels, blockage, and user mobility in an energy- RIS is deployed near it, thereby enhancing system
efficient way [16]. coverage and transmission rate, as well as reducing
4) Simultaneous wireless information and power transfer edge network transmission and processing delays.
(SWIPT): Since RIS is a passive device, it can act Compared to the conventional amplify-and-forward
as a relay-like function to reflect signals mixed with relay scheme, the RIS-assisted MEC system is shown
information and energy. The drawback of low energy to have higher energy efficiency [33]. In addition,
efficiency at the energy Rxs can be solved with the it can achieve a good tradeoff between computation
aid of RIS. The wireless power transfer range can be efficiency and energy consumption [34].
enlarged, and the number of required energy beams 6) Physical layer security (PLS): As RIS can realize real-
can be reduced, as shown in [16]. time reconfigurable electromagnetic environment and
5) Mobile edge computing (MEC): MEC is essential to dynamic programmable wireless channel, it will fur-
offload the data traffic of wireless networks by ther enhance the signal quality of intended users,
decrease the interference signals from other users, 1) The available RIS-based channel measurements and
and then minimize the uncertainty of the environ- experiments are comprehensively surveyed, unveil-
ment to the utmost extent. Therefore, it is a potential ing the new applications and performance gains of
key technology for PLS application [12]. RIS-assisted wireless communications. Many new
7) Sensing and localization: Radio localization includes RIS-based channel observations have been thoroughly
three parts, i.e., measurements, a reference system, analyzed.
and inference algorithms [35]. RIS can be utilized 2) The various large-scale path loss models are investi-
to create strong and consistent multipath in the gated and compared, pointing out their applicabilities
propagation environment and, thus, can improve the and limitations.
accuracy of localization. For example, RIS can support 3) The available small-scale multipath fading models are
localization in the harsh factory by circumventing LOS reviewed. The features of each modeling approach are
blockage. Especially, RIS-assisted integrated sensing discussed in detail.
and communication (ISAC) has attracted much atten-
The remainder of this article is organized as follows.
tion [36]. In NLOS scenarios, the massive elements
Section II investigates the available RIS-based channel
of RIS can help to combat the high path loss and
measurements and experiments. In Section III, the impor-
improve sensing and communication performance. In
tant channel characteristics brought by RIS are comprehen-
addition, as RIS is usually mounted on fixed objects,
sively surveyed. In Section IV, the state-of-the-art channel
it can benefit mapping and localization in ISAC
modeling methods for RIS-based channels are presented,
systems.
including large-scale path loss models and small-scale mul-
8) Smart radio environment (SRE): Due to its ability to
tipath fading models. In Section V, future research direc-
intelligently control the propagation environments,
tions on RIS-based channel measurements and modeling
RIS can be deployed to benefit our daily life, includ-
are discussed. Conclusions are drawn in Section VI.
ing smart city/home/transportation/factory [9]. In
general, these applications are essential parts
of SREs. II. R I S C H A N N E L M E A S U R E M E N T S
We remark in each application scenario mentioned AND EXPERIMENTS
above that the specific RIS-assisted channel characteristics Recently, many types of RISs have been designed and
need to be considered in channel modeling. For example, fabricated for different purposes related to 6G research.
in high-frequency band communications, large bandwidth Up until now, most of the RIS-assisted channel measure-
will bring frequency nonstationarity. In space–air–ground– ments and experiments in the literature are conducted in
sea integrated networks, temporal nonstationarity needs relatively simple indoor environments at sub-6-GHz and
to be considered due to the movement of satellites and mmWave bands. Although the prototypes and demonstra-
UAVs, which will cause the Doppler effect and multipath tions are not enough, they are essential for the validation,
fast fading. When the size of RIS is large enough, the theoretical analysis, and potential insights.
near-field and spatial nonstationarity effects should also be A key question is how to realize the RIS-assisted wireless
considered. communication systems. In [37], an RIS-assisted wire-
Current research mainly concentrates on the theoretical less communication system prototype was presented. By
analysis of performance improvements provided by RIS. changing the biasing voltage, the capacitance value of
However, the underlying wireless channel, which is crucial the varactor diode was adjusted, and thus, the reflection
to the performance analysis, is not fully investigated. Many coefficient of each element was controlled, while the phase
simplified assumptions in the theoretical analysis need to response was limited to 255◦ . The system was based
be validated by real RIS-based channel measurements and on National Instruments (NI) Universal Software Radio
experiments. Meanwhile, only a few of them take the Peripheral (USRP) Platform. At the Tx side, the USRP
RIS electromagnetic properties and their effects on the generated a single tone incident electromagnetic wave to
channel characteristics into account, such as the hardware programmable RIS as the carrier signal. At the Rx side, the
impairments and mutual coupling between different ele- USRP downmixed the received modulated radio frequency
ments. Moreover, most of the current works consider only (RF) signal and sent the baseband signal to the host com-
the large-scale path loss, while the small-scale multipath puter for synchronization and demodulation processing.
fading is ignored or simplified to independent identically A real-time video streaming experiment was conducted at
distributed (i.i.d.) Rayleigh fading, which may lead to an 4 GHz in a typical indoor environment with the RIS-Rx
overestimated performance gain. To fill these gaps, we distance of 4 m, achieving a 2.048-Mb/s transmission
focus on the detailed investigation of RIS-based channel rate. Experiment results showed that the QPSK constel-
measurements and experiments, channel characteristics, lation diagram after equalization was clear and stable,
and channel models, which are of importance to the and the video stream can be transmitted smoothly and
design, optimization, and performance evaluation of RIS- clearly even in the absence of channel coding. Then, the
assisted wireless communication systems. The novelty and authors improved the design of RIS with a full 360◦ phase
main contributions of this article are given as follows. response [38]. A real-time video streaming experiment
was conducted at 4.25 GHz in an indoor environment The new path loss models were validated. The authors also
with the RIS-Rx distance of 3 m, achieving a 6.144-Mb/s compared the received power of the 27-GHz RIS and a
transmission rate by using 8PSK modulation. Furthermore, metal plate with the same shape and size in the anechoic
the real-time RIS-assisted 2 × 2 MIMO-16-QAM wireless chamber. The results showed that the RIS with a uniform
communication experiment was conducted at 4.25 GHz coding state and a high reflection efficiency had the same
in an indoor environment with the RIS-Rx distance of reflection characteristics as a metal plate of the same shape
1.5 m [39]–[42]. The 16-QAM constellation diagrams of and size.
the two bitstreams recovered by the Rx were clear and In [47], the aforementioned RISs at 4.25 and 27 GHz
dense, achieving a 20-Mb/s transmission rate. Since the were employed to study the channel reciprocity. The RIS
reflection amplitude and phase responses of an element at 4.25 GHz was made with varactor diodes, and its phase
were coupled, a nonlinear modulation technique was intro- can be continuously manipulated when the voltage of the
duced to enable RIS-based QAM modulation under con- RIS controller varied from 0 to 21 V, whereas the RIS at
stant envelope constraints. In addition, the discrete phase 27 GHz was made with PIN diodes, and its phase was only
shift was shown to be negligible when the quantization binary programmable. The uplink and downlink received
step was larger than 8. complex signals were recorded in an indoor environment
As a comparison, two RISs working at 2.3- and under different distances and angles from the Tx and Rx
28.5-GHz bands with 16 × 16 elements were utilized to the RIS, respectively. The RIS was also configured with
for transmission demonstration on the NI USRP plat- identical, gradient, and stripe patterns. Good agreement
form in [43]. Measured in a compact anechoic cham- between the uplink and downlink wireless channels was
ber, a 21.7-dBi antenna gain was obtained by the RIS at obtained, which validated the channel reciprocity of RIS-
2.3 GHz, while a 19.1-dBi antenna gain was achieved at assisted time division duplex (TDD) wireless communica-
28.5 GHz. A 20-m communication link was formed in an tion systems.
indoor environment, and a variety of modulation modes, In [48] and [49], an RIS at 5.8 GHz was employed
including QPSK, 16 QAM, 64 QAM, and so on, were for indoor and outdoor trials based on the NI USRP
employed, achieving 28.34-Mb/s data transmission rate at platform. A simple channel reciprocity test was conducted
2.3 GHz. by connecting two horn antennas to the two-port R&S
In [44], an RIS-assisted wireless communication pro- vector network analyzer (VNA). Two horn antennas were
totype was presented at 3.6 GHz. The RIS was capable placed close in front of the RIS to minimize the effects of
of reflection and transmission. Each element was com- multipath components (MPCs) other than the one reflected
posed of two mirror-symmetric layers. The NI USRP and by the RIS. The measured S12 and S21 parameters rep-
directional double-ridged horn antennas were employed resented the uplink and downlink channels, respectively.
to transmit and receive signals on both sides. One Rx Similar to [47], the amplitudes and phases of the two
antenna was deployed on the same side, while the other channels showed consistency, which validated the channel
Rx antenna was deployed on the opposite side in an indoor reciprocity. In addition, initial indoor and outdoor over-
environment, where the walls were covered with wave- the-air (OTA) tests were carried out. In the indoor test,
absorbing materials to avoid the signals scattered from the about 26-dBi gain was achieved by using RIS. In out-
environment. The results proved that the proposed beam- door tests, about 27- and 14-dBi gains were obtained for
forming scheme can generate directional beams toward the 50- and 500-m links, respectively. The larger distance had
intended users on both sides of the RIS. performance degradation as the wireless channel became
In [45], path loss channel measurements were con- more complicated.
ducted in an anechoic chamber to validate the proposed In [50], RIS-assisted mmWave channel measurements
path loss channel model. The Keysight E8267D signal were conducted in the indoor laboratory and office envi-
generator, N9010A signal analyzer, two horn antennas, and ronments. Different from the conventional case where the
three different RISs were utilized. The distances from Tx RIS is between the Tx and Rx, the RIS was utilized as
to RIS and from Rx to RIS were tuned to compare the Tx. It was based on linearly polarized binary-phase unit
measurement results with the proposed path loss model cells with 180◦ phase-shifting capability. The phase shift
for the near-field broadcasting and far-field beamforming provided by each unit cell through a proper bias of the
cases. In addition, new refined path loss models were PIN diodes was computed using an optimization tool. The
proposed in [46], and corresponding mmWave channel channel sounder consists of a four-port VNA, the RIS, one
measurements were conducted in an anechoic chamber. monopole antenna, and an antenna positioner. Channel
Specifically, two RISs were employed. The first one worked measurements were conducted at 28 GHz with 4-GHz
at 27 GHz, where the second one worked at 33 GHz. Two bandwidth. The results showed that antenna beamforming
measurement systems were built. The first one was in an can largely reduce the path loss and delay spread in the
anechoic chamber to measure the impact of the angle of main beam directions, which led to an improvement of the
reception on the power distribution of the signal reflected link budget.
from the RIS. The second one was in a large empty room to In [51] and [52], experiments were conducted to assess
measure the impacts of the distances of Tx-RIS and RIS-Rx. the RIS-assisted ambient backscatter communications. The
source, the tag, and the reader were all equipped with propagation environment was leveraged as an analog
dipole antennas. The USRP was used on both sides. The multiplexer to localize noncooperative objects using wave
transmitted signal was at 5.4 GHz with a 500-MHz band- fingerprints, even when they were outside the LOS propa-
width. The reader received the combined signals of the gation region.
source, the RIS, and the tag. As the RIS controlled the In [62], measurements were conducted to show that RIS
reflected signal, a hot spot or coherent spots on the tag can enable attackers to perform complex signal manipula-
and the reader can be created to assist the communication tion attacks on the physical layer in real time at low cost.
between the tag and the reader. A laptop was connected to the network to measure the
In [53], the concept of active RIS was proposed to end-to-end connection quality. The proposed environment
break the double fading effect of passive RIS. Different reconfiguration attack scheme was able to largely reduce
from the existing passive RIS that reflects signals passively the throughput.
without amplification, active RIS can actively amplify the In [63], an RIS at 28 GHz was designed to perform as
reflected signals. Experiments were conducted by using an reflectarray and transmitarray on each side. Five functions
active RIS at 2.36 GHz. The incident signal and the pump were achieved, including reflection magnitude, transmis-
input were coupled in a varactor-diode-based reflection- sion magnitude, reflection phase difference, transmission
type amplifier to generate the reflected signal with phase difference, and polarization conversion ratio. The
amplification. channel reciprocity was also validated by connecting two
In [54], an RIS was fabricated at 28 GHz based on horn antennas to a VNA and putting the RIS between them.
a binary-phase tunable metasurface. Experiments were A summary of the abovementioned RIS channel
conducted in near- and far-field cases. In the near-field measurements and experiments is given in Table 1, clas-
case, the RIS was designed to be a typical reflectarray sifying the frequency bands, scenarios, system configura-
antenna and proved to be able to beamform mmWave tions, RIS constructions, experiment results, and channel
signals very efficiently, achieving directivity around 30 dBi observations. As can be seen, initial channel measurements
and reaching angles as high as 60◦ as predicted by the and experiments have unveiled various characteristics of
analytical model. In the far-field case, two horn antennas RIS-assisted wireless communication systems, such as the
were connected to a VNA without LOS propagation. The transmission rate, received signal power, path loss, and
RIS was proven to be able to create an LOS propagation, channel reciprocity. Meanwhile, many new applications
which provided a 25-dBi gain over the received power. are proved to be possible, including RIS-assisted ambi-
Based on the design described in [54], an RIS was ent backscatter communications, active RIS, RIS as a
fabricated at 28 GHz in [55]. Four RISs with unit cells of transmitarray/reflectarray, RIS for physical layer attack,
20 × 20 were assembled to achieve 40 × 40 unit cells. RIS-assisted rich scattering environment, and so on.
Indoor measurements were conducted to demonstrate the
functions of RIS. The RIS was placed on a wall, and two III. R I S C H A N N E L C H A R A C T E R I S T I C S
horn antennas were used at the Tx and Rx sides to emulate As RIS has the ability to tune the amplitude and/or the
the base station and user equipment, respectively. The phase of incident signals, many new channel character-
Tx-RIS and RIS-Rx distances were 5.5 and 2 m, respec- istics will appear, which are substantially different from
tively. To create an RIS-assisted LOS path, the positions of conventional wireless channels. Some of the specific chan-
Tx and Rx were estimated in the beamforming algorithm. nel characteristics have been validated by channel mea-
A 30-dBi gain was achieved in a wide frequency range of surements and experiments, while others are theoretically
3 GHz. In addition, the NI USRP platform was utilized analyzed. Here, we mainly introduce channel characteris-
for QPSK signal transmission, where the four modulated tics of reflection and transmission, the Doppler effect and
symbols were well distinguished. multipath fading mitigation, channel reciprocity, channel
In [56], a prototype of the binary-phase metasurface hardening, rank improvement, near field and far field,
was designed and tested. The simulation showed that the half-power beamwidth (HPBW), and aperture efficiency.
reflection amplitudes had little changes, but the reflection Other new channel characteristics include multipath label-
phases were opposite at 28 GHz when the PIN diode ing [64], which intended to inject a label on propagation
switches. The functions of steerable sum and difference paths through the RISs; OTA equalization [65], which
beams were realized. In addition, the beemsteering per- adjusted the phases of RIS elements according to the
formance was analyzed in [57]–[60], which can achieve incoming signals to maximize the magnitude of the first
precise tracking with an angle mismatch of 0.5◦ –2.5◦ and tap and made the intersymbol interference negligible; elec-
minimizes the beam training overhead through flexible tromagnetic interference [66], which had a nonnegligible
beam refinement. impact on communication performance, especially when
In [61], experiments were conducted for RIS-assisted the size of the RIS grew large; and so on.
wireless communications in rich scattering environments.
Two cases were studied. First, an RIS was deployed to A. Reflection and Transmission
shape the channel impulse response (CIR) to enable higher In traditional wireless communication systems, trans-
achievable communication rates. Second, the RIS-tunable mission signals will have a large penetration loss caused by
Table 1 Summary of RIS Experiments, Channel Measurements, and Channel Observations at Sub-6-GHz and mmWave Bands
objects, which can be up to several tens of dB, especially simultaneously transmitting and reflecting (STAR) mode
for mmWave and THz bands. Thus, the wireless signals [15], [67]–[69]. In the reflecting mode, the users at the
are mainly received by the Rx antenna via reflection, Tx’s same half-space are served, while the users at the
scattering, and diffraction. other half-space are served in the transmission mode. In
For RIS-assisted communications, the RIS can be con- the simultaneous reflection and transmission modes, the
figured in the reflecting mode, the transmitting mode, or users in the full space are able to be served. Preliminary
simulation results showed that STAR-RISs can significantly and electronic components, often conform to the Rayleigh–
reduce the base station power consumption compared to Carson reciprocity theorem [45], [47]. The experiments at
conventional reflecting/transmitting-only RISs [68], [69]. 4.25- and 27-GHz bands demonstrate channel reciprocity,
Up until now, most of the works on RIS-assisted commu- which shows consistency between uplink and downlink
nications mainly focus on the reflecting mode, while the channels. In addition, by using active nonreciprocal cir-
other two modes need further investigation. cuits, performing time-varying controls, and employing
nonlinearities and structural asymmetries, the RIS channel
reciprocity can be broken. An example of reciprocity break-
B. Doppler Effect and Multipath Fading Mitigation ing by RIS was shown in [17]. The channel reciprocity was
When the user equipment is moving, it will cause the also validated in [48] and [63].
Doppler effect and fast multipath fading. As the RIS can An angle-dependent phase shifter model for varactor-
tune the phase of incident signals, it can align the phases based RISs was proposed based on the equivalent circuit
of MPCs arriving from different directions, i.e., the rapid in [76]. The simulation results indicated that the angle-
fluctuations in the received signal strength due to the reciprocity of varactor-based RIS only held under small
Doppler effect can be effectively reduced by using the real- incident angles of both forward and reverse incident elec-
time tunable RIS [70]. The simulation results proved that tromagnetic waves and, thus, limited the channel reci-
the multipath fading effect can be totally eliminated when procity in RIS-assisted TDD systems.
all reflectors in a propagation environment are coated In [77], an RIS-assisted pilot decontamination scheme
with RIS, while even a few RIS can significantly reduce was proposed for TDD massive MIMO system by breaking
the Doppler spread and the deep fading in the received the channel reciprocity. Specifically, the phases of RIS
signal for general propagation environments with several elements were configured differently during the uplink
interacting objects [70]. Matthiesen et al. [71] showed transmission and downlink transmission. In this case, the
that the received power can be maximized without adding intercell interference due to the RIS-assisted link can be
Doppler spread to the system. mitigated, while the non-RIS link interference still existed.
In [72], an RIS-assisted high-mobility vehicular com- If there was only the RIS-assisted link between the Tx and
munication system was considered. The RIS is fixed Rx, by increasing the RIS elements and randomly config-
between the roadside BS and the high-mobility vehicle. uring the phases, the pilot contamination can be mitigated
The predictable information of the vehicle, such as speed, asymptotically. The system performance was then proved
trajectory, and timetable, was utilized to design a subop- to be significantly improved.
timal phase shift set of RIS, which made the RIS capable
of maximizing instantaneous signal-to-noise ratio (SNR), D. Channel Hardening
minimizing Doppler spread, and keeping low delay spread MIMO channels can provide spatial diversity to reduce
simultaneously. SNR variations. In particular, i.i.d. Rayleigh fading chan-
Different from [72], an RIS-assisted high-mobility vehic- nels give rise to channel hardening, where the SNR vari-
ular communication system was studied in [73] and [74], ations average out as the number of antennas increases.
where the RIS was mounted on the top of a high-mobility Björnson and Sanguinetti [78] provided a new definition of
vehicle. An efficient two-stage transmission protocol was channel hardening in RIS-assisted communications, which
proposed to achieve RIS channel estimation and refraction stated that the ratio between the system SNR and the
optimization for data transmission by exploiting the quasi- square of the number of RIS elements tends to be a
static channel between the high-mobility RIS and user, as constant value when the number of RIS elements tends to
well as the LoS-dominant channel between the BS and RIS. infinity. The simulation results revealed that the random
Simulation results showed that the proposed protocol can SNR variations in the case of random phases remain large
achieve the full RIS passive beamforming gain and convert since there is no hardening. When the direct path is
the overall BS-user channel from fast to slow fading to present, the SNR increases very slowly with the number of
enable reliable transmission. RIS elements. Hence, an RIS should be properly configured
to benefit from the channel hardening and SNR gain.
In [79], the channel hardening effect was investigated
C. Channel Reciprocity for the RIS-assisted massive MIMO system. Once the chan-
Channel reciprocity means that the wireless channel nel hardening effect is achieved, the effective channel gain
from the Tx side to the Rx side is the same as the can be replaced by its mean value. The base station can
wireless channel from the Rx side to the Tx side [75]. replace the instantaneous channel gain by its mean value
This phenomenon comes from the symmetry of Maxwell’s for signal detection in the uplink. In the downlink, each
equations with respect to time [47]. In general, when the user can treat the mean value of the effective channel gain
control signals applied to the unit cells remain unchanged, as the true one to detect the desired signal, and no down-
commonly designed and fabricated RISs inherently obey link pilot overhead is required for channel estimation.
the reciprocity theorem. This is because the materials that In [80], the channel hardening effect was theoretically
constitute RISs, such as metal patches, dielectric layers, analyzed, and the performance bound was asymptotically
derived for the RIS-assisted system. The derived ergodic IV. R I S C H A N N E L M O D E L I N G

rates showed that hardware impairments, noise, and inter- A. Large-Scale Path Loss Models
ference from estimation errors and the NLOS path became As the RIS-assisted channel is the cascade of three
negligible as the number of antennas increases. subchannels, the large-scale path loss models are affected
by the electromagnetic properties of RIS and show great
E. Rank Improvement
differences from traditional channels. Several path loss
In point-to-point MIMO communications, large spatial models have been proposed for different scenarios and
multiplexing gains can be achieved at the high SNR region, taken different parameters into account. A comprehen-
while high SNR mainly appears in LOS scenarios where sive survey of the path loss models is given in the
the channel matrix has low rank and, therefore, does not following.
support spatial multiplexing [81]. In this case, RIS can Basar et al. [3] illustrated the mechanism of RIS by
enrich the propagation environment by adding MPCs with revisiting the two-ray path loss model, which consisted
distinctively different spatial angles; thus, a multiplexing of an LOS ray and a ground-reflected ray. The RIS with
gain is achieved even when the direct path has a low rank. Q elements was assumed to be laid on the ground
The simulation results in [81] revealed that the perfor- to assist the communications. The received power was
mance greatly depends on the path loss and deployment approximated as
angles.
The rank deficiency can be analyzed in the form of the 2
λ
eigenvalues of the covariance matrix [79]. The RIS was Pr ≈ (Q + 1)2 Pt (1)
4πdtr
proven to be able to compensate for the rank deficiency
appearing in massive MIMO communications.
where Pt is the transmitted power and dtr is the Tx–Rx
To solve the problem of rank deficiency due to the unob-
distance. The received power is proportional to Q2 and
structed LOS in UAV communications, RIS was utilized to
decay with the inverse of (dtr )2 . By increasing the RIS
enrich the channel in [82]. An RIS placement approach
elements, the path loss is able to be combated.
was proposed to improve the spatial multiplexing gains
Özdogan et al. [88] proposed a far-field path loss model
and maximize the average channel capacity in a prede-
for RIS communications based on physical optics tech-
fined drone corridor. The RIS-assisted channel was shown
nique. It also explained why the surface consists of many
to have similar performance at sub-6-GHz and mmWave
elements that individually act as diffuse scatterers but can
bands.
jointly beamform the signal in a desired direction with a
F. Near Field and Far field certain beamwidth. Specifically, the far-field path loss was
expressed as
In the far field, the total channel gain grows linearly
with the number of antenna elements N , while it grows
(4πd1 d2 )2
as N 2 for RIS [19], [83], [84]. However, the growth PL = πb (2)
Gt Gr (ab)2 cos2 θi sinc λ (sinθs − sinθr )
rate eventually tapers off when entering the near field.
In general, the Rayleigh distance criterion is applied to
distinguish the near- and far-field cases. In near field, the where sinc(x) = sin(x)/x is the sampling function, and a
spherical wavefront effect needs to be modeled, while the and b are the sizes of the rectangular RIS. Gt and Gr are
plane wavefront can be assumed in far field. Meanwhile, the Tx and Rx antenna gains, respectively. d1 and d2 are the
the distances and geometry relationship for each pair of distances from the Tx and Rx to the RIS, respectively. θi , θr ,
Tx/Rx and RIS elements will influence the electromagnetic and θs are the incident angle, the desired reflection angle,
response of the RIS-assisted channel directly. An exact and the observed reflection angle, respectively. When we
near-field channel model is needed to evaluate the system observe at the desired reflection angle, i.e., θs = θr , the
performance whenever the distance between the Tx/Rx path loss model simplifies to
and RIS is comparable with the dimension of RIS [85].
(4πd1 d2 )2
G. HPBW and Aperture Efficiency PL = . (3)
Gt Gr (ab)2 cos2 θi
The HPBW of RIS was investigated in uniform linear
array (ULA) and uniform rectangular array (URA) con- The received power is proportional to the square of
figurations under the far-field condition without consid- the RIS area and 1/(d1 d2 )2 , which was different from
ering path loss and fading [86]. It showed that RIS’s the conjecture in [3] that the received power would be
HPBW is equal to the antenna array’s HPBW at the proportional to 1/(d1 + d2 )2 . That conjecture might hold
same reflect/transmit angle with maximal ratio combining for an infinitely large RIS or in the near field if the RIS is
applied, while it is greater than or equal to the antenna configured to act as a mirror, but probably not in the far-
array’s HPBW when all weights are equal. Another feature field setup.
is that RIS’s HPBW will change with the incident angles. In Björnson and Sanguinetti [84] compared conventional
addition, the aperture efficiency of RIS was studied in [87]. massive MIMO with RIS-assisted communications and
proved that the RIS-assisted communications can never were also validated by channel measurements at 4.25 and
provide a higher information rate or SNR. Numerical 10.5 GHz in an anechoic chamber.
results showed that thousands of reflecting elements are Tang et al. [46] further refined their proposed path loss
needed as the RIS cannot amplify the transmitted signal. models in [45] to make it easier to use. The impact of the
A general RIS-assisted path loss model was proposed radiation patterns of the antennas and RIS elements was
in [45], which was expressed as considered as an angle-dependent loss factor. In the far-
field beamforming case, the path loss model was refined
64π 3 as
PL =
Gt Gr Gdx dy λ2
−2 16π 2 (d1 d2 )2

M N combine
Fn,m Γn,m t
−j2π(rn,m r
+rn,m ) PL =
Gt Gr G(M N dx dy )2 F (θt , φt )F (θr , φr )A2
. (8)
× e λ

m=1 n=1 n,m n,m
r t r r
(4) The near-field beamforming path loss was refined as
16π 2
where dx and dy are the size of each element along the PL =
M N √F
2 . (9)
Gt Gr (dx dy )2
combine
x-axis and the y -axis, respectively. M and N are the t
n,m
r
r r
n,m n,m
number of elements along the x-axis and the y -axis, respec- m=1 n=1
combine
tively. G is the gain of one element, Fn,m denotes the
effect of the normalized power radiation pattern of each In addition, the path loss model for one element was
element on the received signal power, Γn,m is the reflection proposed as
coefficient of each element with amplitude of A and phase
t r 2
of φ, and rn,m and rn,m denote the distances from the 16π 2 rn,m
t r
rn,m
PL = . (10)
Tx and Rx to each element. As we can see, the received Gt Gr (dx dy )2 Fn,m
combine
|Γn,m |2
power is related to the Tx/Rx antenna gains, the gain,
size, number, and specific electromagnetic properties of The refined path loss models were then validated by
elements, and the travel distances. In addition, the general channel measurements at 27- and 33-GHz bands in both
path loss model was simplified to three cases, i.e., far-field anechoic chambers and indoor environments.
beamforming (both Tx and Rx are in the far field), near- In [89], a path loss model for RIS communications
field beamforming (at least one in the near field), and near- was proposed by considering the number of elements, the
field broadcasting (at least one in the near field). physical dimensions of the RIS, and the radiation pattern
In far-field beamforming case, the path loss was and spacing of constituent elements. It was expressed as
expressed as
Q −2
256π 4 An G(θt , φt )G(θr , φr ) jφn
e
3 2
64π (d1 d2 ) PL =
PL =
Gt Gr GM 2 N 2 dx dy λ2 F (θt , φt )F (θr , φr )A2
(5) ηGt Gr
q=1
d1n d2n (11)
where F (θt , φt ) and F (θr , φr ) are the normalized radiation where η is the power efficiency of RIS, which is less than 1
patterns of each element at the incident direction (θt , φt ) for passive RIS. An and φn are the amplitude and phase of
and the reflection direction (θr , φr ). the nth element, respectively. d1n and d2n are the distances
In near-field beamforming case, the distances of Tx and from the Tx and Rx to the nth element, respectively. The
Rx to each element is different, and the corresponding path path loss expression for far-field scenario was also derived.
loss was expressed as The proposed path loss model was then applied to a
reflectarray-type RIS, yielding insights into performance as
a function of the size of the RIS, proximity of the RIS to the
64π 3
PL =
M √ 2 . (6) Tx and Rx, and the criteria used to control the elements.
Gt Gr Gdx dy λ2 A2
N Fn,m
combine
t
rn,m r
rn,m
Garcia et al. [90] unified the opposite behavior of RIS
m=1 n=1
as a scatterer and as a mirror based on the free-space
Green function. The derived path loss model is similar
In the near-field broadcasting case, the path loss was to [89]. The results showed that, depending on its size and
approximated as distance, the RIS can be observed as a 0-D, 1-D, or 2-D
object whose radiated power exposes a dependence on the
16π 2 (d1 + d2 )2 fourth, third, or second power of the distance, respectively.
PL ≈ . (7)
Gt Gr λ2 A2 In [91], a simple path loss model was proposed for
RIS communications based on the general scalar theory
Note that, for each case, the phase shift of each element of diffraction and the Huygens–Fresnel principle. The RIS
was differently designed. The proposed path loss models was modeled as a sheet of electromagnetic material of
negligible thickness. The proposed approach can iden-

tify the conditions under which an RIS of finite size
can be approximated as an anomalous mirror or as a
scatterer.
In [92], a physics-consistent analytical free-space path
loss model for RIS was proposed based on the vector gen-
eralization of Green’s theorem. The path loss model can be
applied to 2-D homogenized metasurfaces. The path loss is
formulated in terms of a computable integral that depends
on the transmission distances, the polarization of the radio
waves, the size of the surface, and the desired surface
transformation. Closed-form expressions were obtained in
far-field and near-field deployments.
In [93], a path loss model was proposed for RIS-assisted
THz communication systems, taking into account the THz
band and RIS characteristics. Specifically, the proposed
path loss model revealed the relationships between the Fig. 4. Calculated path loss with d1 increasing from 140 to 200 m.
RIS specifications, such as size, the number of RIS ele-

ments, element’s size, reflection coefficient, radiation pat-
tern, phase shift, and the transmission parameters, such as respectively. In addition, the continuous and discrete tile
frequency band, Tx–Rx distance, THz specific parameters, response functions were derived for a given transmission
and so on. mode.
In [49], a radar cross section (RCS)-based path loss In [100], a similar path loss model was proposed consis-
model for RIS communications was proposed. The received tent with the path loss model in [45] by using the concept
power was related to the distances from the Tx/Rx to the of RCS, which is more common for scattering problems.
RIS, the angles in the Tx-RIS-Rx triangle, and the effective The maximum RCS of the unit cell can be approximated as
area and the reflection coefficient of each element. In addi- the gain of the unit cell multiplied by its area.
tion, channel measurements were conducted in different To illustrate the far-field path loss, Fig. 3 shows an RIS-
scenarios to validate the proposed model in both near- and assisted communication setup. The Tx and Rx are on the
far-field conditions. focal of an ellipse with 2-D coordinates of (−c, 0) and
In [94] and [95], the RIS was partitioned into tiles, (c, 0), respectively. The Tx–Rx distance is d = 2c = 200 m.
where each tile consisted of several unit cells. The far-field The long semiaxis of the ellipse is a = 200 m. Thus,
tile response function g is obtained based on the concept b = 173 m and d1 + d2 = 400 m. The Rayleigh distance
of RCS. Then, the free-space path loss for RIS-based link is for a 1 m × 1.02 m RIS at 10.5 GHz in [45] is about
given as 140 m. We simulate the path loss when the RIS moves
4π |g|2 from d1 = 140 m to d1 = 200 m. The path loss is shown
PLRIS = PLt PLr (12)
λ2 in Fig. 4. As can be seen, the path loss decreases from
where PLt = ((λ/4πd1 ))2 and PLr = ((λ/4πd2 ))2 are 73 to 69 dB. In case of free-space path loss with a distance
the free-space path losses for Tx-RIS and RIS-Rx links, of d1 + d2 = 400 m, the path loss at 10.5 GHz is 105 dB.
If the 21-dB antenna gains at both sides are excluded, the
multiplied path losses of d1 and d2 would be larger than
the free-space path loss. When the size of the RIS reduces
from 102 × 100 to 51 × 50, the path loss increases about
10 dB, which indicates that a very large size up to ten
thousands of RIS elements are needed to combat the path
loss.
In Fig. 5, we compare the condition of RIS with and
without phase control capability. The general physical
model in [95] and the path loss model in [46] are
consistent. Note that the received power is relative to a
transmitted power of 0 dBm. The simulated frequency
band is 5.4 GHz, M = N = 60, dx = dy = 0.33λ, and
d1 = 20 m. The incident angle is 60◦ , and the reflection
angle is 45◦ . For the programmable RIS, each element is set
to a different phase as the way in [46]. It is clear that the
Fig. 3. RIS-assisted communication setup with RIS moving on an received power is largely improved after the phase control
ellipse. of RIS.
channel and UAV to ground channel in [96] and for

RIS-assisted downlink NOMA cellular system in [97].
In [98], the performance of the RIS-assisted communica-
tion system was analyzed under the Nakagami-m fading
model. In [99], a generalized fading model was applied to
RIS-assisted communication system performance analysis
in the presence of phase noise, where Fox’s H model was
utilized as a general fading distribution with Rayleigh and
Rician as special cases.
A few RIS-assisted small-scale channel models have
been proposed recently, including the phase shift model,
the transmission-mode model, the physics-based channel
model, the beamspace channel model, and the geometry-
based stochastic model (GBSM), which are summarized in
Table 2 and, thereby, overviewed in the following.
Fig. 5. Comparison of RIS-assisted path loss with and without
phase control. 1) Phase Shift Model and Transmission-Mode Model:
The phase shift model is widely used in the literature,
which means the cascaded three subchannels, i.e., the
direct Tx–Rx channel, the Tx-RIS channel, and the RIS-
B. Small-Scale Multipath Fading Models Rx channel. However, the phase shift model has two
drawbacks [94], [95]. The first one is that the ranks of
In the literature, some simple end-to-end fading chan- the Tx-RIS matrix and the RIS-Rx matrix are limited by
nel models were used to assess the RIS-assisted wireless the number of scatterers, which is small compared to the
communication system performance. The Rician fading number of RIS elements. The second one is that the phys-
model was used for RIS-assisted base station to UAV ical properties of RIS, the incident and reflected angles,
Table 2 Summary of RIS Small-Scale Channel Models
and polarization are not considered. Thus, a transmission-

mode model was proposed in [95] by three key ideas.
First, it partitioned the large RIS into tiles and calculated
the tile response function. Then, the low-rank channel
matrices were decomposed. Finally, different transmission
modes were selected for each tile, where each transmission
mode effectively corresponded to a given configuration of
the phase shifts that the unit cells of the tile apply to an
impinging electromagnetic wave.
2) Physics-Based Channel Model: In [101] and [102],

a physics-based channel model was proposed for RIS
communications. It considers the RIS and the scattering
environment as a whole by studying the signal’s multipath
propagation. The joint channel contains a specular com- Fig. 6. Beamspace channel model for RIS-assisted communication
scenario [108].
ponent by the RIS LOS link and a scatter component by
the NLOS direct link. Channel statistics were analyzed by
considering M RIS MPCs and N MPCs for the scattering
environment. The model suggests that the RIS-aided wire-
less channel can be approximated by a Rician distribution. 4) GBSM: The GBSM approach has been widely used
In [103], the Tx-RIS channel was modeled by the sum- in 5G and 6G channel modeling, such as the pro-
mation of impinging plane waves. Then, the joint spatial– posed channel models in [109]–[114]. In [115]–[118],
temporal correlation functions for RIS under both isotropic an open-source SimRIS channel simulator was presented
and nonisotropic scattering were derived. A closed-form for RIS-assisted communications, as shown in Fig. 7.
4-D sinc function is derived to describe the joint spatial– The proposed channel model was applicable to various
temporal correlation in the isotropic scattering environ- indoor and outdoor scenarios and various frequency bands,
ment. For the more practical nonisotropic scattering, a including the mmWave band. The model is similar to
realistic 3-D angular distribution extracted from mmWave the 3GPP cluster channel model. It incorporated the LOS
channel measurements in the urban microscenario was probabilities between terminals, array responses of RISs
utilized. and Tx/Rx units, RIS element gains, realistic path loss
In [104]–[106], a spatially stationary channel model and shadowing models, and environmental characteristics
was proposed for holographic MIMO small-scale fading in in different scenarios and frequency bands. However, it
the far filed case. It modeled the array with a massive is only applied to far-field cases. In [119], the SimRIS
number of antennas in a compact space, which is the channel simulator was adopted to provide performance
case of RIS. The small-scale fading was modeled as a insights for the design of RIS-assisted SREs. The impact
zero-mean, spatially stationary, and correlated Gaussian of positioning/placement of RISs, the number of reflect-
scalar random field satisfying the Helmholtz equation in ing elements, and the tilt/rotation of RISs in both single
the frequency domain, which was equivalent to the scalar and multi-RIS environments were investigated. Results
wave equation in the time domain. revealed that these aspects are crucial to achieving capacity
In [107], an analytical channel model was proposed and reliability enhancements in SREs.
for RIS-assisted free-space optical (FSO) systems based In [120], a GBSM was proposed for RIS communica-
on the Huygens–Fresnel principle. Several effects, such as tions, as shown in Fig. 8, where the three ellipses were
the nonuniform power distribution of Gaussian beams, the used to model Tx-RIS, RIS-Rx, and Tx–Rx channels, respec-
impact of RIS size, the positions of the laser source, the tively. The Tx and Rx were assumed to be equipped with
lens, and the phase shift configuration of the RIS were a single antenna, whereas the RIS was a rectangular array.
considered. The simulation results validated the accuracy As the proposed model assumed that the Tx, RIS, and Rx
of the proposed analytical model. were placed in the x–y plane of a Cartesian coordinate
system, it is a 2-D GBSM. The spatial cross correlation
3) Beamspace Channel Model: In [108], a beamspace functions (CCFs) of subchannels associated with different
channel model was proposed for RIS communications, as RIS elements were considered.
shown in Fig. 6. Different from the phase shift model, In [121], the 2-D GBSM proposed in [120] was extended
the proposed model viewed RIS as a controlled scattering to the 3-D wideband case by adopting three ellipsoids,
cluster reflecting its own MPCs. The model is based on as shown in Fig. 9. The Tx and Rx were equipped with
the extended Saleh Valenzuela (SV) channel model. An multiple antennas. The nonstationary property was consid-
antenna segmentation method was proposed to fit RIS in ered. Statistical properties, such as temporal autocorrela-
the developed far-field model and showed the characteris- tion function (ACF), spatial CCF, and frequency correlation
tics of its MPCs and how they can be controlled. function (FCF), were analyzed.
Fig. 7. GBSM for the RIS-assisted communication scenario [118].
In [122], we proposed a 3-D nonstationary RIS-assisted

MIMO GBSM. The twin-cluster concept was applied. The
channel model supports movements of Tx, Rx, and clus-
ters. The RIS-assisted channel is divided into three sub-
channels, i.e., the Tx-RIS channel, the RIS-Rx channel,
and the Tx–Rx channel. The path loss model in [47] was
incorporated into the channel coefficients. The temporal
ACF and spatial CCF were analyzed.
We further take the practical discrete RIS phase shifts
into account in [123]. The cluster evolution in the space
domain and practical discrete phase shifts are considered.
The statistical properties, including temporal ACF, spatial
CCF, local Doppler spread, and rms delay spread, are
investigated. We find that RIS plays a role in separating the Fig. 9. 3-D ellipsoid GBSM for the RIS-assisted communication
whole channel. We take the temporal ACF as an example scenario [121].
to compare channel statistical properties with and without

RIS, as shown in Fig. 10. It can be seen that the absolute
value of temporal ACF employing RIS is much larger than differences among LOS components, and the strength of
the situation without RIS. In fact, temporal ACF reflects the received signal is always strong at different time
the correlation among different time instances. A larger instances.
value indicates that the channel is more stable with RIS. In [124], an RIS-assisted channel model was proposed
The phenomenon proves that RIS can improve the stability for the THz band. The Tx and Rx were assumed to be
of the channel. This is because RIS can eliminate the phase
Fig. 8. GBSM for the RIS-assisted communication scenario [120]. Fig. 10. Comparison of temporal ACFs with and without RIS.
equipped with a single antenna. The spherical wavefront,

i.e., the phase difference between adjacent RIS elements,
was considered. The path loss model for THz bands was
employed, which relied on the plate scattering paradigm
and molecular absorption. The simulation results showed
that an RIS can significantly improve the energy efficiency
of the MIMO system in near field and beamfocusing.
Other channel models have also been proposed for
different scenarios based on GBSM approach, e.g., out-
door [125], fixed-to-mobile [126], double-RIS [127], high-
altitude platform [128], UAV [129]–[131], and general
MIMO [132] scenarios.
5) Keyhole Channel Model: In the concept of the keyhole

MIMO channel, the channel was considered as a dyad with
one degree of freedom (DoF), where each RIS element
Fig. 11. RMSEE for different SNRs with different numbers of
was a perfect example of the keyhole. In [133], the RIS- transmission modes K 4, 5, and 6.
MIMO channel was modeled as a keyhole MIMO channel.
The direct Tx–Rx channel was ignored. The joint Tx-RIS-Rx
channel was modeled as the sum of the contributions from
transceiver designs. Most of the experiments are con-
each RIS element. Based on that, a channel estimation
ducted in simple environments within a short distance,
framework was proposed using single value decomposition
which is far away from reality. Only path loss channel
(SVD) to separate the cascaded channel links and estimate
measurements were conducted in an anechoic chamber,
each link separately. Compared to the phase shift model of
while the multipath fading channel measurements are
RIS, this method considers the amplitude response of the
missing. The physical properties of RIS will have great
RIS element, making the channel model more accurate.
effects on the incident signals. However, RIS can only tune
With the number of RIS elements increasing, the dimen-
part of the wireless channel, i.e., the MPCs that interact
sion of the Tx-RIS channel coefficient matrix and the RIS-
with RIS. Other MPCs, such as the LOS component and
Rx channel coefficient matrix will increase linearly, and the
reflected MPCs by other scatterers in the environments,
complexity of this channel model also increases.
will have no change. Although channel characteristics have
6) ML-Based Channel Model: The ML-based approach been theoretically analyzed, to further validate the theoret-
has been a new method in channel modeling [134]. Com- ical analysis, RIS experiments and channel measurements
pared to the traditional channel modeling methods, the are indispensable.
most important two advantages are that it can predict In the RIS-assisted communication scenario, the MPCs
channel characteristics by training channel measurement that interact with RIS will be affected by the physical
data and can explore the complex nonlinear relationships properties of RIS. Other MPCs reflected by other scatterers
between channel characteristics, frequency band, scenario, will be independent of RIS. We designed an algorithm to
and RIS-assisted system configurations. In [135], the identify whether an MPC interacts with RIS is carried out
unsupervised machine learning (ML) algorithm, i.e., the by investigating this MPC’s responses under the regulation
expectation-maximization (EM) algorithm, was applied of different RIS transmission modes. Similar to the space-
to RIS-assisted channel modeling. Several aspects were alternating generalized EM (SAGE) algorithm, the max-
considered, including beamforming, spatial channel cor- imum likelihood estimation and the grouped coordinate
relation, phase shift errors, arbitrary fading conditions, ascent method are used to estimate MPCs’ general parame-
and coexistence of direct and RIS channels. With the ters, i.e., the amplitude, delay, angle of arrival (AoA), angle
approximation of a mixture of two Nakagami-m distrib- of departure (AoD), and Doppler frequency. For MPCs that
utions, the EM algorithm was used to estimate the mix- interact with RIS, an extra and novel maximum likelihood
ture weight, mean powers, and fading parameters, which estimation for the incident angle and extra reflect angle at
can characterize the RIS’s composite channels as corre- the RIS is considered in each iteration of the algorithm.
lated/independent over conventional/generalized fading This extra estimation is realized based on these MPCs’
paths in the presence of phase errors. amplitudes under the regulations of different RIS transmis-
sion modes [136].
The root mean square estimation error (RMSEE) for
V. F U T U R E R E S E A R C H D I R E C T I O N S different SNRs and different numbers of RIS transmission
A. RIS Channel Measurements and modes is shown in Fig. 11. It can be seen that RMSEE
Parameters Estimation reduces with an increase in the number of RIS transmission
Current experiments are mainly conducted to demon- modes, meaning that more RIS transmission modes lead to
strate RIS-assisted wireless communication systems and higher algorithm’s accuracy.
B. Pervasive RIS-Assisted Channel Model Table 3 Applications of RIS Channel Research to RIS-Assisted
Communications
None of the mentioned RIS-assisted channel models is
general enough to take all channel characteristics into
account and can be applied to various frequency bands
and scenarios. Different channel models will lead to under-
estimates or overestimates of RIS-assisted communication
system performance evaluation. This situation makes it
difficult to acquire a standardized RIS-assisted channel
model, which is pervasive and can be an add-on compo-
nent. A pervasive RIS-assisted channel model can apply a
unified channel modeling framework, such as the GBSM
method, use the unified expression of CIR, and integrate
different channel characteristics brought by RIS [137].
By adjusting channel model parameters, the pervasive
RIS-assisted channel model is able to adapt to various
frequency bands, scenarios, and RIS configurations. It is or calculating the system capacity. Second, there is mutual
also vital for further performance analysis. The new pro- coupling between a large number of RIS subwavelength
posed circuit-based channel model considering antenna units. Therefore, the entire antenna array can be regarded
size and mutual coupling can be a promising method for as continuous when compactly arranged (such as holo-
RIS-assisted channel modeling [138]. graphic MIMO). This requires starting from Maxwell’s
electromagnetic theory to determine the assumptions and
C. AI-Enabled RIS Channel Modeling applications of the unit from discrete to continuous. An
Artificial intelligence (AI) has now been widely applied accurate and easy-to-handle model that can describe the
to wireless communications and channel modeling, includ- function of the electromagnetic characteristics of the RIS
ing MPC clustering, scenario classification, and channel is needed. Third, analysis and optimization of point-to-
statistical properties prediction [2]. However, AI-enabled point RIS wireless networks are open tasks, especially for
wireless channel modeling is still not fully explored, espe- optimizing the phase of the surface and system capacity
cially AI-enabled RIS channel modeling. Recent works have under different configurations.
shown that a generative adversarial network (GAN) is able
to learn the end-to-end wireless channels in some sim- E. Applications of RIS Channel Research to
ple scenarios [139]–[142]. Different from the traditional RIS-Assisted Communications
channel, the RIS-assisted channel is a cascade of three The RIS channel research will be essential for
subchannels, i.e., Tx–Rx, Tx-RIS, and RIS-Rx channels. RIS-assisted communications since the performance of
The inherent nonideal channel characteristics depend on RIS-assisted communication systems is largely affected by
the specific analysis of the three subchannels. However, as the underlying channel characteristics, which is summa-
RIS is passive, it is not able to obtain the sole responses rized in Table 3.
of Tx-RIS and RIS-Rx channels from traditional channel Take the channel capacity as an example. The chan-
measurements. Currently, there are no open RIS-assisted nel capacity is a strict upper limit of the rate at which
channel datasets, which would limit the application of information can be reliably transmitted on the communi-
AI-enabled RIS channel modeling. cation channel. RIS can improve the channel capacity by
enhancing the received signal power. The ergodic capac-
D. RIS and Electromagnetic Information Theory ity of the RIS-assisted single-input–single-output (SISO)
In principle, the current channel models are for the system was analyzed in [143]. In [144] and [145], the
propagation channel, while the radio channel model is channel capacities of the RIS-assisted MIMO system in the
lacking. The radio channel takes into account the effects unicast scenario and the multicast scenario were studied,
of Tx and Rx antennas. A novel circuit-based radio channel respectively. The alternating optimization algorithm was
model was proposed in [138], which can be applied to RIS- used to jointly optimize the phase shift matrix of RIS
assisted radio channel modeling. Due to its unique char- and covariance matrix of the transmitted signal vector.
acteristics, RIS can be a bridge between electromagnetic The numerical results illustrated that the channel capacity
theory and information theory, building the foundation of increased logarithmically with the number of antennas and
electromagnetic information theory. Here, some challenges the square of the number of reflecting elements, and it
are pointed out. tended to be a constant when both of them went to infinity
First, the essence of RIS amplitude and phase control at a fixed ratio. However, it is inversely proportional to the
is actually realized by the circuit behind it, so the RF link user number. DoF offered by RIS will be limited by the
may need more consideration. Certain circuit parameters number of elements of RIS since it is impractical to create
may be included when obtaining channel characteristics an RIS containing infinite reflecting elements. To solve
this gap, Su et al. [146] proposed irregular RIS, which communication Tx systems [161]–[167]. Based on the
added a selection matrix in the channel transform function. information metasurfaces, new electromagnetic infor-
The current analysis of RIS-assisted communications is mation theory has also been presented by considering
mainly based on the simplification of RIS and the wire- both electromagnetic waves and bit-stream information
less channel, thus enabling closed-form expressions. More simultaneously [168], [169]. The digital coding and
accurate RIS-based channel characteristics and channel programmable metasurfaces, space–time-coding digital
models can help to improve the design and performance metasurfaces, and information metasurfaces can be
analysis of RIS-assisted wireless communication systems. regarded as the advanced RISs. In the future, investigating
the channel characteristics and channel models of such
advanced RISs may produce revolutionary developments
F. Advanced Hardware of RISs in wireless communications.
In most RISs reported above, their main functions
are focused on reconfiguring the beam directions and VI. C O N C L U S I O N
coverages. Recently, some advanced metasurfaces have In this article, we have provided a comprehensive survey
been developed, such as digital coding and programmable on state-of-the-art RIS experiments and channel measure-
metasurfaces [22], [147]–[153], which can not only ments, channel characteristics, and large-scale path loss
manipulate the beam directions and coverages but models and small-scale multipath fading channel models.
also control the number of beams [22], [147], [148], Although various RIS prototypes have been made, the
polarization states [149], and waveforms (e.g., cone experiments and especially channel measurements are still
beams [150] and orbital-angular-momentum vortex in their infancy. RIS-based communications bring several
beams [151]–[153]) in real time and programmable merits to wireless channel characteristics, including simul-
way. The introduction of time-domain coding increases taneous reflection and transmission, the Doppler effect and
an additional DoF to control the frequency spectra of multipath fading mitigation, channel reciprocity, channel
electromagnetic waves [154], [155]. Hence, space–time- hardening, rank improvement, and so on. The large-scale
coding digital metasurfaces can manipulate both spatial path loss models and small-scale multipath fading mod-
waves and their spectra simultaneously and independently els have been compared. In addition, we have pointed
[156], [157]. More importantly, the digital coding out future directions that need further investigation, i.e.,
metasurfaces make a bridge between the digital world and RIS prototypes, experiments, and channel measurements,
the physical world, and hence, they can be used to process pervasive RIS channel models, AI-enabled RIS channel
the digital information directly, resulting in information modeling, RIS and electromagnetic information theory,
metasurfaces [158]–[160]. The information metasurfaces applications of RIS channel research to RIS-assisted com-
have been applied to build new-architecture wireless munications, and advanced hardware of RISs.
REFERENCES
[1] X.-H. You et al., “Towards 6G wireless Tuts., vol. 22, no. 4, pp. 2283–2314, Principles and opportunities,” IEEE Commun.
communication networks: Vision, enabling 4th Quart., 2020. Surveys Tuts., vol. 23, no. 3, pp. 1546–1577,
technologies, and new paradigm shifts,” Sci. China [9] Q. Wu et al., “Intelligent reflecting surface aided 3rd Quart., 2021.
Inf. Sci., vol. 64, no. 1, pp. 1–74, Jan. 2021. wireless communications: A tutorial,” IEEE Trans. [17] J. Y. Dai et al., “Wireless communication based on
[2] C. X. Wang, J. Huang, H. Wang, X. Gao, X. You, Commun., vol. 69, no. 5, pp. 3313–3351, information metasurfaces,” IEEE Trans. Microw.
and Y. Hao, “6G wireless channel measurements May 2021. Theory Techn., vol. 69, no. 3, pp. 1493–1510,
and models: Trends and challenges,” IEEE Veh. [10] X. Yu, V. Jamali, D. Xu, D. W. K. Ng, and Mar. 2021.
Technol. Mag., vol. 15, no. 4, pp. 22–32, R. Schober, “Smart and reconfigurable wireless [18] W. Long, R. Chen, M. Moretti, W. Zhang, and J. Li,
Dec. 2020. communications: From IRS modeling to algorithm “A promising technology for 6G wireless networks:
[3] E. Basar, M. D. Renzo, J. D. Rosny, M. Debbah, design,” IEEE Wireless Commun., vol. 28, no. 6, Intelligent reflecting surface,” J. Commun. Inf.
M. Alouini, and R. Zhang, “Wireless pp. 118–125, Dec. 2021. Netw., vol. 6, no. 1, pp. 1–16, Mar. 2021.
communications through reconfigurable [11] K.-K. Wong, K.-F. Tong, Z. Chu, and Y. Zhang, [19] E. Björnson, Ö. Özdogan, and E. G. Larsson,
intelligent surfaces,” IEEE Access, vol. 7, “A vision to smart radio environment: Surface “Reconfigurable intelligent surfaces: Three myths
pp. 116753–116773, 2019. wave communication superhighways,” IEEE and two critical questions,” IEEE Commun. Mag.,
[4] J. Zhao, “A survey of intelligent reflecting surfaces Wireless Commun., vol. 28, no. 1, pp. 112–119, vol. 58, no. 12, pp. 90–96, Dec. 2020.
(IRSs): Towards 6G wireless communication Feb. 2021. [20] Y.-C. Liang et al., “Reconfigurable intelligent
networks,” 2019, arXiv:1907.04789. [12] X. Yuan et al., “Reconfigurable-intelligent-surface surfaces for smart wireless environments: Channel
[5] M. D. Renzo et al., “Smart radio environments empowered wireless communications: estimation, system design and applications in 6G
empowered by reconfigurable intelligent surfaces: Challenges and opportunities,” IEEE Wireless networks,” Sci. China Inf. Sci., vol. 64, no. 10,
How it works, state of research, and the road Commun., vol. 28, no. 2, pp. 136–143, pp. 1–21, Oct. 2021.
ahead,” IEEE J. Sel. Areas Commun., vol. 38, Apr. 2021. [21] H. Tataria, M. Shafi, A. F. Molisch, M. Dohler,
no. 11, pp. 2450–2525, Jul. 2020. [13] C. Pan et al., “Reconfigurable intelligent surfaces H. Sjöland, and F. Tufvesson, “6G wireless
[6] H. Gacanin and M. D. Renzo, “Wireless 2.0: for 6G systems: Principles, applications, and systems: Vision, requirements, challenges,
Toward an intelligent radio environment research directions,” IEEE Commun. Mag., vol. 59, insights, and opportunities,” Proc. IEEE, vol. 109,
empowered by reconfigurable meta-surfaces and no. 6, pp. 14–20, Jun. 2021. no. 7, pp. 1166–1199, Jul. 2021.
artificial intelligence,” IEEE Veh. Technol. Mag., [14] E. Basar and H. V. Poor, “Present and future of [22] T. J. Cui, M. Q. Qi, X. Wan, J. Zhao, and Q. Cheng,
vol. 15, no. 4, pp. 74–82, Dec. 2020. reconfigurable intelligent surface-empowered “Coding metamaterials, digital metamaterials, and
[7] Q. Wu and R. Zhang, “Towards smart and communications,” IEEE Signal programmable metamaterials,” Light, Sci. Appl.,
reconfigurable environment: Intelligent reflecting Process. Mag., vol. 38, no. 6, pp. 146–152, vol. 3, pp. 1–9, Oct. 2014.
surface aided wireless network,” IEEE Commun. Nov. 2021. [23] C. Liaskos et al., “ABSense: Sensing
Mag., vol. 58, no. 1, pp. 106–112, Nov. 2020. [15] M. Zeng et al., “IRS-empowered wireless electromagnetic waves on metasurfaces via
[8] S. Gong et al., “Toward smart wireless communications: State-of-the-art, key techniques, ambient compilation of full absorption,” in Proc.
communications via intelligent reflecting surfaces: and open issues,” 2021, arXiv:2101.07394. 6th Annu. ACM Int. Conf. Nanosc. Comput.
A contemporary survey,” IEEE Commun. Surveys [16] Y. Liu et al., “Reconfigurable intelligent surfaces: Commun., Sep. 2019, pp. 1–6.
[24] Z. Chen, X. Ma, C. Han, and Q. Wen, “Towards [42] W. Tang et al., “Realization of reconfigurable rich scattering wireless communications: Recent
intelligent reflecting surface empowered 6G intelligent surface-based Alamouti space-time experiments, challenges, and opportunities,” IEEE
terahertz communications: A survey,” China transmission,” in Proc. Int. Conf. Wireless Commun. Commun. Mag., vol. 59, no. 6, pp. 28–34,
Commun., vol. 18, no. 5, pp. 93–119, May 2021. Signal Process. (WCSP), Oct. 2020, pp. 1–6. Jun. 2021.
[25] Z. Chen, B. Ning, C. Han, Z. Tian, and S. Li, [43] L. Dai et al., “Reconfigurable intelligent [62] P. Staat, H. Elders-Boll, M. Heinrichs, C. Zenger,
“Intelligent reflecting surface assisted terahertz surface-based wireless communications: Antenna and C. Paar, “Mirror mirror on the wall: Wireless
communications toward 6G,” IEEE Wireless design, prototyping, and experimental results,” environment reconfiguration attacks based on fast
Commun., vol. 28, no. 6, pp. 110–117, Dec. 2021. IEEE Access, vol. 8, pp. 45913–45923, 2020. software-controlled surfaces,” 2021,
[26] Z. Wan, Z. Gao, F. Gao, M. D. Renzo, and [44] H. Zhang et al., “Intelligent omni-surfaces for arXiv:2107.01709.
M.-S. Alouini, “Terahertz massive MIMO with full-dimensional wireless communications: [63] C.-C. Chung, F.-P. Lai, S.-X. Huang, and Y.-S. Chen,
holographic reconfigurable intelligent surfaces,” Principle, technology, and implementation,” 2021, “Anisotropic metasurface with asymmetric
IEEE Trans. Commun., vol. 69, no. 7, arXiv:2104.12313. propagation of electromagnetic waves and
pp. 4732–4750, Jul. 2021. [45] W. Tang et al., “Wireless communications with enhancements of antenna gain,” IEEE Access,
[27] S.-W. Qu, H. Yi, B. J. Chen, K. B. Ng, and reconfigurable intelligent surface: Path loss vol. 9, pp. 90295–90305, 2021.
C. H. Chan, “Terahertz reflecting and transmitting modeling and experimental measurement,” IEEE [64] C.-J. Wang, S.-L. Shih, C.-K. Wen, and S. Jin,
metasurfaces,” Proc. IEEE, vol. 105, no. 6, Trans. Wireless Commun., vol. 20, no. 1, “Labeling multipath via reconfigurable intelligent
pp. 1166–1184, Jun. 2017. pp. 421–439, Jan. 2021. surface,” 2021, arXiv:2104.12062.
[28] H. Sarieddeen, M.-S. Alouini, and [46] W. Tang et al., “Path loss modeling and [65] E. Arslan, I. Yildirim, F. Kilinc, and E. Basar,
T. Y. Al-Naffouri, “An overview of signal processing measurements for reconfigurable intelligent “Over-the-air equalization with reconfigurable
techniques for terahertz communications,” Proc. surfaces in the millimeter-wave frequency band,” intelligent surfaces,” 2021, arXiv:2106.07996.
IEEE, vol. 109, no. 10, pp. 1628–1665, Oct. 2021. 2021, arXiv:2101.08607. [66] A. de Jesus Torres, L. Sanguinetti, and
[29] H. Abumarshoud, L. Mohjazi, O. A. Dobre, [47] W. Tang et al., “On channel reciprocity in E. Bjornson, “Electromagnetic interference in
M. D. Renzo, M. A. Imran, and H. Haas, “LiFi reconfigurable intelligent surface assisted wireless RIS-aided communications,” IEEE Wireless
through reconfigurable intelligent surfaces: A new networks,” IEEE Wireless Commun., vol. 28, no. 6, Commun. Lett., vol. 11, no. 4, pp. 668–672,
frontier for 6G?” 2021, arXiv:2104.02390. pp. 94–101, Dec. 2021. Apr. 2022, doi: 10.1109/LWC.2021.3124584.
[30] M. Najafi, B. Schmauss, and R. Schober, [48] X. Pei et al., “RIS-aided wireless communications: [67] J. Xu, Y. Liu, X. Mu, and O. A. Dobre, “STAR-RISs:
“Intelligent reflecting surfaces for free space Prototyping, adaptive beamforming, and Simultaneous transmitting and reflecting
optical communication systems,” IEEE Trans. indoor/outdoor field trials,” IEEE Trans. Commun., reconfigurable intelligent surfaces,” IEEE
Commun., vol. 69, no. 9, pp. 6134–6151, vol. 69, no. 12, pp. 8627–8640, Dec. 2021. Commun. Lett., vol. 25, no. 9, pp. 3134–3138,
Sep. 2021. [49] Z. Wang, L. Tan, H. Yin, K. Wang, X. Pei, and Sep. 2021.
[31] H. Wang et al., “Performance analysis of D. Gesbert, “A received power model for [68] Y. Liu et al., “STAR: Simultaneous transmission
multi-branch reconfigurable intelligent reconfigurable intelligent surface and and reflection for 360◦ coverage by intelligent
surfaces-assisted optical wireless communication measurement-based validations,” in Proc. IEEE surfaces,” 2021, arXiv:2103.09104.
system in environment with obstacles,” IEEE 22nd Int. Workshop Signal Process. Adv. Wireless [69] X. Mu, Y. Liu, L. Guo, J. Lin, and R. Schober,
Trans. Veh. Technol., vol. 70, no. 10, Commun. (SPAWC), Sep. 2021, pp. 1–5. “Simultaneously transmitting and reflecting
pp. 9986–10001, Oct. 2021. [50] A. Mudonhi, M. Lotti, A. Clemente, R. D’Errico, (STAR) RIS aided wireless communications,” IEEE
[32] C. You, Z. Kang, Y. Zeng, and R. Zhang, “Enabling and C. Oestges, “RIS-enabled mmWave channel Trans. Wireless Commun., vol. 21, no. 5,
smart reflection in integrated air-ground wireless sounding based on electronically reconfigurable pp. 3083–3098, May 2022, doi: 10.1109/
network: IRS meets UAV,” IEEE Wireless Commun., transmitarrays,” in Proc. 15th Eur. Conf. Antennas TWC.2021.3118225.
vol. 28, no. 6, pp. 138–144, Dec. 2021. Propag. (EuCAP), Mar. 2021, pp. 1–5. [70] E. Basar, “Reconfigurable intelligent surfaces for
[33] C. Huang, A. Zappone, G. C. Alexandropoulos, [51] R. Fara, D.-T. Phan-Huy, P. Ratajczak, A. Ourir, Doppler effect and multipath fading mitigation,”
M. Debbah, and C. Yuen, “Reconfigurable M. D. Renzo, and J. D. Rosny, “Reconfigurable 2019, arXiv:1912.04080.
intelligent surfaces for energy efficiency in intelligent surface-assisted ambient backscatter [71] B. Matthiesen, E. Bjornson, E. D. Carvalho, and
wireless communication,” IEEE Trans. Wireless communications—Experimental assessment,” in P. Popovski, “Intelligent reflecting surface
Commun., vol. 18, no. 8, pp. 4157–4170, Proc. IEEE Int. Conf. Commun. Workshops (ICC operation under predictable receiver mobility: A
Aug. 2019. Workshops), Jun. 2021, pp. 1–7. continuous time propagation model,” IEEE
[34] X. Yu, K. Yu, X. Huang, X. Dang, K. Wang, and [52] R. Fara, P. Ratajczak, D.-T. Phan-Huy, A. Ourir, Wireless Commun. Lett., vol. 10, no. 2,
J. Cai, “Computation efficiency optimization for M. D. Renzo, and J. D. Rosny, “A prototype of pp. 216–220, Feb. 2021.
RIS-assisted millimeter-wave mobile edge reconfigurable intelligent surface with continuous [72] K. Wang, C.-T. Lam, and B. K. Ng, “IRS-aided
computing systems,” IEEE Trans. Commun., early control of the reflection phase,” IEEE Wireless predictable high-mobility vehicular
access, Jun. 13, 2022, doi: 10.1109/TCOMM. Commun., vol. 29, no. 1, pp. 70–77, Feb. 2022. communication with Doppler effect mitigation,” in
2022.3181673. [53] Z. Zhang et al., “Active RIS vs. passive RIS: Which Proc. IEEE 93rd Veh. Technol. Conf. (VTC-Spring),
[35] H. Wymeersch, J. He, B. Denis, A. Clemente, and will prevail in 6G?” 2021, arXiv:2103.15154. Apr. 2021, pp. 1–6.
M. Juntti, “Radio localization and mapping with [54] J.-B. Gros, V. Popov, M. A. Odit, V. Lenets, and [73] Z. Huang, B. Zheng, and R. Zhang, “Transforming
reconfigurable intelligent surfaces: Challenges, G. Lerosey, “A reconfigurable intelligent surface at fading channel from fast to slow: IRS-assisted
opportunities, and research directions,” IEEE Veh. mmWave based on a binary phase tunable high-mobility communication,” in Proc. IEEE Int.
Technol. Mag., vol. 15, no. 4, pp. 52–61, metasurface,” IEEE Open J. Commun. Soc., vol. 2, Conf. Commun., Jun. 2021, pp. 1–6.
Dec. 2020. pp. 1055–1064, 2021. [74] Z. Huang, B. Zheng, and R. Zhang, “Transforming
[36] F. Liu et al., “Integrated sensing and [55] V. Popov et al., “Experimental demonstration of a fading channel from fast to slow: Intelligent
communications: Toward dual-functional wireless mmWave passive access point extender based on a refracting surface aided high-mobility
networks for 6G and beyond,” IEEE J. Sel. Areas binary reconfigurable intelligent surface,” 2021, communication,” IEEE Trans. Wireless Commun.,
Commun., vol. 40, no. 6, pp. 1728–1767, arXiv:2107.02087. early access, Dec. 22, 2021, doi: 10.1109/
Jun. 2022, doi: 10.1109/JSAC.2022.3156632. [56] X. Wan et al., “Reconfigurable sum and difference TWC.2021.3135685.
[37] W. Tang et al., “Wireless communications with beams based on a binary programmable [75] V. S. Asadchy, M. S. Mirmoosa, A. Díaz-Rubio,
programmable metasurface: Transceiver design metasurface,” IEEE Antennas Wireless Propag. S. Fan, and S. A. Tretyakov, “Tutorial on
and experimental results,” China Commun., Lett., vol. 20, no. 3, pp. 381–385, Mar. 2021. electromagnetic nonreciprocity and its origins,”
vol. 16, no. 5, pp. 46–61, May 2019. [57] Y. Li et al., “A novel wireless communication Proc. IEEE, vol. 108, no. 10, pp. 1684–1727,
[38] W. Tang et al., “Programmable metasurface-based system using programmable metasurface Oct. 2020.
RF chain-free 8PSK wireless transmitter,” Electron. operating at 28 GHz,” in Proc. Int. Conf. Microw. [76] W. Chen, L. Bai, W. Tang, S. Jin, W. X. Jiang, and
Lett., vol. 55, no. 7, pp. 417–420, Apr. 2019. Millim. Wave Technol. (ICMMT), Sep. 2020, T. J. Cui, “Angle-dependent phase shifter model
[39] W. Tang et al., “Design and implementation of pp. 1–3. for reconfigurable intelligent surfaces: Does the
MIMO transmission through reconfigurable [58] Y. Li et al., “Beamsteering for 5G mobile angle-reciprocity hold?” IEEE Commun. Lett.,
intelligent surface,” in Proc. IEEE 21st Int. communication using programmable vol. 24, no. 9, pp. 2060–2064, Sep. 2020.
Workshop Signal Process. Adv. Wireless Commun. metasurface,” IEEE Wireless Commun. Lett., [77] X. Luo and H. Zhu, “IRS-based TDD reciprocity
(SPAWC), May 2020, pp. 1–5. vol. 10, no. 7, pp. 1542–1546, Jul. 2021. breaking for pilot decontamination in massive
[40] W. Tang et al., “MIMO transmission through [59] Y. Li et al., “Ka-band programmable metasurface MIMO,” IEEE Wireless Commun. Lett., vol. 10,
reconfigurable intelligent surface: System design, demonstrations in wireless communication no. 1, pp. 102–106, Jan. 2021.
analysis, and implementation,” IEEE J. Sel. Areas scenarios,” in Proc. 15th Eur. Conf. Antennas [78] E. Björnson and L. Sanguinetti, “Rayleigh fading
Commun., vol. 38, no. 11, pp. 2683–2699, Propag. (EuCAP), Mar. 2021, pp. 1–5. modeling and channel hardening for
Nov. 2020. [60] Y. Li et al., “A programmable-metasurface-based reconfigurable intelligent surfaces,” IEEE Wireless
[41] W. Tang et al., “Wireless communications with TDMA fast beam switching communication system Commun. Lett., vol. 10, no. 4, pp. 830–834,
programmable metasurface: New paradigms, at 28 GHz,” IEEE Antennas Wireless Propag. Lett., Apr. 2021.
opportunities, and challenges on transceiver vol. 20, no. 5, pp. 658–662, May 2021. [79] T. V. Chien, H. Q. Ngo, S. Chatzinotas, and
design,” IEEE Wireless Commun., vol. 27, no. 2, [61] G. C. Alexandropoulos, N. Shlezinger, and P. del B. Ottersten, “Reconfigurable intelligent
pp. 180–187, Apr. 2020. Hougne, “Reconfigurable intelligent surfaces for surface-assisted massive MIMO: Favorable
propagation, channel hardening, and rank vol. 69, no. 12, pp. 8551–8568, Dec. 2021. [115] E. Basar and I. Yildirim, “SimRIS channel
deficiency,” IEEE Signal Process. Mag., vol. 39, [97] M. Elhattab, M. A. Arfaoui, C. Assi, and simulator for reconfigurable intelligent
no. 3, pp. 97–104, May 2022. A. Ghrayeb, “RIS-assisted joint transmission in a surface-empowered communication systems,” in
[80] M. Jung, W. Saad, Y. Jang, G. Kong, and S. Choi, two-cell downlink NOMA cellular system,” IEEE J. Proc. IEEE Latin-American Conf. Commun.
“Performance analysis of large intelligent surfaces Sel. Areas Commun., vol. 40, no. 4, (LATINCOM), Nov. 2020, pp. 1–6.
(LISs): Asymptotic data rate and channel pp. 1270–1286, Apr. 2022, doi: 10.1109/ [116] E. Basar, I. Yildirim, and F. Kilinc, “Indoor and
hardening effects,” IEEE Trans. Wireless Commun., JSAC.2022.3143211. outdoor physical channel modeling and efficient
vol. 19, no. 3, pp. 2052–2065, Mar. 2020. [98] Y. Ni, Y. Liu, J. Wang, Q. Wang, H. Zhao, and positioning for reconfigurable intelligent surfaces
[81] O. Ozdogan, E. Bjornson, and E. G. Larsson, H. Zhu, “Performance analysis for RIS-assisted in mmWave bands,” IEEE Trans. Commun., vol. 69,
“Using intelligent reflecting surfaces for rank D2D communication under Nakagami-m fading,” no. 12, pp. 8600–8611, Dec. 2021.
improvement in MIMO communications,” in Proc. IEEE Trans. Veh. Technol., vol. 70, no. 6, [117] I. Yildirim, A. Uyrus, and E. Basar, “Modeling and
IEEE Int. Conf. Acoust., Speech Signal Process. pp. 5865–5879, Jun. 2021. analysis of reconfigurable intelligent surfaces for
(ICASSP), May 2020, pp. 9160–9164. [99] I. Trigui, W. Ajib, W.-P. Zhu, and M. D. R. Renzo, indoor and outdoor applications in future wireless
[82] E. Ozturk, C. K. Anjinappa, F. Erden, I. Guvenc, “Performance evaluation and diversity analysis of networks,” IEEE Trans. Commun., vol. 69, no. 2,
H. Dai, and A. Bhuyan, “Channel rank RIS-assisted communications over generalized pp. 1290–1301, Feb. 2021.
improvement in urban drone corridors using fading channels in the presence of phase noise,” [118] E. Basar and I. Yildirim, “Reconfigurable
passive intelligent reflectors,” 2021, IEEE Open J. Commun. Soc., vol. 3, pp. 593–607, intelligent surfaces for future wireless networks: A
arXiv:2108.02179. 2022. channel modeling perspective,” IEEE Wireless
[83] E. Bjornson and L. Sanguinetti, “Power scaling [100] F. Costa and M. Borgese, “Electromagnetic model Commun., vol. 28, no. 3, pp. 108–114, Jun. 2021.
laws and near-field behaviors of massive MIMO of reflective intelligent surfaces,” IEEE Open J. [119] M. Toumi and A. Aijaz, “System performance
and intelligent reflecting surfaces,” IEEE Open J. Commun. Soc., vol. 2, pp. 1577–1589, 2021. insights into design of RIS-assisted smart radio
Commun. Soc., vol. 1, pp. 1306–1324, 2020. [101] J. Xu and Y. Lin, “A novel channel model for environments for 6G,” in Proc. IEEE Wireless
[84] E. Bjornson and L. Sanguinetti, “Demystifying the reconfigurable intelligent surface-assisted wireless Commun. Netw. Conf. (WCNC), Mar. 2021,
power scaling law of intelligent reflecting surfaces networks,” in Proc. GLOBECOM IEEE Global pp. 1–6.
and metasurfaces,” in Proc. IEEE 8th Int. Workshop Commun. Conf., Dec. 2020, pp. 1–6. [120] J. Dang et al., “A geometry-based stochastic
Comput. Adv. Multi-Sensor Adapt. Process. [102] J. Xu and Y. Liu, “A novel physics-based channel channel model and its application for intelligent
(CAMSAP), Dec. 2019, pp. 549–553. model for reconfigurable intelligent reflecting surface assisted wireless
[85] A. de Jesus Torres, L. Sanguinetti, and surface-assisted multi-user communication communication,” IET Commun., vol. 15, no. 3,
E. Bjornson, “Near- and far-field communications systems,” IEEE Trans. Wireless Commun., vol. 21, pp. 421–434, Feb. 2021.
with large intelligent surfaces,” in Proc. 54th no. 2, pp. 1183–1196, Feb. 2022, doi: 10.1109/ [121] H. Jiang et al., “A general wideband
Asilomar Conf. Signals, Syst., Comput., Nov. 2020, TWC.2021.3102887. non-stationary stochastic channel model for
pp. 1–5. [103] S. Sun and H. Yan, “Small-scale spatial-temporal intelligent reflecting surface-assisted MIMO
[86] H. Han, Y. Liu, and L. Zhang, “On half-power correlation and degrees of freedom for communications,” IEEE Trans. Wireless Commun.,
beamwidth of intelligent reflecting surface,” IEEE reconfigurable intelligent surfaces,” IEEE Wireless vol. 20, no. 8, pp. 5314–5328, Aug. 2021.
Commun. Lett., vol. 25, no. 4, pp. 1333–1337, Commun. Lett., vol. 10, no. 12, pp. 2698–2702, [122] Y. Sun, C.-X. Wang, J. Huang, and J. Wang, “A 3D
Apr. 2021. Dec. 2021. non-stationary channel model for 6G wireless
[87] S.-K. Chou, O. Yurduseven, H. Q. Ngo, and [104] A. Pizzo, T. Marzetta, and L. Sanguinetti, systems employing intelligent reflecting surface,”
M. Matthaiou, “On the aperture efficiency of “Holographic MIMO communications under in Proc. WCSP, Nanjing, China, Oct. 2020,
intelligent reflecting surfaces,” IEEE Wireless spatially-stationary scattering,” in Proc. 54th pp. 19–25.
Commun. Lett., vol. 10, no. 3, pp. 599–603, Asilomar Conf. Signals, Syst., Comput., Nov. 2020, [123] Y. Sun, C.-X. Wang, J. Huang, and J. Wang, “A 3D
Mar. 2021. pp. 702–706. non-stationary channel model for 6G wireless
[88] O. Özdogan, E. Björnson, and E. G. Larsson, [105] A. Pizzo, T. L. Marzetta, and L. Sanguinetti, systems employing intelligent reflecting surfaces
“Intelligent reflecting surfaces: Physics, “Spatially-stationary model for holographic MIMO with practical phase shifts,” IEEE Trans. Cognit.
propagation, and pathloss modeling,” IEEE small-scale fading,” IEEE J. Sel. Areas Commun., Commun. Netw., vol. 7, no. 2, pp. 496–510,
Wireless Commun. Lett., vol. 9, no. 5, pp. 581–585, vol. 38, no. 9, pp. 1964–1979, Sep. 2020. Jun. 2021.
May 2020. [106] A. Pizzo, L. Sanguinetti, and T. L. Marzetta, [124] K. Dovelos, S. D. Assimonis, H. Q. Ngo, B. Bellalta,
[89] S. W. Ellingson, “Path loss in reconfigurable “Fourier plane-wave series expansion for and M. Matthaiou, “Intelligent reflecting surfaces
intelligent surface-enabled channels,” in Proc. holographic MIMO communications,” 2021, at terahertz bands: Channel modeling and
IEEE 32nd Annu. Int. Symp. Pers., Indoor Mobile arXiv:2105.01535. analysis,” in Proc. IEEE Int. Conf. Commun.
Radio Commun. (PIMRC), Sep. 2021, pp. 829–835. [107] H. Ajam, M. Naja, V. Jamali, and R. Schober, Workshops (ICC Workshops), Jun. 2021,
[90] J. C. B. Garcia, A. Sibille, and M. Kamoun, “Channel modeling for IRS-assisted FSO systems,” pp. 1–6.
“Reconfigurable intelligent surfaces: Bridging the in Proc. IEEE Wireless Commun. Netw. Conf. [125] Y. Yang, W. Zhang, J. Sun, and C.-X. Wang, “A 3D
gap between scattering and reflection,” IEEE J. (WCNC), Mar. 2021, pp. 1–7. wideband non-stationary GBSM for RIS-assisted
Sel. Areas Commun., vol. 38, no. 11, [108] M. Alayasra and H. Arslan, “IRS-enabled communications in outdoor scenarios,” in Proc.
pp. 2538–2547, Nov. 2020. beam-space channel,” IEEE Trans. Wireless IEEE 4th Int. Conf. Electron. Commun. Eng.
[91] M. D. Renzo, F. H. Danufane, X. Xi, J. de Rosny, Commun., vol. 21, no. 6, pp. 3822–3835, (ICECE), Dec. 2021, pp. 210–214.
and S. Tretyakov, “Analytical modeling of the Jun. 2022, doi: 10.1109/TWC.2021.3124288. [126] G. Sun, R. He, Z. Ma, B. Ai, and Z. Zhong, “A 3D
path-loss for reconfigurable intelligent [109] S. Wu, C.-X. Wang, E.-H.-M. Aggoune, geometry-based non-stationary MIMO channel
surfaces—Anomalous mirror or scatterer,” in Proc. M. M. Alwakeel, and X. You, “A general 3-D model for RIS-assisted communications,” in Proc.
IEEE 21st Int. Workshop Signal Process. Adv. non-stationary 5G wireless channel model,” IEEE IEEE 94th Veh. Technol. Conf. (VTC-Fall),
Wireless Commun. (SPAWC), May 2020, pp. 1–5. Trans. Commun., vol. 66, no. 7, pp. 3065–3078, Sep. 2021, pp. 1–5.
[92] F. H. Danufane, M. D. Renzo, J. de Rosny, and Jul. 2018. [127] H. Jiang, Z. Zhang, B. Xiong, J. Dang, L. Wu, and
S. Tretyakov, “On the path-loss of reconfigurable [110] J. Bian, C.-X. Wang, X. Gao, X. You, and M. Zhang, J. Zhou, “A 3D stochastic channel model for 6G
intelligent surfaces: An approach based on Green’s “A general 3D non-stationary wireless channel wireless double-IRS cooperatively assisted MIMO
theorem applied to vector fields,” IEEE Trans. model for 5G and beyond,” IEEE Trans. Wireless communications,” in Proc. 13th Int. Conf. Wireless
Commun., vol. 69, no. 8, pp. 5573–5592, Commun., vol. 20, no. 5, pp. 3211–3224, Commun. Signal Process. (WCSP), Oct. 2021,
Aug. 2021. May 2021. pp. 1–5.
[93] A.-A.-A. Boulogeorgos and A. Alexiou, “Pathloss [111] C.-X. Wang, J. Bian, J. Sun, W. Zhang, and [128] Z. Lian, Y. Su, Y. Wang, and L. Jiang,
modeling of reconfigurable intelligent surface M. Zhang, “A survey of 5G channel measurements “A non-stationary 3-D wideband channel model
assisted THz wireless systems,” in Proc. IEEE Int. and models,” IEEE Commun. Surveys Tuts., vol. 20, for intelligent reflecting surface-assisted
Conf. Commun., Jun. 2021, pp. 1–6. no. 4, pp. 3142–3168, 4th Quart., 2018. HAP-MIMO communication systems,” IEEE Trans.
[94] M. Najafi, V. Jamali, R. Schober, and H. V. Poor, [112] J. Huang, C.-X. Wang, H. Chang, J. Sun, and Veh. Technol., vol. 71, no. 2, pp. 1109–1123,
“Physics-based modeling of large intelligent X. Gao, “Multi-frequency multi-scenario Feb. 2022, doi: 10.1109/TVT.2021.3131765.
reflecting surfaces for scalable optimization,” in millimeter wave MIMO channel measurements [129] C. Cao, Z. Lian, Y. Wang, Y. Su, and B. Jin,
Proc. 54th Asilomar Conf. Signals, Syst., Comput., and modeling for B5G wireless communication “A non-stationary geometry-based channel model
Nov. 2020, pp. 559–563. systems,” IEEE J. Sel. Areas Commun., vol. 38, for IRS-assisted UAV-MIMO channels,” IEEE
[95] M. Najafi, V. Jamali, R. Schober, and H. V. Poor, no. 9, pp. 2010–2025, Sep. 2020. Commun. Lett., vol. 25, no. 12, pp. 3760–3764,
“Physics-based modeling and scalable [113] Y. He et al., “A novel 3D non-stationary maritime Dec. 2021.
optimization of large intelligent reflecting wireless channel model,” IEEE Trans. Commun., [130] H. Jiang, R. He, C. Ruan, J. Zhou, and D. Chang,
surfaces,” IEEE Trans. Commun., vol. 69, no. 4, vol. 70, no. 3, pp. 2102–2116, Mar. 2022, doi: “Three-dimensional geometry-based stochastic
pp. 2673–2691, Apr. 2021. 10.1109/TCOMM.2021.3134275. channel modeling for intelligent reflecting
[96] M. Al-Jarrah, A. Al-Dweik, E. Alsusa, Y. Iraqi, and [114] J. Wang et al., “A novel 3D non-stationary GBSM surface-assisted UAV MIMO communications,”
M.-S. Alouini, “On the performance of IRS-assisted for 6G THz ultra-massive MIMO wireless IEEE Wireless Commun. Lett., vol. 10, no. 12,
multi-layer UAV communications with imperfect systems,” IEEE Trans. Veh. Technol., vol. 70, no. 12, pp. 2727–2731, Dec. 2021.
phase compensation,” IEEE Trans. Commun., pp. 12312–12324, Dec. 2021. [131] B. Xiong, Z. Zhang, H. Jiang, J. Zhang, L. Wu, and
J. Dang, “A 3D non-stationary MIMO channel “ChannelGAN: Deep learning-based channel pp. 231–238, 2018.
model for reconfigurable intelligent surface modeling and generating,” IEEE Wireless Commun. [155] J. Y. Dai, J. Zhao, Q. Cheng, and T. J. Cui,
auxiliary UAV-to-ground mmWave Lett., vol. 11, no. 3, pp. 650–654, Mar. 2022, doi: “Independent control of harmonic amplitudes and
communications,” IEEE Trans. Wireless Commun., 10.1109/LWC.2021.3140102. phases via a time-domain digital coding
early access, Jan. 20, 2022, doi: 10.1109/ [143] A.-A.-A. Boulogeorgos and A. Alexiou, “Ergodic metasurface,” Light: Sci. Appl., vol. 7, no. 1,
TWC.2022.3142437. capacity analysis of reconfigurable intelligent pp. 1–10, Dec. 2018.
[132] B. Xiong et al., “A statistical MIMO channel model surface assisted wireless systems,” in Proc. IEEE [156] L. Zhang et al., “Space-time-coding digital
for reconfigurable intelligent surface assisted 3rd 5G World Forum (5GWF), Sep. 2020, metasurfaces,” Nature Commun., vol. 9, no. 1,
wireless communications,” IEEE Trans. Commun., pp. 395–400. pp. 1–11, Dec. 2018.
vol. 70, no. 2, pp. 1360–1375, Feb. 2022, doi: 10. [144] S. Zhang and R. Zhang, “Capacity characterization [157] L. Zhang et al., “Breaking reciprocity with
1109/TCOMM.2021.3129926. for intelligent reflecting surface aided MIMO space-time-coding digital metasurfaces,” Adv.
[133] S. E. Zegrar, L. Afeef, and H. Arslan, communication,” IEEE J. Sel. Areas Commun., Mater., vol. 31, Oct. 2019, Art. no. 1904069.
“Reconfigurable intelligent surface (RIS): vol. 38, no. 8, pp. 1823–1838, Aug. 2020. [158] T.-J. Cui, S. Liu, and L.-L. Li, “Information entropy
Eigenvalue decomposition-based separate channel [145] L. Du, S. Shao, G. Yang, J. Ma, Q. Liang, and of coding metasurface,” Light, Sci. Appl., vol. 5,
estimation,” 2020, arXiv:2010.05623. Y. Tang, “Capacity characterization for no. 11, Nov. 2016, Art. no. e16172.
[134] J. Huang, C. X. Wang, L. Bai, J. Sun, and Y. Yang, reconfigurable intelligent surfaces assisted [159] T. J. Cui, S. Liu, and L. Zhang, “Information
“A big data enabled channel model for 5G wireless multiple-antenna multicast,” IEEE Trans. Wireless metamaterials and metasurfaces,” J. Mater. Chem.
communication systems,” IEEE Trans. Big Data, Commun., vol. 20, no. 10, pp. 6940–6953, C, vol. 5, no. 15, pp. 3644–3668, 2017.
vol. 6, no. 2, pp. 211–222, Jun. 2020. Oct. 2021. [160] T. J. Cui et al., “Information metamaterial
[135] J. D. V. Sanchez, L. Urquiza-Aguiar, [146] R. Su, L. Dai, J. Tan, M. Hao, and R. MacKenzie, systems,” iScience, vol. 23, no. 8, Aug. 2020,
M. C. P. Paredes, and F. J. Lopez-Martinez, “Capacity enhancement for irregular Art. no. 101403.
“Expectation-maximization learning for wireless reconfigurable intelligent surface-aided wireless [161] J. Y. Dai et al., “Wireless communications through
channel modeling of reconfigurable intelligent communications,” in Proc. IEEE Globecom, Taipei, a simplified architecture based on time-domain
surfaces,” IEEE Wireless Commun. Lett., vol. 10, Taiwan, Dec. 2020, pp. 1–6. digital coding metasurface,” Adv. Mater. Technol.,
no. 9, pp. 2051–2055, Sep. 2021. [147] S. Liu et al., “Convolution operations on coding vol. 4, no. 7, Jul. 2019, Art. no. 1900044.
[136] Y. Xu, C.-X. Wang, Z. Zhou, R. Feng, L. Xin, and metasurface to reach flexible and continuous [162] X. Wan et al., “Multichannel direct transmissions
J. Huang, “A novel SAGE-based channel parameter controls of terahertz beams,” Adv. Sci., vol. 3, of near-field information,” Light, Sci. Appl., vol. 8,
estimation scheme for 6G RIS-assisted wireless no. 10, Oct. 2016, Art. no. 1600156. no. 1, pp. 1–8, Dec. 2019.
channel measurements,” in Proc. IEEE ICC, Seoul, [148] L. Li et al., “Electromagnetic reprogrammable [163] T. J. Cui, S. Liu, G. D. Bai, and Q. Ma, “Direct
South Korea, May 2022, pp. 1–6. coding-metasurface holograms,” Nature Commun., transmission of digital message via programmable
[137] C.-X. Wang, Z. Lv, X. Gao, X. You, Y. Hao, and vol. 8, no. 1, pp. 1–7, Aug. 2017. coding metasurface,” Research, vol. 2019,
H. Haas, “Pervasive wireless channel modeling [149] Q. Ma, C. B. Shi, G. D. Bai, T. Y. Chen, A. Noor, pp. 1–12, Jan. 2019.
theory and applications to 6G GBSMs for all and T. J. Cui, “Beam-editing coding metasurfaces [164] J. Y. Dai et al., “Realization of multi-modulation
frequency bands and all scenarios,” IEEE Trans. based on polarization bit and OAM-mode bit,” schemes for wireless communication by
Veh. Technol., early access, Jun. 2, 2022, doi: Adv. Opt. Mater., vol. 5, no. 23, Dec. 2017, time-domain digital coding metasurface,” IEEE
10.1109/TVT.2022.3179695. Art. no. 1700548. Trans. Antennas Propag., vol. 68, no. 3,
[138] J. Gao, C.-X. Wang, J. Huang, and Y. Yang, [150] S. Liu and T. J. Cui, “Flexible controls of terahertz pp. 1618–1627, May 2020.
“A novel circuit-based MIMO channel model waves using coding and programmable [165] H. Zhao, Y. Shuang, M. Wei, T. J. Cui,
considering antenna size and mutual coupling,” in metasurfaces,” IEEE J. Sel. Topics Quantum P. D. Hougne, and L. Li, “Metasurface-assisted
Proc. 13th Int. Conf. Wireless Commun. Signal Electron., vol. 23, no. 4, pp. 1–12, Aug. 2017. massive backscatter wireless communication with
Process. (WCSP), Oct. 2021, pp. 1–5. [151] L. Zhang, S. Liu, L. Li, and T. J. Cui, commodity Wi-Fi signals,” Nature Commun.,
[139] Y. Yang, Y. Li, W. Zhang, F. Qin, P. Zhu, and “Spin-controlled multiple pencil beams and vortex vol. 11, no. 1, p. 3926, Dec. 2020.
C.-X. Wang, beams with different polarizations generated by [166] L. Zhang et al., “A wireless communication scheme
“Generative-adversarial-network-based wireless pancharatnam-berry coding metasurfaces,” ACS based on space- and frequency-division
channel modeling: Challenges and opportunities,” Appl. Mater. Interfaces, vol. 9, pp. 36447–36455, multiplexing using digital metasurfaces,”
IEEE Commun. Mag., vol. 57, no. 3, pp. 22–27, Oct. 2017. Nature Electron., vol. 4, no. 3, pp. 218–227,
Mar. 2019. [152] Y. Shuang, H. Zhao, W. Ji, T. J. Cui, and L. Li, Mar. 2021.
[140] H. Ye, L. Liang, G. Y. Li, and B.-H. F. Juang, “Deep “Programmable high-order OAM-carrying beams [167] M. Z. Chen et al., “Accurate and broadband
learning-based end-to-end wireless for direct-modulation wireless communications,” manipulations of harmonic amplitudes and phases
communication systems with conditional GANs as IEEE J. Emerg. Sel. Topics Circuits Syst., vol. 10, to reach 256 QAM millimeter-wave wireless
unknown channels,” IEEE Trans. Wireless no. 1, pp. 29–37, Mar. 2020. communications by time-domain digital coding
Commun., vol. 19, no. 5, pp. 3133–3143, [153] X. G. Zhang et al., “An optically driven digital metasurface,” Nat. Sci. Rev., vol. 9, no. 1,
May 2020. metasurface for programming electromagnetic Jan. 2022, Art. no. nwab134.
[141] L. Sun, Y. Wang, A. L. Swindlehurst, and X. Tang, functions,” Nature Electron., vol. 3, pp. 165–171, [168] H. T. Wu et al., “Information theory of
“Generative-adversarial-network enabled signal Mar. 2020. metasurfaces,” Nat. Sci. Rev., vol. 7, no. 3,
detection for communication systems with [154] J. Zhao et al., “Programmable time-domain digital pp. 561–571, Mar. 2020.
unknown channel models,” IEEE J. Sel. Areas coding metasurface for nonlinear harmonic [169] H. Wu et al., “Harmonic information transitions of
Commun., vol. 39, no. 1, pp. 47–60, Jan. 2021. manipulation and new wireless communication spatiotemporal metasurfaces,” Light, Sci. Appl.,
[142] H. Xiao, W. Tian, W. Liu, and J. Shen, systems,” Nat. Sci. Rev., vol. 6, no. 2, vol. 9, no. 1, pp. 1–13, Dec. 2020.
ABOUT THE AUTHORS

Jie Huang (Member, IEEE) received the and also a Researcher with the Purple Mountain Laboratories,
B.E. degree in information engineering from Nanjing. His research interests include millimeter wave, ter-
Xidian University, Xi’an, China, in 2013, ahertz (THz), massive multiple-input multiple-output (MIMO),
and the Ph.D. degree in communication reconfigurable intelligent surface channel measurements and
and information systems from Shandong modeling, wireless big data, and sixth-generation (6G) wireless
University, Jinan, China, in 2018. communications.
From October 2018 to October 2020, Dr. Huang received the Best Paper Awards from the International
he was a Postdoctoral Research Associate Symposium on Wireless Personal Multimedia Communications
with the National Mobile Communications (WPMC) 2016 and International Conference on Wireless Commu-
Research Laboratory, Southeast University, Nanjing, China, nications and Signal Processing (WCSP) 2020 and 2021. He has
supported by the National Postdoctoral Program for Innovative also delivered four tutorials in IEEE/CIC International Conference
Talents. From January 2019 to February 2020, he was a Postdoctoral on Communications in China (ICCC) 2021, IEEE International
Research Associate with Durham University, Durham, U.K. He Symposium on Personal, Indoor and Mobile Radio Communications
is currently an Associate Professor with the National Mobile (PIMRC) 2021, IEEE International Conference on Communications
Communications Research Laboratory, Southeast University, (ICC) 2022, and IEEE VTC-Spring 2022.
Cheng-Xiang Wang (Fellow, IEEE) Yingzhuo Sun received the B.E. degree in
received the B.Sc. and M.Eng. degrees in information engineering from Southeast Uni-
communication and information systems versity, Nanjing, China, in 2020, where he is
from Shandong University, Jinan, China, currently pursuing the M.E. degree with the
in 1997 and 2000, respectively, and the National Mobile Communications Research
Ph.D. degree in wireless communications Laboratory, School of Information Science
from Aalborg University, Aalborg, Denmark, and Engineering.
in 2004. He is also an Intern with the Purple Moun-
He was a Research Assistant with the tain Laboratories, Nanjing. His research
Hamburg University of Technology, Hamburg, Germany, from interests include reconfigurable intelligent surface channel mea-
2000 to 2001, a Visiting Researcher with Siemens AG Mobile surements and modeling.
Phones, Munich, Germany, in 2004, and a Research Fellow with Mr. Sun received the Best Paper Award from the International
the University of Agder, Grimstad, Norway, from 2001 to 2005. Conference on Wireless Communications and Signal Processing
He has been with Heriot-Watt University, Edinburgh, U.K., since (WCSP) 2020.
2005, where he was promoted to a Professor in 2011. In 2018, he
joined the National Mobile Communications Research Laboratory,
Southeast University, Nanjing, China, as a Professor. He is also
a part-time Professor with the Purple Mountain Laboratories,
Nanjing. He has authored four books, three book chapters,
and more than 460 papers in refereed journals and conference
proceedings, including 26 Highly Cited Papers. He has also
delivered 23 invited keynote speeches/talks and ten tutorials at
international conferences. His current research interests include
wireless channel measurements and modeling, sixth-generation
(6G) wireless communication networks, and electromagnetic
information theory.
Dr. Wang is currently a member of the Academia Europaea Rui Feng (Member, IEEE) received the
(Academy of Europe) and the European Academy of Sciences and M.Eng. degree in signal and information
Arts and a Fellow of the Royal Society of Edinburgh (FRSE), IEEE, processing from Yantai University, Yantai,
Institution of Engineering and Technology (IET), and the China China, in 2014, and the Ph.D. degree in
Institute of Communications (CIC). He is also an Executive Editorial information and communication engineer-
Committee Member of the IEEE TRANSACTIONS ON WIRELESS ing from Shandong University, Jinan, China,
COMMUNICATIONS. He has served as a TPC member, the TPC in 2018.
chair, and the General Chair for more than 80 international From July 2018 to September 2020, she
conferences. He received the 2020–2022 “AI 2000 Most was a Lecturer with Ludong University,
Influential Scholar Award Honourable Mention” in recognition Yantai. Since October 2020, she has been a Postdoctoral Research
of his outstanding and vibrant contributions in the field of the Associate with the Purple Mountain Laboratories, Nanjing, China,
Internet of Things. He received 14 Best Paper Awards from IEEE and the School of Information Science and Engineering, Southeast
Global Communications Conference (GLOBECOM) 2010, IEEE University, Nanjing. Her research interests include (ultra) massive
International Conference on Communication Technology (ICCT) multiple-input multiple-output (MIMO) channel modeling theory
2011, International Conference on ITS Telecommunications (ITST) and beam domain channel modeling.
2012, IEEE Vehicular Technology Conference (VTC) 2013 Spring, Dr. Feng received the Best Paper Award from the Interna-
International Wireless Communications and Mobile Computing tional Symposium on Wireless Personal Multimedia Communica-
Conference (IWCMC) 2015, IWCMC 2016, IEEE/CIC International tions (WPMC) 2016.
Conference on Communications in China (ICCC) 2016, International
Symposium on Wireless Personal Multimedia Communications
(WPMC) 2016, Wireless and Optical Communication Conference
(WOCC) 2019, IWCMC 2020, International Conference on
Wireless Communications and Signal Processing (WCSP) 2020,
International Conference in Communications, Signal Processing,
and Systems (CSPS) 2021, and WCSP 2021. He has served
as an Editor for nine international journals, including the IEEE
TRANSACTIONS ON WIRELESS COMMUNICATIONS from 2007 to 2009,
the IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY from 2011
to 2017, and the IEEE TRANSACTIONS ON COMMUNICATIONS from
2015 to 2017. He was a Guest Editor of the IEEE JOURNAL ON
SELECTED AREAS IN COMMUNICATIONS, the Special Issue on Jialing Huang received the B.E. degree in
Vehicular Communications and Networks (Lead Guest Editor), communication engineering from Soochow
the Special Issue on Spectrum and Energy Efficient Design of University, Suzhou, China, in 2019. She is
Wireless Communication Networks, and the Special Issue on currently pursuing the M.E. degree with the
Airborne Communication Networks. He was a Guest Editor of the National Mobile Communications Research
IEEE TRANSACTIONS ON BIG DATA and the Special Issue on Wireless Laboratory, School of Information Science
Big Data. He is also a Guest Editor of the IEEE TRANSACTIONS ON and Engineering, Southeast University,
COGNITIVE COMMUNICATIONS AND NETWORKING and the Special Nanjing, China.
Issue on Intelligent Resource Management for 5G and Beyond. She is also an Intern with the Purple
He was an IEEE Communications Society Distinguished Lecturer Mountain Laboratories, Nanjing. Her research interests include
in 2019 and 2020, and a Highly-Cited Researcher recognized by millimeter wave, terahertz (THz), and ray tracing channel
Clarivate Analytics from 2017 to 2020. modeling.
Bolun Guo received the B.E. degree in elec- Tie Jun Cui (Fellow, IEEE) received the
tronic information engineering and the M.S. B.Sc., M.Sc., and Ph.D. degrees in elec-
degree in signal and information process- trical engineering from Xidian University,
ing from the Dalian University of Tech- Xi’an, China, in 1987, 1990, and 1993,
nology, Dalian, China, in 2005 and 2008, respectively.
respectively. In March 1993, he joined the Depart-
He joined Huawei Technologies Company ment of Electromagnetic Engineering, Xid-
Ltd., Beijing, China, in 2008. His responsi- ian University, where he was promoted to
bilities have included International Telecom- an Associate Professor in November 1993.
munication Union (ITU) and 3rd Generation Partnership Project From 1995 to 1997, he was a Research Fellow with the Insti-
(3GPP) standard radio channel model research, fourth-generation tut für Höchstfrequenztechnik und Elektronik (IHE), University of
(4G)/fifth-generation (5G) system-level simulation and evaluation, Karlsruhe, Karlsruhe, Germany. In July 1997, he joined the Center
and spectrum-sharing studies. He has participated in ITU-R WP for Computational Electromagnetics, Department of Electrical and
3J/3K/3M and SG3 Meetings regularly since 2014. He has made Computer Engineering, University of Illinois at Urbana–Champaign,
considerable contributions to the work of several ITU-R recommen- Champaign, IL, USA, first as a Postdoctoral Research Associate
dations. He also participates regularly in ITU-R WP 5D meetings and and then a Research Scientist. In September 2001, he was a
contributes actively to spectrum-sharing studies. In 3GPP, he has Cheung-Kong Professor with the Department of Radio Engineering,
contributed to RAN1 channel model standards in 3GPP TR38901, Southeast University, Nanjing, China. He is currently a Chief Pro-
relating to vehicle-to-vehicle (V2V), non-terrestrial network (NTN), fessor with Southeast University. He is also the Founding Director
Industrial Internet of Things (IIOT), and so on. His current interests of the Institute of Electromagnetic Space, Southeast University. He
include research into 3GPP/ITU radio propagation characteristics is the first author of books Metamaterials: Theory, Design, and
at frequencies that may be considered for systems beyond fifth Applications (Springer, November 2009), Metamaterials: Beyond
generation (5G)/sixth generation (6G), fixed links and satellite com- Crystals, Noncrystals, and Quasicrystals (CRC Press, March 2016),
munication systems, the use of reconfigurable intelligent surface and Information Metamaterials (Cambridge University Press, 2021).
(RIS), and the modeling of radio propagation in spectrum-sharing He has published over 500 peer-reviewed journal papers, which
studies. have been cited more than 46 000 times (H-Factor 108; Google
Scholar), and licensed over 160 patents. His research interests
include metamaterials and computational electromagnetics. He
proposed the concepts of digital coding and programmable meta-
materials, and realized their first prototypes, based on which he
founded the new direction of information metamaterials, bridging
the physical world and digital world.
Dr. Cui is also an Editorial Board Member of National Sci-
ence Review, eLight, PhotoniX, Advanced Optical Materials, Small
Structure, and Advanced Photonics Research. He was awarded a
Research Fellowship from the Alexander von Humboldt Foundation,
Bonn, Germany, in 1995, and a Cheung Kong Professor by the
Ministry of Education, China, in 2001. He received the Young Sci-
entist Award from the International Union of Radio Science in 1999
and the National Science Foundation of China for Distinguished
Young Scholars in 2002. He also received the Natural Science Award
Zhimeng Zhong (Senior Member, IEEE) (first class) from the Ministry of Education, China, in 2011, and the
received the B.E., M.S., and Ph.D. degrees National Natural Science Awards of China (second class, twice) in
from Xi’an Jiaotong University, Xi’an, China, 2014 and 2018. His research has been selected as one of the most
in 2002, 2005, and 2008, respectively, all in exciting peer-reviewed optics research “Optics in 2016” by Optics
electronic engineering. & Photonics News magazine, ten breakthroughs of China Science
He joined Huawei Technologies Company in 2010, and many research highlights in a series of journals. From
Ltd., Shanghai, China, in 2009. He has about 2019 to 2021, he was ranked in the top 1% for the highly cited
11 years of experience in wireless com- papers in the field of physics by Clarivate Web of Science (Highly
munication system research and develop- Cited Researcher). His work has been widely reported by Nature
ment. During these years, he mainly focused on wireless chan- News, MIT Technology Review, Scientific American, Discover, New
nel measurement, channel modeling, and multiple-input–multiple- Scientists, and so on. He is also the Academician of the Chinese
output (MIMO) algorithm and system design. He is currently the Academy of Sciences. He has served as an Associate Editor for
Team Leader of wireless channel research with Huawei. Espe- IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING and a
cially, he was in charge of the 3GPP fifth-generation (5G) channel Guest Editor for Science China—Information Sciences, Science Bul-
model standardization work in Huawei and gave contributions to letin, IEEE Journal on Emerging and Selected Topics in Circuits
the 3GPP millimeter-wave (mmWave) channel model, vehicle-to- and Systems (JETCAS), and Research. He is also the Chief Editor
vehicle (V2V) channel model, and IIOT channel model standard- of Metamaterial Short Books (Cambridge University Press) and
ization. Recently, he also focused on massive MIMO algorithm an Associate Editor of Research. He presented more than 100
design based on some special channel characteristics and machine keynote and plenary talks at academic conferences, symposiums,
learning technologies. or workshops.

Reconfigurable_Intelligent_Surfaces_Channel_Characterization_and_Modeling

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Reconfigurable_Intelligent_Surfaces_Channel_Characterization_and_Modeling

Uploaded by

Copyright:

Available Formats

Reconfigurable Intelligent

1290 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

communication key technologies has been launched,

energy efficiency, and spectral efficiency.

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1291

Fig. 2. Potential RIS-assisted application scenarios.

1292 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1293

1294 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1295

1296 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1297

derived for the RIS-assisted system. The derived ergodic IV. R I S C H A N N E L M O D E L I N G

1298 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

(4) The near-field beamforming path loss was refined as

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1299

negligible thickness. The proposed approach can iden-

RIS specifications, such as size, the number of RIS ele-

1300 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

channel and UAV to ground channel in [96] and for

Table 2 Summary of RIS Small-Scale Channel Models

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1301

and polarization are not considered. Thus, a transmission-

2) Physics-Based Channel Model: In [101] and [102],

1302 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Fig. 7. GBSM for the RIS-assisted communication scenario [118].

In [122], we proposed a 3-D nonstationary RIS-assisted

to compare channel statistical properties with and without

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1303

equipped with a single antenna. The spherical wavefront,

5) Keyhole Channel Model: In the concept of the keyhole

1304 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1305

1306 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1307

1308 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

ABOUT THE AUTHORS

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1309

1310 P ROCEEDINGS OF THE IEEE | Vol. 110, No. 9, September 2022

Vol. 110, No. 9, September 2022 | P ROCEEDINGS OF THE IEEE 1311

You might also like