This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2021.3082214, IEEE
Communications Letters
1

STAR-RISs: Simultaneous Transmitting and

Reflecting Reconfigurable Intelligent Surfaces
Jiaqi Xu Graduate Student Member, IEEE, Yuanwei Liu Senior Member, IEEE, Xidong Mu Graduate Student
Member, IEEE, and Octavia A. Dobre Fellow, IEEE.

Abstract—In this letter, simultaneous transmitting and reflect- ting and reflecting RISs (STAR-RISs)1 . From a manufacturing
ing reconfigurable intelligent surfaces (STAR-RISs) are studied. perspective, the proposed new RISs can be implemented
Compared with the conventional reflecting-only RISs, the cover- using various metasurface-based designs. For example, NTT
age of STAR-RISs is extended to 360 degrees via simultaneous
transmission and reflection. A general hardware model for STAR- DOCOMO, INC. recently announced a dynamic metasurface
RISs is presented. Then, channel models are proposed for the achieving dynamic manipulation of both reflection and pene-
near-field and the far-field scenarios, based on which the diversity tration in a highly transparent package [9]. Other successful
gain of the STAR-RISs is analyzed and compared with that of the implementations achieving manipulation of transmission and
conventional RISs. Numerical simulations are provided to verify reflection are also reported, including graphene-based dynamic
analytical results and to demonstrate that full diversity order can
be achieved on both sides of the STAR-RIS. metasurfaces [10] and multi-layer metasufaces [11].
However, existing research contributions [9–11] mainly fo-
Index Terms—Channel modelling, electromagnetics, perfor- cused on the designing of metasurface prototypes or studying
mance analysis, reconfigurable intelligent surfaces (RISs), simul-
taneous transmission and reflection. particular physical properties of the designed surface. There
is a lack of communication models for the STAR-RISs that
are both physically-compliant and mathematically-tractable.
To facilitate the research on STAR-RISs in the field of wireless
I. I NTRODUCTION communications, we separately present a general hardware
model and two channel models for the near-field region and the
Recently, the concept of smart radio environments and
far-field region. Based on the field equivalence principle [3],
reconfigurable intelligent surfaces (RISs) has received heated
the proposed hardware model characterizes each STAR-RIS
discussion in the research community. It is envisioned that
element by the transmission and reflection coefficients. The
the RISs can assist wireless communication networks by
channel model gives the closed-form formulation of channel
intelligently controlling part of the radio environment [1–3].
gains in terms of the transmission and reflection coefficients,
Specifically, the RIS is a two-dimensional structure that con-
as well as the geometric settings of the system. Based on
sists of a large number of low-cost elements. These elements
the proposed models, we analyze and compare the diversity
are of sub-wavelength sizes and are reconfigurable in terms of
gain of the STAR-RISs and conventional RISs. Our numerical
their electromagnetic responses. There are diverse applications
results confirm that the STAR-RIS can significantly extend
of introducing RISs into wireless networks, including but
the coverage and achieve full diversity order on both sides
not limited to energy efficiency enhancement [4], multi-user
of the surface, yielding a superior performance gain over
support [5], visible light communication [6], and robotic
conventional RISs.
communication [7]. Moreover, the potentials of the RISs can
be much more than just passively reflecting signals. In [8], a
hybrid RIS was proposed, where the sensing capability was II. A G ENERAL H ARDWARE M ODEL
integrated into the RIS to facilitate tasks such as channel
estimation and localization. As shown in Fig. 1, a schematic representation of the
structure of STAR-RISs is presented. Based on the field equiv-
Nevertheless, we notice that existing research contri-
alence principle [3], as the STAR-RIS elements are excited
butions [3–8] mainly considered the employment of the
by the incident signal, the transmitted and reflected signals
reflecting-only RISs, which brings extra topological con-
can be equivalently treated as waves radiated from the time-
straints to the wireless system. Specifically, to receive signals
varying surface equivalent electric currents Jp and magnetic
from the reflecting-only RISs, users have to be on the same
currents Jb (also referred to as the bound currents). Within
side as the transmitter. To mitigate this drawback and to facili-
each element, the strengths and distribution of these surface
tate more flexible system designs of the RIS-assisted network,
equivalent currents are determined by the incident narrowband
in this letter, we propose the concept of simultaneous transmit-
signal sm as well as the local surface averaged electric and
magnetic impedances Ym and Zm . Assume that the STAR-
J. Xu and Y. Liu are with the School of Electronic Engineering and RIS produces both transmitted and reflected signals with the
Computer Science, Queen Mary University of London, London E1 4NS, UK.
(email:{jiaqi.xu, yuanwei.liu}@qmul.ac.uk).
X. Mu is with School of Artificial Intelligence, Beijing University of Posts 1 In prior works, “refracting” is also used to describe the “transmitting”
and Telecommunications, Beijing, China. (email: muxidong@bupt.edu.cn). of wireless signal at the RIS. Since they have the same meaning from a
O. A. Dobre is with Memorial University, Newfoundland and Labrador, physical perspective, we use the term “transmitting” in the rest of the article
Canada. (email: odobre@mun.ca). for consistency.

1089-7798 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: East Carolina University. Downloaded on June 15,2021 at 22:08:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2021.3082214, IEEE
Communications Letters
2

same polarization. At the mth element, these signals can be

expressed as:
Reflected Signal
sTm = Tm sm , sR
m = Rm sm , (1) mth
elem mitted Signal
ent

where Tm and Rm are the transmission and reflection coeffi-

STAR-RIS
cients of the mth element, respectively. According to the law
Equivalent
of energy conservation, for passive STAR-RIS elements, the Magnetic Circuit

following constraint on the local transmission and reflection

coefficients must be satisfied: Equivalent
Electrical Circuit Incident Signal

2 2
|Tm | + |Rm | ≤ 1. (2)
According to electromagnetic theory, the phase delays of Fig. 1. Schematic illustration of the STAR-RIS.
both the transmitted and reflected field are related to Ym
A. Far-field channel model
and Zm . In Fig. 1, the reconfigurability of the element is
reflected in the change of the surface impedances, since the In this case, Rxs are located in the far-field region of the
transmission and reflection coefficients of the mth element is STAR-RIS. We denote the channel between Tx and the STAR-
related to the surface impedances as: Tm = 2−η 0 Ym RIS by h = (h1 , · · · , hM )T , where hm is the channel between
2+η0 Ym − Rm ,
2(η 2 Y −Z )
χ T
Tx and the mth element. In addition, let rχ = (r1χ , · · · , rM )
and Rm = − (2+η0 Y0mm m
)(2η0 +Zm ) , where η0 is the impedance denote the channel between STAR-RIS and receiver χ. Since
of free space [12]. From the perspective of metasurface both Rxs are located in the far-field region of the STAR-
design, supporting the magnetic currents is the key to achieve RIS, the ray tracing technique can be adopted by studying
independent control of both the transmitted and reflected a number of M geometrical rays, each corresponding to a
signals. According to [13], single-layered RISs with non- multipath signal propagating through an element. This leads
magnetic elements can only produce identical radiation on to the channel model as follows:
different sides, which is referred to as the symmetry limitation. q χ
q
χ jφχ
By introducing the equivalent surface electric and magnetic g χ = (rχ )H diag( β1χ ejφ1 , · · · , βM e M )h, (4)
H
currents into the model, the proposed hardware model is able where a denotes the conjugate transpose of a complex vector
to independently characterize the transmission and reflection a. For convenience, we denote diag(T1 , T2 , · · · , TM ) by ΦT
of each element. To facilitate the design of the STAR-RISs in and diag(R1 , R2 , · · · , RM ) by ΦR . In addition, in the far-
wireless communication systems, we rewrite these narrowband field channel model, rχ and h can be written in the form of
frequency-flat coefficients in the form of their amplitudes and path loss (large-scale fading) multiplied with the normalized
phase shifts as follows: small-scale fading channels. The large-scale fading depends on
q T
q R
the distance between Tx, the STAR-RIS, and Rxs, while the
Tm = βm T ejφm , R = βmR ejφm , (3)
m small-scale fading depends on the scattering environment. Let
where βm T R
, βm ∈ [0, 1] are real-valued coefficients satisfying dχ denote the distances between the STAR-RIS and receiver
βm + βm ≤ 1, and φTm , φR
T R χ, and d0 denote the distance between Tx and the STAR-RIS.
m ∈ [0, 2π) are the phase shifts
introduced by element m for the transmitted and reflected The channel gains can be expressed as:
1
signals, ∀m ∈ {1, 2, · · · , M }. Note that different from the |g χ | = χ αχ rχ )H Φχ h|,
|(e e (5)
T R (d ) (d0 )α0
amplitudes (βm and βm ) which are coupled due to the law of
energy conservation, the phase shifts (i,e, φTm and φR where erχ and he are the corresponding small-scale fading com-
m ) can be
independently adjusted. ponents. In addition, αχ and α0 are the path loss coefficients
of the corresponding channels.
III. C HANNEL M ODELS
Based on the proposed hardware model, in this section, B. Near-field channel model
we study the communication channels of the STAR-RISs. As In scenarios where Rxs are located within the near-field
shown in Fig. 2, a narrowband STAR-RIS-assisted wireless region of the STAR-RIS, the conventional ray tracing tech-
network is considered, where the transmitter (Tx) and receivers nique based channel models can not be adopted. Based on
(Rxs) are equipped with single antenna and the STAR-RIS the Huygens-Fresnel principle [3], A. Fresnel and Kirchhoff
consists of M reconfigurable elements. As an initial analysis arrived at the analytical result which is known as the Fresnel-
to demonstrate the extended coverage of the STAR-RIS, we Kirchhoff diffraction formula. As illustrated in Fig. 2, the
consider a two-user setup. In particular, Tx sends information electromagnetic signal at Rx can be calculated by summing
to the two Rxs employing orthogonal frequency bands of equal up the contribution of every elements on the wavefront. The
sizes. Suppose that receiver T is located on the opposite side wavefront is chosen as the plane where the STAR-RIS is
of Tx, with respect to the STAR-RIS, and thus, it can only located.
receives the transmitted signals of the STAR-RIS. Reicerver 1
ZZ χ
e2jπdm /λ
R and Tx are located on the same side of STAR-RIS. Let hχ gχ = U χ (Q)F (θχ ) dΣ, (6)
jλ (Σ) dχ
denote the direct link between Tx and receiver T or receiver
R, where the notation χ ∈ {T, R} is an indicator representing where j is the imaginary unit, (Σ) denotes a closed surface
“T ” or “R”. Let g χ denote the effective channel between Tx (wavefront) which contains the RIS elements and the scatters
and receiver χ through STAR-RIS transmission or reflection. in the environment, U χ (Q) is the aperture distributions for
1089-7798 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: East Carolina University. Downloaded on June 15,2021 at 22:08:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2021.3082214, IEEE
Communications Letters
3

In the following, we compare the performance of the STAR-

RIS with the conventional RIS and demonstrate the benefit of
deploying STAR-RISs in wireless systems. The power ratio
between transmitted signals and the incident signal is fixed to
β T , and the power ratio of the reflected signal is fixed to β R ,
Transmitter for all STAR-RIS elements2 . Based on the setup described
in Section III, we further assume that Tx and the STAR-
RIS can be positioned with a line-of-sight (LoS) Tx-RIS
link. Both Rxs are located in the far-field region of the RIS.
Receiver T
Receiver R
The direct Tx-Rx links follows Ricean distribution as: hχ ∼
R(Kdχ , Ωχd ), and the links between each STAR-RIS element
χ
and Rxs (rm ) are identically and independently distributed,
χ
Fig. 2. Illustration of the considered STAR-RIS-assisted network. following Ricean distribution as: rm ∼ R(Ksχ , Ωχs ). Messages
intended for different Rxs are sent on different orthogonal
the transmitted or reflected waves at point Q on (Σ), F (θχ )
frequency
√ bands. The signal sent by Tx can be expressed as
is the leaning factor at point Q, dχ are the distances between
x = Pt (αR sR + αT sT ), where Pt is the transmit power,
Q and the receiver χ, and λ is the free-space wavelength of
αχ is the power allocation coefficient, and sχ is the intended
the signal. The integral in (6) can be evaluated element by
message for receiver χ. The signal-to-noise ratio (SNR) of
element. Assuming that the aperture distributions U χ (Q) is
receiver χ for the case of STAR-RIS or conventional RIS can
uniform within each element, at the mth element, we have
be expressed as γµχ = |Hµχ |2 Pt (αχ )2 /σ02 , where µ ∈ {S, C} is
U T (Q) = ΦTm hm and U R (Q) = ΦR m hm . Thus, (6) can be an indicator representing “S” for the corresponding quantities
expressed as:
Ae X χ 2jπdχ
m /λ
in the STAR-RIS case, or “C” for the conventional RIS case,
χ e
gχ = Φm hm F (θm ) χ , (7) Hµχ is the overall channel, and σ02 is the power spectrum
jλ m dm
χ density of the Gaussian white noise. According to the far-field
where Ae is the area of each element, and θm is the di-
channel model, the overall channel of receiver χ is:
rection of receiver χ, with respect to the normal direction
of STAR-RIS, as illustrated in Fig. 2. According to the Hµχ = (rχ )H Φχµ h + hχ , (9)
Fresnel-Kirchhoff diffraction formula, the leaning factor is
F (θmχ
) = (1 + cos θm χ
)/2, which holds for both receiver T where ΦTS = diag(T1 , · · · , TM ), ΦR S = diag(R1 , · · · , RM ),
and receiver R. Thus, the channel gain can be expressed as: ΦTC = diag(0, · · · , 0, T1 , · · · , TMt ), and ΦR C =
2jπdχ
m /λ
diag(R1 , · · · , RMr , 0, · · · , 0). For the case of STAR-
χ Ae X χ χ e χ
|g | = Φm hm (1 + cos θm ) χ . (8) RIS, under given transmit power Pt and target SNR γtarget
λ m 2dm for receiver χ, the outage probability can be formulated as
By comparing these results with the channel gains of the
follows:
far-field channel model, it can be noticed that the distances
M χ
between STAR-RIS elements and Rx can not be treated as the χ
np X
χ jφχ
2 σ02 γtarget o
Pout,S =Pr βχ rm e m +hχ < , (10)
same and be brought outside of the summation. In addition, Pt (αχ )2
m=1
the contribution of the leaning factor should be explicitly
considered in the near-field model. where β T and β R are the power ratios for the transmitted and
Remark 1. Comparing the channel models given in (5) and reflected signals of all STAR-RIS elements. For the case of
(8), it can be noticed that the angular distribution of the conventional RIS, the outage probability is:
far-field channel gain is independent of the distance between Mχ
n X 2
χ
σ02 γtarget o
χ χ
χ jφm
Rx and the RIS. However, this is not the case for near- Pout,C = Pr rm e + hχ < . (11)
field scenarios. This difference is the essential criteria that m=1
Pt (αχ )2
set the distinction between the two field regions, and it is
further illustrated by our simulation results. According to Note that the summation in (11) has Mt terms for receiver T
recent study [14], the boundary between the near-field and and Mr terms for receiver R. In the following, we compare the
the far-field is 2L2a /λ, where La is the largest dimension of maximum achievable diversity orders of Rxs for STAR-RISs
the RIS aperture. and conventional RISs.
IV. D IVERSITY A NALYSIS Theorem 1. For STAR-RISs, the asymptotic outage probability
To concisely highlight the benefits of STAR-RISs, in this of receiver χ can be expressed as follows (superscripts “χ”
section, we present a comparison between the following two for K and Ω are omitted):
RISs with specific configurations: 2M +1 (Ks + 1)M (Kd + 1)
χ χ
• STAR-RIS: All STAR-RIS elements share common values Pout,S (γtarget )=
for the power ratios of the transmitted signal and the re- (2M + 2)! ΩM χ 2M +2
s Ωd (α ) (12)
χ −M/2 −M Ks −Kd χ −(M +1)
flected signal. However, the phase delays of each element · (β ) e σ02M +2 (γtarget )M +1 Pt .
can be independently configured.
2 It is worth mentioning that for STAR-RISs, there is a trade-off between
• Conventional RIS: A composite smart surface consists of
the received power of receiver T and receiver R by changing the power ratio
transmitting-only and reflecting-only RISs, each with Mt of each element. Including these power ratios in the optimization can further
and Mr elements, respectively. improve the performance of the system.

1089-7798 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: East Carolina University. Downloaded on June 15,2021 at 22:08:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2021.3082214, IEEE
Communications Letters
4

√
For conventional RISs, the asymptotic outage probability of αT = αR = 1/ 2.
receiver χ can be expressed as follows: In Fig. 3, we compare the coverage of the proposed STAR-
0 0 RIS with conventional RIS by simulating their radiation pat-
χ χ 2M +1 (Ks + 1)M (Kd + 1)
Pout,C (γtarget ) = 0 χ 2M 0 +2
terns. The simulation is carried out as follows: Based on (6),
(2M 0 + 2)!ΩM
s Ωd (α ) (13) we calculate the channel gain of each point on the plane Σe .
−M 0 Ks −Kd 0 χ 0 −(M 0 +1)
·e σ02M +2 (γtarget )M +1 Pt , For STAR-RIS, we set β T = β R = 1/2. The phase shift
configurations of both the STAR-RIS and conventional RIS is
where M 0 = Mt for receiver T and M 0 = Mr for receiver R. chosen according to the cophase condition [3]. Specifically,
Proof. Following the assumption of Ricean fading channels, the target angles of the transmitted signal and the reflected
the probability density functions (PDFs) of Tx-Rx direct link signal are set as 7.6◦ and 16.6◦ , respectively. As illustrated in
and RIS-Rx links can be expressed in the following series Fig. 3(b), the power density of the STAR-RIS is significantly
form: higher near the target angles, forming a beam-like radiation
2(Ksχ + 1) −Ksχ pattern. In contrast, in Fig. 3(c), the transmitted and reflected
p|rm
χ (x) =
| √ χ χ e x + o(x2 ), (14)
β Ωs beams of the conventional RIS are less focused with weaker
2(Kdχ + 1) −K χ channel gains. This observation is consistent with our ana-
p|hχ | (x) = e d x + o(x2 ), (15) lytical results presented in Section. IV. Moreover, it can be
Ωχd
observed that the angular distribution of power density has an
where o(·) is the little-O notation, and o(x2 ) denotes a function irregular pattern which highly depends on the distance in the
which is asymptotically smaller than x2 . To obtain the PDF near-field region, while in the far-field region, a fixed radiation
of the magnitude of the overall channel (|Hkµ |), we calculate pattern can be observed. These observations of the radiation
its Laplace transform according to the convolution theorem. patterns are consistent with the results of the two proposed
Consider the case of STAR-RIS, we have: channel models.

M Fig. 4 shows the channel gain obtained by the 16×16 STAR-
L{p|HSχ | (x)}(t) = L{p|rm χ
|(x) } L{p|hχ | (x)}
RIS. The power ratios are set as β T = 0.4 and β R = 0.6. For
2M +1
(Ks + 1) (Kd + 1) −M Ks −Kd −2M −2 (16)
M
the horizontal axis, d denotes the distance between Rx and
= e ·t .
(β χ )M/2 ΩM s Ωd the center of STAR-RIS. The signal is reflected to the region
Thus, by applying the inverse Laplace transform, the outage where d > 0, and transmitted to the region where d < 0.
probability can be calculated according to: In the simulation, we assume that Rx moves on both sides
r
χ
of STAR-RIS, along a straight line in Σe with an angle of
Z σ0
αχ
γtarget
Pt
60◦ with respect to the normal direction of STAR-RIS. As
χ
Pout,S = p|HSχ | (x)dx. (17) illustrated in Fig. 4, the dash-dotted blue line represents the
0 channel gains calculated using the near-field channel model
The proof for the case of conventional RISs follows the same without considering the leaning factor. In other words, we
strategy, thus is omitted here for brevity. T R
assume F (θm ) = F (θm ) = 1. The solid red line represents
In the following remark, we consider the diversity order the channel gains calculated using the near-field channel model
χ χ
defined through outage probability as follows: with leaning factor F (θm ) = (1+cos θm )/2. The dashed black
χ χ
line represents the channel gains calculated using the far-field
log Pout (γtarget ) channel, according to (5). It can be observed that results for the
dχ = − lim . (18)
Pt →∞ log Pt near-field model start to approach the far-field model after Rx
Remark 2. According to Theorem 1, the diversity orders is several wavelength away from STAR-RIS. Within the near-
of different Rxs can be readily obtained from (12) and (13). field region, the far-field model fails to provide a physically
For STAR-RIS, the maximum achievable diversity orders of meaningful result as the channel gain tends to infinity when
both Rxs are dTS = dR the distance approaches zero.
S = M + 1. Thus, the combined
diversity order is dTS + dR Fig. 5 illustrates the outage probabilities and diversity orders
S = 2M + 2. For conventional
RISs, the maximum achievable diversity order for receiver T of STAR-RIS and conventional RIS. The red markers represent
and receiver R follows: dTC + dR the simulated outage probability for Rxs assisted by an STAR-
C = M + 2, yielding smaller
achievable diversity orders compared to STAR-RISs. RIS with eight elements (M = 8), β T = 0.4, and β R = 0.6.
The blue markers represent a conventional RIS with Mt = 3
V. N UMERICAL R ESULTS and Mr = 5. It can be observed that the simulated outage
In this section, we numerically demonstrate our proposed probabilities fit well with the analytical asymptotic results.
models to compare the coverage of the STAR-RISs and the The diversity orders of Rxs are consistent with our analytical
conventional RISs. Then, we simulate the outage probabilities results. For STAR-RIS, both Rxs achieve full diversity order
of Rxs served by STAR-RISs and the conventional RISs of M + 1, for conventional RIS, the diversity orders follow
to verify our analytical results. The geographical setup is dTC + dR C = M + 2, which is consistent with Remark 2.
shown in Fig. 3(a), where the RIS is positioned at the origin
in the z = 0 plane. We simulate a 16 × 16 STAR-RIS, VI. C ONCLUSIONS
where the spacing of each element is chosen as λ/2, and a In this letter, the concept of STAR-RISs was proposed.
conventional RIS with Mt = Mr = 128. Tx sends information A general hardware model and two channel models were
to receiver T and receiver R with equal power allocation, i.e., proposed. To evaluate the performance of STAR-RISs, we
1089-7798 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: East Carolina University. Downloaded on June 15,2021 at 22:08:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2021.3082214, IEEE
Communications Letters
5

7
20 20 6
Receiver T
ST 6

A
40 40

R- STAR-RIS Transmitting-only RIS 5

R
60 60

S
5

80 Reflected beam 80 4
4
Reflected beam
100 100
3
120 3 120 Transmitted beam
Transmitted beam
Receiver R 140 140 2
2
Reflecting-only RIS
160 160
Near-field radiation 1
180
1
180

200 200 0
100 200 300 400 500 600 100 200 300 400 500 600

(a) Simulation setup (b) STAR-RIS (c) Conventional RIS

Fig. 3. Illustration of the simulation setup and simulated radiation coverage observed on plane Σe .

R EFERENCES
6
Near-field model
without leaning factors [1] M. Di Renzo, M. Debbah, D.-T. Phan-Huy, A. Zappone, M.-S.
5
Near-field model Alouini, C. Yuen, V. Sciancalepore, G. C. Alexandropoulos, J. Hoydis,
with leaning factors
Far-field model H. Gacanin et al., “Smart radio environments empowered by recon-
figurable ai meta-surfaces: An idea whose time has come,” EURASIP
4
Journal on Wireless Communications and Networking, vol. 2019, no. 1,
pp. 1–20, May 2019.
3 [2] C. Huang, S. Hu, G. C. Alexandropoulos, A. Zappone, C. Yuen,
R. Zhang, M. D. Renzo, and M. Debbah, “Holographic MIMO surfaces
2 Transmitted signal Reflected signal for 6G wireless networks: Opportunities, challenges, and trends,” IEEE
Wireless Commun., vol. 27, no. 5, pp. 118–125, Oct. 2020.
[3] Y. Liu, X. Liu, X. Mu, T. Hou, J. Xu, Z. Qin, M. Di Renzo,
1 and N. Al-Dhahir, “Reconfigurable intelligent surfaces: Principles and
opportunities,” IEEE Commun. Surv. Tutor., Early Access, 2021, doi:
0 10.1109/COMST.2021.3077737.
-20 -15 -10 -5 0 5 10 15 20 [4] C. Huang, A. Zappone, G. C. Alexandropoulos, M. Debbah, and
C. Yuen, “Reconfigurable intelligent surfaces for energy efficiency in
wireless communication,” IEEE Trans. Wireless Commun., vol. 18, no. 8,
Fig. 4. Channel gain computed using far-field and near-field channel models. pp. 4157–4170, Aug. 2019.
[5] J. Xu and Y. Liu, “A novel physics-based channel model for reconfig-
urable intelligent surface-assisted multi-user communication systems,”
100 C-RIS (R) arXiv preprint arXiv:2008.00619, 2020.
STAR-RIS (R)
C-RIS (T)
[6] A. R. Ndjiongue, T. M. N. Ngatched, O. A. Dobre, and H. Haas,
STAR-RIS (T) “Towards the use of re-configurable intelligent surfaces in VLC systems:
Asymptotic (C)
10-1
Asymptotic (S) Beam steering,” arXiv preprint arXiv:2009.06822, 2020.
[7] X. Mu, Y. Liu, L. Guo, J. Lin, and R. Schober, “Intelligent reflecting
surface enhanced indoor robot path planning: A radio map based
approach,” IEEE Trans. Wireless Commun., Early Access, 2021, doi:
10-2 10.1109/TWC.2021.3062089.
[8] G. C. Alexandropoulos, N. Shlezinger, I. Alamzadeh, M. F. Imani, H.
Conventional RIS (C)
Zhang, and Y. C. Eldar, “Hybrid reconfigurable intelligent metasurfaces:
Enabling simultaneous tunable reflections and sensing for 6G wireless
10-3
communications,” arXiv preprint arXiv:2104.04690, 2021.
[9] N. DOCOMO. “DOCOMO conducts world’s first successful
STAR-RIS (S)
trial of transparent dynamic metasurface”. [Online]. Available:
10 -4 www.nttdocomo.co.jp/english/info/media center/pr/2020/0117 00.html
1 2 3 4 5 6 7 8 9 10 11
[10] J. Zhang, Z. Li, L. Shao, and W. Zhu, “Dynamical absorption manipula-
tion in a graphene-based optically transparent and flexible metasurface,”
Carbon, vol. 176, pp. 374–382, May 2021.
Fig. 5. Outage probabilities of for STAR-RIS and conventional RIS (C-RIS). [11] X. Wang, J. Ding, B. Zheng, S. An, G. Zhai, and H. Zhang, “Si-
multaneous realization of anomalous reflection and transmission at two
derived expressions for the asymptotic behaviour of the outage frequencies using bi-functional metasurfaces,” Sci. Rep., vol. 8, no. 1,
probability. Numerical simulations demonstrated the differ- pp. 1–8, Jan. 2018.
[12] N. M. Estakhri and A. Alu, “Wave-front transformation with gradient
ences between the channel gains in the far-field and the near- metasurfaces,” Phys. Rev. X, vol. 6, no. 4, p. 041008, Oct. 2016.
field region and verified our proposed channel models. It is [13] L. La Spada, C. Spooner, S. Haq, and Y. Hao, “Curvilinear metasurfaces
also demonstrated that the proposed STAR-RISs can extend for surface wave manipulation,” Sci. Rep., vol. 9, no. 1, pp. 1–10, Feb.
2019.
coverage and achieve higher diversity order on both sides [14] F. H. Danufane, M. Di Renzo, J. de Rosny, and S. Tretyakov,
of the surface, compared to conventional RISs. Based on “On the path-loss of reconfigurable intelligent surfaces: An approach
the proposed models, the joint optimization of transmission based on green’s theorem applied to vector fields,” arXiv preprint
arXiv:2007.13158, 2020.
and reflection beamforming of the STAR-RISs constitutes an
interesting topic for future research.

1089-7798 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: East Carolina University. Downloaded on June 15,2021 at 22:08:55 UTC from IEEE Xplore. Restrictions apply.