1

Joint Beamforming Optimization for Active

STAR-RIS-Assisted ISAC Systems
Shuang Zhang, Wanming Hao, Senior Member, IEEE, Gangcan Sun, Chongwen Huang, Member, IEEE,
Zhengyu Zhu, Senior Member, IEEE, Xingwang Li, Senior Member, IEEE, Chau Yuen, Fellow, IEEE

Abstract—In this paper, we investigate an active simulta- are conducted through numerical simulations to show their
neously transmitting and reflecting reconfigurable intelligent benefits and efficiency.
surface (STAR-RIS) assisted integrated sensing and commu-
arXiv:2308.06064v2 [[Link]] 14 Nov 2024

nications (ISAC) system, where the dual-function base station Index Terms—Integrated sensing and communication, simul-
(DFBS) operates in full-duplex (FD) mode to provide commu- taneously transmitting and reflecting reconfigurable intelligent
nication services and performs targets sensing simultaneously. surface, alternating optimization.
Meanwhile, we consider multiple targets and multiple users
scenario as well as the self-interference at the FD DFBS.
Through jointly optimizing the DFBS and active STAR-RIS I. I NTRODUCTION
beamforming under different work modes, our purpose is to
achieve the maximum communication sum-rate, while satisfy- Recently, with the development of radar and communi-
ing the minimum radar signal-to-interference-plus-noise ratio cation technologies, their differences in antenna structure,
(SINR) constraint, the active STAR-RIS hardware constraints
hardware component architecture and working bandwidth
and the total power constraint of DFBS and active STAR-
RIS. To tackle the complex non-convex optimization problem have been gradually reduced, which attracts widespread
formulated, an efficient alternating optimization algorithm is attention on integrated sensing and communication (ISAC)
proposed. Specifically, the fractional programming method in academic and industrial fields [1], [2]. By the utilization of
is first leveraged to turn the original problem into a more shared hardware platforms, spectrum and signal processing
tractable one, and subsequently the transformed problem is
frameworks, ISAC can integrate sensing and communication
decomposed into several sub-problems. Next, we develop a
derivation method to obtain the closed-form expression of the tightly and seek their performance balance, thereby sig-
radar receiving beamforming, and then the DFBS transmit nificantly enhancing the spectral efficiency, decreasing the
beamforming is optimized under the radar SINR requirement hardware cost and system power consumption, which has
and total power constraints. After that, the active STAR- made it gain recognition as a promising technology for the
RIS reflection and transmission beamforming are optimized
forthcoming sixth generation (6G) wireless communication
by majorization minimization, complex circle manifold and
convex optimization techniques. Finally, the proposed schemes [3]. However, ISAC still faces several challenges in practical
applications, such as high environmental dependence, lim-
ited coverage, and many blind spots. Fortunately, owing to
This work was supported by the China National Natural Science Foun- the ability to create efficient line-of-sight (LOS) links and
dation under Grant 62101499, the open research fund of National Mobile provide additional degrees of freedom (DoFs) by smartly
Communications Research Laboratory Southeast University under grant
2024D12 and 2024D15, and the Program for Science & Technology Innova-
manipulating the wireless propagation environment, the re-
tion Talents in Universities of Henan Province under Grant 24HASTIT038. configurable intelligent surface (RIS) offers a novel approach
The work of C. Huang was supported by the China National Key R&D to meet these challenges [4]–[7]. Typically, RIS consists of
Program under Grant 2021YFA1000500 and 2023YFB2904800, and the
National Natural Science Foundation of China under Grant 62101492. The
numerous reconfigurable elements with low power and low
work of X. Li was supported by the Key Research and Development Project cost, each of which can dynamically adjust the phase shift
of Henan Province under Grant 231111210500. The work of C. Yuan was and amplitude of the incident signal, allowing for a con-
supported by the Ministry of Education Singapore MOE Tier 2 (Award
number MOET2EP50220-0019). (Corresponding author: Wanming Hao)
trollable propagation environment [8], [9]. Deploying RIS
S. Zhang, W. Hao, G. Sun, Z. Zhu are with the School of Electrical can provide a more controllable propagation environment,
and Information Engineering, Zhengzhou University, Zhengzhou 450001, improving the coverage range and transmission quality [10]–
China, and W. Hao is also with the National Mobile Communications
Research Laboratory, Southeast University, Nanjing 214135, China (E-mail:
[13].
yayue96@[Link], {iewmhao,iegcsun,iezyzhu}@[Link]).
C. Huang is with College of Information Science and Electronic En-
gineering, Zhejiang University, Hangzhou 310027, China, and Zhejiang A. Related works
Provincial Key Laboratory of Info. Proc., Commun. & Netw. (IPCAN),
Hangzhou 310027, China (E-mail: chongwenhuang@[Link]). Nowadays, there have been several works considering
X. Li is with the School of Physics and Electronic Information Engi-
neering, Henan Polytechnic University, Jiaozuo, China, and also with the integrating RIS into ISAC systems [14]–[22], in which the
National Mobile Communications Research Laboratory, Southeast Univer- joint beamforming optimization of RIS and base station
sity, Nanjing 214135, China (email: lixingwang@[Link]). (BS) is proposed to improve the system performance. To
C. Yuen is with the School of Electrical and Electronics Engi-
neering, Nanyang Technological University, 639798, Singapore (e-mail: be specific, in [14], a RIS-assisted dual-function radar and
[Link]@[Link]). communication system was designed for the first time with

November 15, 2024 DRAFT

 2

single-user and single-target scenario. Then, the radar signal- RIS [23]–[26], which can achieve full-space coverage. A
to-noise ratio (SNR) maximization problem was formu- STAR-RIS-enabled ISAC framework was first brought up in
lated under transmit power budget and quality of service [27]. To overcome the severe path loss and interference from
(QoS) constraint, and an iterative optimization algorithm clutter in sensing, a new type of STAR-RIS sensor structure
was introduced based on majorization minimization (MM), was introduced, where the dedicated sensors were installed at
semidefinite programming (SDP) and semidefinite relaxation the STAR-RIS. Furthermore, to explore the ability of the typ-
(SDR) techniques to solve it. The authors [15] extended ical STAR-RIS in ISAC systems, the authors [28] considered
their work to the multi-user scenario, where the RIS re- an ISAC network enhanced by energy-splitting STAR-RIS.
flection coefficients, BS transmit beamforming and receive Under the objective of maximizing sensing SNR, an iterative
filter were jointly designed to maximize the sum-rate. To algorithm was created based on MM, sequential rank-one
address the optimization problem, an iterative algorithm constraint relaxation and SDR techniques. After that, the
framework was introduced, utilizing the alternative direction authors [29] undertook the initial exploration of leveraging
method of multipliers (ADMM), fractional programming STAR-RIS to enhance the security efficacy of ISAC systems,
(FP) and MM techniques. In [16], the authors considered the and a secrecy rate maximization problem was formulated
interference of strong multiple clutters on target detection, while ensuring the sensing minimum SNR. The authors
and aimed to maximize the radar signal-to-interference-plus- [30] further considered the physical layer security of ISAC
noise ratio (SINR) while ensuring the QoS requirements for systems with the presence of multiple eavesdropping targets.
communication, while ensuring the QoS requirements for To maximize the average received radar sensing power,
communication. To deal with it, an algorithm framework a low-complexity approach based on distance-majorization
based on ADMM and MM was introduced. In [17], the was conducted. As a further advance, the authors [31] uti-
authors utilized RIS to assist the multiple targets detection, lized the long-term synergistic effect between STAR-RIS and
and a minimum gain of RIS’s beampattern effective algo- ISAC and the two deep reinforcement learning algorithms to
rithm framework was exploited to obtain the maximized augment legitimate users’ mean long-term security rate.
expected detection angles, while meeting the minimum Nevertheless, the above works in ISAC systems mainly
communication SNR constraint. The authors [18] further focus on the passive STAR-RIS. Although the ability to
considered a more general scenario, where the BS per- simultaneously reflect and transmit signals enables passive
formed multiple users communication and multiple targets STAR-RIS to realize full-space coverage, it still has the
sensing. The weighted summation of target detection SNRs inherent flaws of passive RIS, namely multiplicative fading
was maximized by comprehensively applying penalty-based, [32]. This severe twin-hop path-loss fading will limit the
manifold optimization and MM methods. In [19] and [20], a performance improvement of passive STAR-RIS. Therefore,
new joint optimization metric was introduced. Specifically, inspired by the active RIS introduced [33]–[35], to overcome
work [19] attempted to utilize the joint optimization metric the multiplicative fading and significantly improve space
to maximize the radar SINR, and meanwhile minimizing the coverage, a new concept has recently been proposed, namely
multi-user interferences. In work [20], the cross-correlation active STAR-RIS. The advantage of active STAR-RIS is that
pattern, beam pattern error and total interference between it can not only reflect and transmit the incident signals, but
multiple users and radar were minimized. The work [21] also amplify the incident signal to compensate for the multi-
concentrated on minimizing the BS transmit power by taking plicative fading. Hence, it can achieve the same performance
into account the interference caused by the cross-correlation as passive STAR-RIS, but requires much fewer elements,
between radar and RIS, and an alternating optimization (AO) which means that the complexity of channel estimation and
algorithm based on SDR was proposed. The authors [22] beamforming can be significantly reduced. Compared to the
created a novel strategy of RIS-assisted beamforming in performance gain attained by enlarging the STAR-RIS size,
consideration of the dimensions of the target, where the the gain brought by active nature is more appealing and
conception of ultimate detection resolution (UDR) based practical.
on target size was introduced to evaluate the capability of Currently, there have been some preliminary studies on
target detection the first time. Then an optimization problem active STAR-RIS, including [36]–[42]. The concept of active
of maximizing user SNR was formulated subject to the STAR-RIS was first brought up in [36] to assist vehicular
minimum UDR requirement. communication system, where the transmit precoding matrix
However, the aforementioned researches assume that RIS at the BS and active STAR-RIS coefficient matrices were
only reflects the incident signals. In this case, it is necessary jointly optimized to minimize the BS’s transmit power.
for the communication users or targets to be placed on the The authors [37] introduced active STAR-RIS into the non-
same side of RIS and BS, resulting in a coverage limited orthogonal multiple access system, where the beamforming
to half-space. Given that the spatial confinement may not vector at the BS and the active STAR-RIS coefficients were
always be satisfied, it significantly restricts the flexibility and jointly optimized to obtain the maximized sum achievable
efficacy of RIS. Recently, there has been a growing interest rate. The authors [38] considered the joint BS transmit
in a burgeoning concept known as intelligent omni-surface beamforming and active STAR-RIS coefficients design in
(IOS) or simultaneously transmitting and reflecting (STAR)- multiuser multiple-input single-output systems, in which

November 15, 2024 DRAFT

 3

both sum-rate maximization and power minimization prob- beamforming and active STAR-RIS beamforming, sub-
lems were studied. The authors [39] proposed an efficient ject to the minimum sensing SINR constraint per target,
beamforming scheme for active STAR-RIS-aided wireless the active STAR-RIS hardware constraints and the
communications, where the sum-rate was maximized. The transmit power constraints at DFBS and active STAR-
authors [40] investigated the problem of maximizing the RIS. Due to the high coupling of variables, the objective
secrecy rate underlaying an active STAR-RIS-enabled com- function is non-convex with fractional form, as well
munication system, where the access point’s beamform- as the sensing SINR constraint and active STAR-RIS
ing and active STAR-RIS configuration were jointly de- phase shift constraint are both non-convex, so it poses
signed. The authors [41] studied the joint design of transmit a significant challenge to solve the formulated problem
beamforming and the active STAR-RIS configuration in a directly. Thus, the FP method is leveraged first to con-
multi-user MIMO system. With the purpose of maximizing vert the original problem into a more tractable structure
spectral efficiency, an analytic-based, highly computation- using the quadratic transform and the Lagrangian dual
parallelizable, and convergence-guaranteed algorithm was transform. Next, based on AO framework, the trans-
developed. The authors [42] deployed an active STAR-RIS in formed problem is decomposed into three subproblems,
a NOMA downlink multiuser network. By jointly optimizing which can be iteratively optimized.
the active beamforming and active STAR-RIS coefficients, • Then, we offer a derivation method for the closed
the sum-rate was maximized. expression of the radar receiving beamforming by re-
ducing it into a typical generalized “Rayleigh quotient”
optimization problem. Subsequently, the DFBS transmit
B. Main Contributions beamforming can be transformed into a conventional
quadratic constraint quadratic programming (QCQP)
Based on the above analysis, the recent works of active
optimization problem that can be solved with stan-
STAR-RIS focused on its application in communication
dard convex optimization algorithms. In the end, the
systems. To the best of our knowledge, the utilization of
active STAR-RIS reflection/transmission beamforming
active STAR-RIS to ISAC systems has yet to be investigated.
is designed under three active STAR-RIS work modes.
Especially, the joint beamforming design of active STAR-
For UED mode, with the other variables given, the
RIS and DFBS transceiver, the multiple targets and multiple
subproblem can also be reformulated as a QCQP opti-
users scenario, as well as the self-interference (SI) intro-
mization problem, and the standard convex optimization
duced by DFSB full-duplex (FD) mode, all bring challenges
algorithm can be applied to obtain the solution. For
for the application of active STAR-RIS in ISAC systems.
EED and SD modes, the subproblem can be addressed
Thus, motivated by the above discussions, we investigate the
by complex circle manifold and MM techniques.
joint design and optimization of active STAR-RIS-assisted
• Finally, numerical simulation results demonstrate the
ISAC systems to fill the research gap. Concretely, we
convergence and effectiveness of our proposed al-
propose an ISAC system empowered by active STAR-RIS,
gorithms. Our results show that the proposed active
wherein the DFBS operates in FD mode and transmits the
STAR-RIS schemes obtain a higher performance com-
dual-functional signals, thereby providing communication
pared to the passive STAR-RIS, which validates the
services for multiple user with the help of an active STAR-
benefits of deploying active STAR-RIS in ISAC sys-
RIS and performing multiple targets sensing concurrently.
tems.
For the proposed system, the joint beamforming design of
the DFBS and active STAR-RIS is examined to maximize The rest of this paper is organized as follows. In Section II,
the communication sum-rate, while ensuring the SINR of all the system model of the downlink active STAR-RIS-assisted
targets surpasses a predefined threshold. Overall, the main ISAC system is introduced, and the maximum communica-
contributions of our work can be summarized as follows: tion sum-rate optimization problem is formulated. Section
• We investigate an active STAR-RIS-assisted ISAC sys- III presents the proposed joint beamforming optimization
tem for mitigating the multiplicative fading effect as framework under three active STAR-RIS work modes. The
well as enhancing full-space coverage. This system simulation results are shown in Section IV. Finally, conclu-
leverages a DFBS and an active STAR-RIS to serve sions are given in Section V.
multiple users and detect multiple targets. Meanwhile, Notations: Throughout this paper, the following notations
we consider the existing three work modes of active are employed. Column vectors are typically indicated by
STAR-RIS, i.e., unequal energy division (UED), equal boldface lower-case letters, whereas matrices are commonly
energy division (EED) and space division (SD) modes. represented by boldface upper-case letters. (·)T , (·)∗ , (·)H ,
In order to enhance practicality, we consider the FD and (·)−1 indicate the transpose, conjugate, conjugate-
mode and SI at the DFBS, as well as the echo interfer- transpose, and the inversion operations of a matrix, respec-
ences caused by multiple targets. tively. | · | and ℜ{·} refer to the modulus and real part of
• On this basis, we formulate an optimization problem, a complex variable, respectively. E{·} and Cm×n denote
which aims to maximize the communication sum-rate statistical expectation vaule and the complex space with a
through the joint design of DFBS transmit/receive dimension of m × n, respectively. diag(·) refers to create a

November 15, 2024 DRAFT

 4

TABLE I active STAR-RIS is assumed to be deployed in close prox-

N OTATIONS imity to the users for assisting downlink communications
effectively. In the meantime, we assume that the targets fly
Notation Description
M The number of DFBS transmit/receive antennas
at low altitudes and own the robust LoS link with the DFBS.
N The number of active STAR-RIS elements Considering the round-trip characteristic of the echo signals,
K/Q The total number of users and targets the signals that go through the DFBS-target-active STAR-
Kr /Kt
The number of users at reflection space and RIS-DFBS channel are highly attenuated due to three-hop
transmission space transmissions, and since the active STAR-RIS is far from the
The reflection and transmission coefficient matrices targets, which will cause the three-hop path loss to be more
Ψr / Ψt
of active STAR-RIS
The reflection and transmission phase-shift matrices
severe. Thus, compared to the echoes reflected by STAR-
Φr /Φt RIS and undergoing three-hop attenuation to reach DFBS,
of active STAR-RIS
The reflection and transmission amplitude matrices the target echoes will be much larger1 . Therefore, similar to
Ar /At
of active STAR-RIS the prior works [21], [28], [43], [44], the reflected clutter
Equivalent active STAR-RIS reflection coefficient, interferences through active STAR-RIS in our paper are
ψ r /ϕr /ar
phase-shift and amplitude vectors
omitted due to the severe round-trip path loss and physical
Equivalent active STAR-RIS transmission
ψ t /ϕt /at obstacles.
coefficient, phase-shift and amplitude vectors
x Transmission signal at the DFBS
sc /sr Communication signal and radar signal
Wc /Wr Beamforming matrix for users and targets
A. Active STAR-RIS And Signal Models

W/s
Equivalent beamforming matrix and transmission The active STAR-RIS owns the ability to amplify, trans-
signal at the DFBS mit, and reflect signals simultaneously via amplifiers and
wj Equivalent beamforming vector at W j-th column
phase shifters. Each element of the active STAR-RIS can
G Channel from the DFBS to the active STAR-RIS
fk Channel from the active STAR-RIS to the k-th user have the incident signals’ amplitude and phase-shift adjusted
Direct channels from the DFBS to the k-the user separately using different components [36]–[42]. As shown
hd,k /hd,q
and q-th target in the upper half of Fig. 1(b), the physical structure of each
H H Equivalent cascaded channels from the DFBS to the active STAR-RIS element consists of a reflective layer, an
h̃r,k /h̃t,k k-th user ampliate layer, and a transmissive layer. The reflective layer
SI channel between the DFBS transmitter and
HSI and transmissive layer are two symmetric layers, each one
receiver
b/ξ 2 The target RCS and the corresponding variance contains a metal patch, multiple positive intrinsic negative
Thermal noise of active STAR-RIS and complex (PIN) diodes, and a feedline. By controlling the bias voltage
v/zq /nk
AWGN of the q-th target and the k-th user of PIN diodes through the feedline, they can be work on
σv2 /σz2 /σk2 Variances of v, zq and nk the ON or OFF state so as to realize the different phase
uq The radar receive filter for q-th target shifts, and thus the phase shift can be adjusted flexible. The
Γq The sensing SINR requirement for the q-th target
power amplifier and power divider embedded in the ampliate
B R The power budget at the DFBS and active
Pmax /Pmax layer can be effectively designed to change the amplitude of
STAR-RIS
reflected and transmitted signals [38], [42]. The equivalent
circuit implementation can be found in the lower half of
Fig. 1(b), the active STAR-RIS structure deploys element
diagonal matrix from a vector or the diagonal elements of a arrays on two opposite faces of the plate with the element
matrix. IM indicates the M × M identity matrix. ⊗ refers aligned. Each pair of aligned elements on the opposite faces
to the Kronecker product. ∥ · ∥F indicates the F-norm of the are connected by circuits embedded within the plate. The
input matrix, and arg(·) is the phase of complex variable. incident signals on the active STAR-RIS are first fed into
Besides, we have summarized the main symbols and their the reflection-type amplifier (RA) for amplification. Then,
corresponding means in TABLE I to provide a better reading the amplified signals are divided into two parts by the
experience and facilitate understanding. power divider (PD), and are fed into a pair of opposite
controllable phase shifters in the reflective and transmissive
II. S YSTEM M ODEL AND P ROBLEM F ORMULATION layers. The signals are then emitted from the front and back
An active STAR-RIS-assisted ISAC system is consid- sides of the active STAR-RIS and represented as reflected
ered as depicted in Fig. 1(a), in which a DFBS with M signals and transmitted signals, respectively, thus achieving
transmit/receive antennas provides communication services simultaneous reflection and transmission of the incident
to K single-antenna users with the assistance of an N -
element active STAR-RIS and performs Q targets sensing 1 Deploying active STAR-RIS is used to assist DFBS to communicate
simultaneously. Here, we define the users located in the with users for improving the service of quality. Therefore, the active STAR-
reflection space as the r-user and can be denoted by the RIS reflect/transmit beamforming formulated by the proposed scheme will
set of Kr ≜ {1, 2, . . . , Kr }, while the users located in the toward to the r-/t-users, rather than reflecting back to DFBS. In this case,
even the DFBS receives the echo from the active STAR-RIS, the received
transmission space is the t-user and can be denoted by the echo power from DFBD-active STAR-RIS-DFBS link is very low and can
set of Kt ≜ {1, 2, . . . , Kt }, and we have Kr +Kt = K. The be almost negligible.

November 15, 2024 DRAFT

 5

(a) (b)
Fig. 1. Structure of active STAR-RIS and its deployment in a ISAC system. (a) Active STAR-RIS-assisted ISAC system model. (b) The basic structure
of the active STAR-RIS.

signals. It is worth noting that the active STAR-RIS structure

is indeed feasible and can be implemented based on existing
technologies2 .
Hence, based on the above analysis, the reflection
and transmission matrices for active STAR-RIS are de-
fined as Ψr ≜ Ar Φr = diag(ψ r ) ∈ CN ×N and
Ψt ≜ At Φt = diag(ψ t ) ∈ CN ×N , respectively,
where Φr ≜ diag(ϕr ) = diag([ejφr,1 , . . . , ejφr,N ]T ) ∈
CN ×N denotes the reflection phase-shift matrix, Φt ≜
diag(ϕt ) = diag([ejφt,1 , . . . , ejφt,N ]T ) ∈ CN ×N denotes
the transmission phase-shift matrix, Ar ≜ diag(ar ) =
diag([ar,1 , . . . , ar,N ]T ) denotes the reflection amplitude ma- Fig. 2. The dual-functional signal transmission diagram.
trix, and At ≜ diag(at ) = diag([at,1 , . . . , at,N ]T ) denotes
the transmission amplitude matrix.
The DFBS transmit a dual-functional signal that includes
radar signals and communication signals. As illustrated in [wr,1 , . . . , wr,M ] ∈ CM ×M are their corresponding beam-
Fig. 2, we fix our attention on one typical transmission formers with wc,i or wr,i ∈ CM ×1 . To avoid mutual
frame, which comprises several time slots, thus, the trans- interference in radar and communication, the radar signals
mitted signal in the l-th time slot is given by: and communication signals are assumed  to be independent
ofeach other and own unit power, i.e., E sc (l)sH
r (l) = 0,
x(l) = Wc sc (l) + Wr sr (l) = Ws(l), (1) H
 H
E sc (l)sc (l) = IK and E sr (l)sr (l) = IM . We also
T
where sc (l) = [sc,1 (l), . . . , sc,K (l)] ∈ CK×1 and assume that the communication signals for different users
are uncorrelated, i.e., E sc,i (l)s∗c,j (l) = 0, ∀i ̸= j. For

T
sr (l) = [sr,1 (l), . . . , sr,M (l)] ∈ CM ×1 denote the
communication signals and radar signals, respectively, simplicity, in (1), we define W ≜ [Wc , Wr ] ∈ CM ×(K+M )
Wc = [wc,1 , . . . , wc,K ] ∈ CM ×K and Wr = with wj ∈ CM ×1 denotes the j-th column of W, j =
T
1, 2, ..., (K + M ), and s(l) ≜ sTc (l), sTr (l) ∈ C(K+M )×1 .


2 The key components of the active STAR-RIS are phase shifters, RA,
As our proposed ISAC system needs to serve multi-user
and PD. For phase shifters, they can be implemented by the phase shifters communication as well as target sensing, the communica-
that widely used in passive RIS/STAR-RIS. The implementation of the RA tion and sensing operations are synchronized such that the
and PD, can refer to various existing analog circuit technologies [38], [40], transmit antennas emit the dual-functional signals, while the
[41]. Specifically, RA can be implemented using various analog circuits,
such as, the aperture-coupled microstrip paths, tunnel diodes, field effect receive antennas receive the echoes from the targets during
transistors, and even some complementary metal oxide semiconductors. PD the downlink communication. Note that since the channel
can also be implemented utilizing low-cost analogue components, such as state information remains unchanged in one frame, and the
combining adjustable varactors/diodes with conventional power dividers or
couplers, reconfigurable synthesized transmission line techniques, and the operations are the same for each time slot, without loss of
new Type-II reconfigurable couplers. generality and for notational convenience, we omit the time

November 15, 2024 DRAFT

 6

index in the subsequent description of our paper. simultaneously, we assume that DFBS operate in FD mode,
For downlink communication, let G ∈ CN ×M , fk ∈ which will cause SI [45]. Note that SI can be suppressed
N ×1
C and hd,k ∈ CM ×1 represent the channels from the to an acceptable level by employing contemporary SI can-
DFBS to the active STAR-RIS, from the active STAR- cellation (SIC) techniques, but there still exists residual
RIS to the k-th user and from the DFBS to the k-th user, SI [46], [47]. Thus, due to the existing multiple targets
respectively. In order to fully exploit the potential of active and SI leakage, the received echo signal of the q-th target
STAR-RIS, we assume that it is available for DFBS to y q ∈ CM ×1 at the DFBS can be represented as:
obtain the accurate and complete channel state information. Q
Thus, for r-user, the received signal at the k-th user can be X
y q = bhd,q hH
d,q Ws + b hd,i hH
d,i W s+HSI Ws+zq , (7)
expressed as:
i=1,i̸=q
H
yk = hH
d,k x + fk Ψr (Gx + v) + nk where the scalar b represents the target radar cross section
(2)
= (hH H H
d,k + fk Ψr G)Ws + fk Ψr v + nk , k ∈ Kr ,
(RCS) with E |b|2 = ξ 2 , hd,q ∈ CM ×1 denotes the
channel between
the DFBS and the q-th target, and zq ∼
where v ∈ CN ×1 is the thermal noise that related to the CV 0, σz2 IM is AWGN for the q-th target. HSI ∈ CM ×M
inherent device noise of the

STAR-RIS elements with each represents the SI channel between the DFBS transmitter and
entry obeying CN 0, σv2 , and Ψr v is the dynamic noise receiver, and HSI Ws is the leaked transmission signal which
introduced by the amplifier in active STAR-RIS, also known is also known as SI signal. It is noted that the DFBS-target
as the amplified thermal noise. nk ∼ CN (0, σk2 ) represents links are LOS and the angle of arrival/departure (AoA/AoD)
the complex additive white Gaussian noise (AWGN) at of interest is known a priori.
the k-th user with variance σk2 . Due to the use of active Typically, in order to capture the desired reflected signal
components, active STAR-RIS consumes additional power of the q-th target, we apply a radar receive filter uq ∈ CM ×1
to amplify the incident signals, and the dynamic noise in to process the echo signal y q , and the radar output signal
active STAR-RIS cannot be neglected. Thus, SINR of the for the q-th target at the DFBS can be expressed as:
k-th r-user can be calculated as:
Q
H X
|h̃r,k wk |2 uH H H
q y q = uq bHq Ws+uq b Hi Ws+uH H
q HSI Ws+uq zq ,
SINRk = K+M
, k ∈ Kr , i=1,i̸=q
H
|h̃r,k wi |2 +σv2 ∥fH 2
P 2
k Ψr ∥ +σk (8)
i=1,i̸=k where Hq ≜ hd,q hH d,q ∈ C M ×M
is the equivalent channel
(3)
H H
of the q-th echo signal. Thus, the radar output SINR of the
where h̃r,k ≜ hH d,k + fk Ψ r G ∈ C 1×M
is the equivalent q-th target can be caculated as (9), which is presented at the
cascaded channels from the DFBS to the k-th r-user. top of the next page.
Similarly, the received signal for the k-th t-user can be
calculated as:
H
B. Problem Formulation
yk = hH
d,k x + fk Ψt (Gx + v) + nk
(4) In this paper, through jointly designing the DFBS transmit
H H
= (hH
d,k + fk Ψt G)Ws + fk Ψt v + nk , k ∈ Kt , beamforming W, the radar receive filter {uq }, ∀q, and the
active STAR-RIS reflection & transmission beamforming
and the correspoding SINR can be given by:
matrices, i.e., Ψr and Ψt , we aim to maximize the achiev-
H
|h̃t,k wk |2 able sum-rate, as well as satisfying the worst radar SINR
SINRk = K+M
, k ∈ Kt , requirement Γq , active STAR-RIS hardware constraints, the
H
|h̃t,k wi |2 +σv2 ∥fH 2 R B
P 2
k Ψt ∥ +σk given power budget Pmax at the active STAR-RIS and Pmax
i=1,i̸=k at the DFBS. The optimization problem is thus formulated
(5)
H as:
where ≜
h̃t,k hH
+d,k fH
∈ C
k Ψt G
1×M
represents the K
equivalent cascaded channels from DFBS to the k-th t-user. P0 : max
X
log2 (1 + SINRk ) (10a)
Hence, the sum rate can be calculated as: {wj },{uq },Ψr ,Ψt
k=1
K
X s.t. C1 : SINRq ⩾ Γq , ∀q, (10b)
Rsum = log2 (1 + SINRk ). (6)
K+M
k=1
X
C2 : ∥wj ∥2 ⩽ Pmax
B
, (10c)
From the perspective of radar sensing, to obtain an en- j=1
hanced target sensing performance, one common approach K+M
X
is to maximize the signal power directed towards the target C3 : ∥Ψr Gwj ∥2 + σv2 ∥Ψr ∥2F +
while minimizing it in other directions. On this basis, the j=1
assessment of sensing capabilities is usually conducted by (10d)
K+M
using the SINR of target echo. In addition, considering that
X
2
∥Ψt Gwj ∥ + σv2 ∥Ψt ∥2F ⩽ R
Pmax ,
DFBS needs to perform both communication and sensing j=1

November 15, 2024 DRAFT

 7

2
ξ 2 uH
q Hq Ws
SINRq = PQ 2 2 2
ξ2 i=1,i̸=q uH
q Hi Ws + uH
q HSI W s + uH
q zq
H H
(9)
ξ 2 uH
q Hq WW Hq uq
= Ä PQ ä .
uH
q ξ 2 i=1,i̸=q Hi WWH HH H H 2
i + HSI WW HSI + σz IM uq

C4 : ar,n ⩾ 0, at,n ⩾ 0, ∀n, (10e) By substituting (12) into (11), the original optimization
C5 : |ϕr,n | = 1, |ϕt,n | = 1, ∀n. (10f) problem P0 can be equivalently rewritten as P1 :

where, constraints C1 indicates the radar sensing SINR P1 : max R̃sum (γ, Ψr , Ψt , w, ρ) (13a)
{wj },{uq },Ψr ,Ψt ,γ,ρ
requirement per target, C2 and C3 represent the given
s.t. C1 , C2 , C3 , C4 , C5 . (13b)
maximum power budget at the DFBS and active STAR-
RIS, respectively, C4 means the active nature limitation of Then, P1 can be addressed by adopting the AO frame-
STAR-RIS amplitude, and C5 indicates the unit-modulus work, which involves dividing it into several sub-problems.
constraint of phase shift. It is clear that P0 is a non-convex
optimization problem due to the complex objective function
B. Optimize Auxiliary Variable Vectors γ and ρ
(10a) which includes logarithmic and fractional terms, the
interdependence of variables in both the objective function With given {wj }, Ψr and Ψt , the optimal γ and ρ can
∂f
(10a) and the power constraint C3 , and the constraint C5 be determined explicitly by letting ∂∂γ R̃sum
k
and ∂ρ k
to zero,
which imposes a unit modulus condition. Next, we propose i.e.:
an efficient iterative algorithm to deal with it.  H
|h̃r,k wk |
2



 K+M , k ∈ Kr ,
P H 2 +σ 2 ∥fH Ψ ∥2 +σ 2
|h̃ |

w

III. J OINT B EAMFORMING O PTIMIZATION

 r,k i v k r k
γkopt = i=1,i̸=k H 2
In this section, an AO algorithm is developed to solve |h̃t,k wk |
, k ∈ Kt .



P0 . Specifically, FP method is first utilized to convert the  K+M
 H
|h̃ wi |2 +σ 2 ∥fH Ψt ∥2 +σ 2

 P
objective function into a more tractable form. After that, the  t,k v k k
i=1,i̸=k
transformed problem is divided into multiple sub-problems, (14)
√
1+γk h̃H

that are solved in an alternating way. r,k wk



 K+M , k ∈ Kr ,
P H 2 +σ 2 ∥fH Ψ ∥2 +σ 2
|h̃ |

w


 r,k j v k r k
A. FP-based Transformation ρopt = j=1
√
k H
1+γk h̃t,k wk
We first convert the complicated sum-of-logarithms and ,

k ∈ Kt .


 K+M
fractions in (10a) into an easier-to-compute feasible form by
 H
|h̃ wj |2 +σ 2 ∥fH Ψt ∥2 +σ 2

 P
 t,k v k k
using the closed-form FP algorithm [48], [49]. Subsequently, j=1
by introducing an auxiliary variable γ = [γ1 , . . . , γK ]T and (15)
utilizing the Lagrangian dual transform to reformulate the
original objective function (10a) as: C. Optimize Radar Receive Beamforming {uq }
K+M
X K+M
X For given {wj }, the radar receive beamforming opti-
R̃sum (γ, Ψr , Ψt , {wj }) = (1 + γk ) − γk mization problem can be reduced to the typical generalized
k=1 k=1 “Rayleigh quotient” optimization problem [33] as follows:
H
(1 + γk )|h̃r,k wk |2 ξ 2 uH
q Cq u
X
+ K+M max , (16)
H uq H
uq Eq uq
|h̃r,k wj |2 + σv2 ∥fH 2 (11)
P 2
k Ψr ∥ + σk
k∈Kr
j=1
whose optimal solution uopt
q is the eigenvector that corre-
H
X (1 + γk )|h̃t,k wk |2 sponds to the maximum eigenvalue of matrix E−1 q Cq , where
+ . PQ
K+M H Cq = Hq WWH HH q , Eq = ξ 2
i=1,i̸=q H i WW H H
Hi +
+ σv2 ∥fH 2
P
k∈Kt |h̃t,k wj |2 2
k Ψt ∥ + σk H H 2
HSI WW HSI + σz IM .
j=1

Next, the quadratic transform is employed on the final two

fractional terms of (11) through the introduction of another D. Optimize DFBS Beamforming W
auxiliary variable ρ = [ρ1 , . . . , ρK ]T to further recast them For given {uq }, Ψr , Ψt , γ and ρ, the optimization of
to (12), which is presented at the top of the next page. DFBS transmit beamforming W can be modeled as:

November 15, 2024 DRAFT

 8

Ñ Ñ éé
K+M
X H X H
2 1 +γk ℜ{ρ∗k h̃r,k wk }−|ρk |2 |h̃r,k wj |2 +σv2 ∥fH
p 2 2
f (γ, Ψr , Ψt , w, ρ) = k Ψr ∥ +σk
k∈Kr j=1
Ñ Ñ éé (12)
K+M
X H X H
2 1 +γk ℜ{ρ∗k h̃t,k wk }−|ρk |2 |h̃t,k wj |2 +σv2 ∥fH
p 2 2
+ k Ψt ∥ +σk ,
k∈Kt j=1

is a non-convex constraint, which makes P2 non-convex. In

order to deal with this formidable undertaking, we exploit
max R̃sum (γ, Ψr , Ψt , {wj }, ρ) (17a)
{wj } the first-order Taylor expansion [50] to obtain an suitable
surrogate function for w̃H Ỹq w̃ and transform C1 into a
s.t. C1 , C2 , C3 . (17b)
linear constraint, which is given by:
For convenience, we first give the following definitions: w̃H Ỹq w̃ ⩾ w̃H H
s Ỹq w̃s + 2ℜ{w̃s Ỹq (w̃ − w̃s )}. (28)
w̃ ≜ vec{W}, (18)
p Then, P2 can be rewritten as:
αk ≜ 1 + γk ρk h̃r/t,k , ∀k ∈ K, (19)
P3 : min w̃H Qr w̃ − ℜ{2αH H H
r w̃} + w̃ Qt w̃ − ℜ{2αt w̃}
w̃
αr ≜ [αT1 , . . . , αTk , 01×[M (K+M )−M Kr ] ]T , ∀k ∈ Kr , (20) (29a)
Kr
X H
s.t. C1 : w̃H H
s Ỹq w̃s + 2ℜ{w̃s Ỹq (w̃ − w̃s )} ⩾ ηq , ∀q,
Qr ≜ IK+M ⊗ |ρk |2 h̃r,k h̃r,k , (21) (29b)
k=1
C 2 , C3 . (29c)
αt ≜ [01×M Kr , αT1 , . . . , αTk , 01×M M ] ]T , ∀k ∈ Kt , (22)
One can observe that P3 is a standard QCQP problem,
Kt
X H which can be effectively tackled with standard convex opti-
Qt ≜ IK+M ⊗ |ρk |2 h̃t,k h̃t,k , (23) mization algorithm or tools, e.g., CVX [51].
k=1

Yq ≜ IK+M ⊗ HH H
q ut uq Hq , ∀q, (24)
YSI,q ≜ IK+M ⊗ HH H E. Optimize Active STAR-RIS Beamforming Ψr and Ψt
SI uq uq HSI , ∀q, (25)
Ξr ≜ (IK+M ⊗ GH ΨH In this section, three work modes are considered for the
r Ψr G),
(26) active STAR-RIS, and the active STAR-RIS beamforming is
Ξt ≜ (IK+M ⊗ GH ΨH
t Ψt G). solved under each mode.
Then, we can reformulate the subproblem into explicit and 1) UED mode
compact forms with respect to w̃: In UED mode, each element of the active STAR-RIS can
H H transmit and reflect the incident signals with different am-
P2 : max ℜ{2αH H
r w̃} − w̃ Qr w̃ + ℜ{2αt w̃} − w̃ Qt w̃ plitudes and phases simultaneously, and this mode owns the
w̃
highest DoFs [26]. With given other variables, considering
H H H
(27a) the fact that h̃r,k ≜ hH H
d,k + fk Ψr G = hd,k + ψ r diag(fk )G,
H H H
Q
X ∀k ∈ Kr and h̃t,k ≜ hH H
d,k + fk Ψt G = hd,k + ψ t diag(fk )G,
s.t.C1 : ξ 2 w̃H Yq w̃−ξ 2 Γq w̃H Yi w̃−Γq w̃H YSI,q w̃ ∀k ∈ Kt , by defining (30), (31), (32), (33) and (34), as
i=1,i̸=q shown at the top of the next page, this subproblem with
≥ Γq uH 2 respect to ψ r and ψ t can be reformulated as:
q σz IM uq , ∀q,
(27b) P4 : min ψ H H H H
r Dr ψ r −ℜ{2ψ r dr } + ψ t Dt ψ t −ℜ{2ψ t dt }
ψ r ,ψ t
C2 : ∥w̃∥2 ⩽ Pmax
B
, (27c) (35a)
H H
C3 : w̃ Ξr w̃+ w̃ R
Ξt w̃ ⩽ Pmax −σv2 ∥Ψr ∥2F −σv2 ∥Ψt ∥2F . s.t. C3 : ψH ψH R
r Πψ r + t Πψ t ⩽ Pmax . (35b)
(27d)
Obviously, P4 is also a QCQP problem that can be
For simplicity to calculate, by mathematical transforma- addressed with the standard convex optimization algorithm.
tions, C1 can be rewritten as w̃H Ỹq w̃ ≥ ηq , ∀q, where By considering the constraints C4 and C5 , the associated
Y PQ Y H 2 2
Ỹq = Γqq − i=1,i̸=q Yi − SI,q ξ 2 , ηq = uq σz IM uq /ξ . phase-shift matrix Φopt
r ∈ CN ×N , Φopt
t ∈ CN ×N and
Since Ỹq is a positive semi-definite Hermitian matrix, C1 opt
amplification factor vector ar ∈ C N ×1
, aopt
t ∈ CN ×1

November 15, 2024 DRAFT

 9

X p K+M
X
dr ≜ 1 + γk diag(ρ∗k fH 2 H
k )Gwk − |ρk | diag(fk )G wj wH
j hd,k , (30)
k∈Kr j=1

X K+M
X
Dr ≜ |ρk |2 σv2 diag(fH
k )diag(fk ) + |ρk |
2
diag(fH H H
k )Gwj wj G diag(fk ), (31)
k∈Kr j=1

X p K+M
X
dt ≜ 1 + γk diag(ρ∗k fH 2 H
k )Gwk − |ρk | diag(fk )G wj wH
j hd,k , (32)
k∈Kt j=1

X K+M
X
Dt ≜ |ρk |2 σv2 diag(fH
k )diag(fk ) + |ρk |
2
diag(fH H H
k )Gwj wj G diag(fk ), (33)
k∈Kt j=1

K+M
X
Π= diag(Gwj )(diag(Gwj ))H + σv2 IN . (34)
j=1

are given by: 2) EED mode

In EED mode, the signals reflected and transmitted by
Φopt
r = diag(exp(jarg(ψ opt
r ))), (36a)
active STAR-RIS have the same power, i.e., all elements
Φopt
t = opt
diag(exp(jarg(ψ t ))), (36b) of active STAR-RIS transmit and reflect the incident signal
aopt = diag(exp(−jarg(ψ opt opt with the same amplitude simultaneously, given by Ar =
r r )))ψ r , (36c)
At = A ≜ diag(a) = diag([a1 , . . . , aN ]T ) ∈ CN ×N .
aopt
t = diag(exp(−jarg(ψ opt opt
t )))ψ t . (36d) Recall that Ψr = diag(ψ r ) = diag(Φr a), Ψt =
H
diag(ψ t ) = diag(Φt a), thus the equivalent channels h̃r,k
According to the definition of the reflection and transmis- H H H
sion precoding matrices for active STAR-RIS in Part II.A, and h̃t,k can be rephrased as h̃r,k ≜ hH H
d,k +fk Ψr G = hd,k +
H
i.e., Ψr = diag(ψ r ) and Ψt = diag(ψ t ), we can obtain the aH diag(fH H H
k Φr )G, ∀k ∈ Kr and h̃t,k ≜ hd,k + fk Ψt G =
optimal Ψr and Ψt , respectively. hd,k + a diag(fk Φt )G, ∀k ∈ Kt . Let cr = fH
H H H
k Φr ∈
1×N H 1×N
We summarize the joint beamforming optimizing scheme C , ∀k ∈ Kr and ct = fk Φt ∈ C , ∀k ∈ Kt , we
for UED mode in Algorithm 1. define (37), (38), (39) and (40), as shown at the top of the
next page.
With given other variables, the subproblem can be trans-
Algorithm 1 Joint Beamforming Optimization for UED formed into the following one:
Mode. P5 : min aH Ωr a − ℜ{2aH µr } + aH Ωt a − ℜ{2aH µt }
a
Input: hd,k , fk , σk2 ,
∀k ∈ K, G, hd,q , σv2 , σz2 , B
Pmax , R
Pmax , (41a)
Γt , Qmax , δth H H R
Output: DFBS transmit beamforming W, radar receive s.t. C3 : a Πa + a Πa ⩽ Pmax , (41b)
filter {uq }, active STAR-RIS beamforming Ψr and Ψt , C4 . (41c)
and sum rate Rsum .
1: Initialize: W, Ψr , Ψt , t = 1, δ = ∞, Rsum = 0 It is clear that P5 is also a standard QCQP problem
2: while t ⩽ Qmax and δ ⩾ δth do and we can easily obtain the optimal a by existing convex
3: Rpre = Rsum ; optimization algorithm or optimization tools.
4: Update γkopt and ρopt After obtaining a, to solve Ψr and Ψt , considering that
k , ∀k ∈ K by (14) and (15);
5: Update radar receiver {uq } by solving (16); Ψr = diag(ψ r ) = diag(Aϕr ) and Ψt = diag(ψ t ) =
6: Update DFBS beamforming W by solving P3 ; diag(Aϕt ), we should optimize ϕr and ϕt next. Similarly,
H H
7: Update Ψr and Ψt by solving P4 ; the equivalent channel h̃r,k and h̃t,k firstly be rewritten as
H H H H
8: Update Rsum by (6); h̃r,k ≜ hH H
d,k + fk Ψr G = hd,k + ϕr diag(fk A)G, ∀k ∈ Kr
|R −R | H
9: δ = preRsum sum ; and h̃t,k ≜ hH H H H H
d,k + fk Ψt G = hd,k + ϕt diag(fk A)G, ∀k ∈
10: t = t + 1. H
Kt . Let qr = fk A ∈ C 1×N
, ∀k ∈ Kr and qt = fH k A ∈
11: end while C 1×N
, ∀k ∈ Kt . Considering that the phase-shifts ϕr and

November 15, 2024 DRAFT

 10

X p K+M
X
µr ≜ 1 + γk diag(ρ∗k cr )Gwk − |ρk |2 diag(cr )G wj wH
j hd,k , (37)
k∈Kr j=1

X K+M
X
Ωr ≜ |ρk |2 σv2 diag(fH
k )diag(fk ) + |ρk |
2
diag(cr )Gwj wH H H
j G diag(cr ), (38)
k∈Kr j=1

X p K+M
X
µt ≜ 1 + γk diag(ρ∗k ct )Gwk − |ρk |2 diag(ct )G wj wH
j hd,k , (39)
k∈Kt j=1

X K+M
X
Ωt ≜ |ρk |2 σv2 diag(fH
k )diag(fk ) + |ρk |
2
diag(ct )Gwj wH H H
j G diag(ct ). (40)
k∈Kt j=1

ϕt are independent, we can optimize them separately. Next, the corresponding optimization of ϕt can be modeled as:
we define:
X p P7 : min ϕH H
t Ω̂t ϕt + ℜ{2ϕt (−µ̂t )} (47a)
µ̂r ≜ 1 + γk diag(ρ∗k qr )Gwk ϕt

k∈Kr s.t. |ϕt,n | = 1, ∀n. (47b)

K+M (42)
X We summarize the joint beamforming optimization
− |ρk |2 diag(qr )G wj wH
j hd,k ,
scheme for UED mode in Algorithm 2.
j=1
X
Ω̂r ≜ |ρk |2 σv2 diag(fH
k )diag(fk )+
Algorithm 2 Joint Beamforming Optimization for EED
k∈Kr
(43) Mode.
K+M
Input: hd,k , fk , σk2 , ∀k ∈ K, G, hd,q , σv2 , σz2 , Pmax
B R
X
H , Pmax
|ρk |2 diag(qr )Gwj wH H
j G diag(qr ).
,
j=1 Γt , Qmax , δth
Output: DFBS transmit beamforming W, radar receive
With given other variables, the optimization of ϕr is filter {uq }, active STAR-RIS beamforming Ψr and Ψt ,
equivalent to solve the problem that follows: and sum rate Rsum .
1: Initialize: W, Ψr , Ψt , t = 1, δ = ∞, Rsum = 0
P6 : min ϕH H
r Ω̂r ϕr + ℜ{2ϕr (−µ̂r )} (44a) 2: while t ⩽ Qmax and δ ⩾ δth do
ϕr
3: Rpre = Rsum ;
s.t. |ϕr,n | = 1, ∀n. (44b) 4: Update γkopt and ρopt k , ∀k ∈ K by (14) and (15);
5: Update radar receiver {uq } by solving (16);
The presence of the unit modulus constraint in (44b) 6: Update DFBS beamforming W by solving P3 ;
results in P6 a non-convex optimization problem. Fortu- 7: Update a by solving P5
nately, two methods have been discussed in [52] to solve this 8: Update Ψr and Ψt by solving P6 and P7 ;
problem, i.e., MM and CCM. For a more comprehensive un- 9: Calculate Φr = diag(Φr a) and Φt = diag(Φt a);
derstanding, readers can consult [52] for more information. 10: Update Rsum by (6);
|R −R |
By repeating the same procedure, ϕt can also be opti- 11: δ = preRsum sum ;
mized. Specifically, by defining: 12: t = t + 1.
X p 13: end while
µ̂t ≜ 1 + γk diag(ρ∗k qt )Gwk
k∈Kt
K+M (45)
X 3) SD mode
− |ρk |2 diag(qt )G wj wH
j hd,k , In SD mode, elements of active STAR-RIS are segregated
j=1
into two distinct groups, with one group operating in reflec-
X
Ω̂t ≜ |ρk |2 σv2 diag(fH tion mode and the other group operating in transmission
k )diag(fk )+
k∈Kt mode, i.e., the corresponding amplitude coefficient ar,n &
K+M (46) at,n is fixed as 1 or 0. The algorithm for this mode can
be obtained with the same algorithmic framework in EED
X
H
|ρk |2 diag(qt )Gwj wH H
j G diag(qt ),
j=1
mode as above with fixing the amplitude coefficient as 1 or

November 15, 2024 DRAFT

 11

TABLE II
S YSTEM PARAMETERS

Paramater Value
The number of antennas at DFBS M =8
The number of active STAR-RIS
N = 64
elements
The number of communication users K=4
The number of targets Q=2
τBQ = τRU,k = 1.8, τBR = 3,
Path loss exponents
τBU,k = 3.5, ∀k ∈ K
The path loss at 1 m L0 = 30dB
Rician factor κ=1
total
The total transmit power budget Pmax = 16dBm
σk2 = σv2 = σz2 = −80dBm,
The noise power
∀k ∈ K
The target radar cross section ξ2 = 1
The sensing SINR requirement Γt = 5dB

and STAR-RIS, a three-dimensional coordinate system is

Fig. 3. Simulation setup for active STAR-RIS-assisted ISAC system. utilized. Concretely, the DFBS is positioned at the origin
(0, 0, 0) meters(m), while the active STAR-RIS is posi-
tioned at (0, 15, 0)m. On both sides of the active STAR-
RIS, assuming two users are randomly distributed within
0. To make this paper more concise, we have omitted the
a semicircular area with a radius of 10m centred around
details here.
the active STAR-RIS, i.e., K = 4. The two targets are
assumed to be hovering within a circular area of a radius
F. Analysis of Computational Complexity with 5m, centred around DFBS. We set the number of
Based on the above proposed method, the joint DFBS DFBS transmitting and receiving antennas to be the same,
beamforming and active STAR-RIS beamforming is straight- i.e., M = 8. The number of active STAR-RIS elements
forward. Next, we give a brief complexity analysis of the is modeled as N = 64, and the total transmit power
total
proposed algorithm. In each iteration, the computational budget is se to Pmax = 16dBm. For simplicity, we set
2
complexity of updating γ is O(K 2 N 2 ). The complexity of the RCS as ξ = 1, the sensing SINR constraint for all
updating ρ is O(K(K +1)N 2 ). The complexity of updating targets as Γt = 5dB, and all noise power as the same, i.e.,
the receiver filter {uq } is O(QM 3 ). Given the complexity σk2 = σv2 = σz2 = −80dBm, ∀k ∈ K. Following [32], we
of finding a solution for the standard convex QCQP prob- employ the Rician channel model » for all channels
» except
κ 1
lem, the computational√ complexity of updating the transmit the SI channel with H = PL( κ+1 HLoS + κ+1 HNLoS ),
beamformer W is O( M K + 2(1 + M K)M 3 K 3 )) and
»
where PL = L0 ( dd0 )−τ , τ ∈ {τBR , τBQ , τRU,k , τBU,k },
updating the√ active STAR-RIS precoding matrix Ψr and indicates the path loss associated with H, d indicates the
Ψt is O(2 N + 1(1+2)N 3 )). Similarly,
√ the computational distance between two devices, d0 = 1m is the path loss of
complexity of updating A is O( N + 1(1 + 2)N 3 )). For reference distance, and κ = 1 is the Rician factor. For the
EED mode, the complexity of MM or CCM is given residual SI channel HSI at the DFBS, we follow [45] to
by O(N 3 + Io N 2 ), where Io represents the number of model it. The specific parameters setting can be found in
iterations necessary for the two algorithms to converge. TABLE II. Unless otherwise stated, the above experimental
Thus, the overall computational complexity of Algorithm parameters remain unchanged throughout our simulations.
1 is approximated as the order of O(Ic1 (K 2 N 2 + QM 3 + To verify the effectiveness of our proposed schemes, we
M 4.5 K 4.5 +N 4.5 +Io N 2 )), where Ic1 represents the number conduct a comparison with the following baseline:
of iterations necessary for Algorithm 1 to achieve conver-
• The passive STAR-RIS-assisted ISAC system scheme
gence. Similarly, the overall computational complexity of
(legend “passive STAR-RIS ES”): As verified by [28],
Algorithm 2 is approximated as the order of O(Ic2 (K 2 N 2 +
[29], the optimal mode for passive STAR-RIS is energy
QM 3 +M 4.5 K 4.5 +N 4.5 +Io N 2 )), where Ic2 represents the
splitting (ES), which means that the transmission and
number of iterations necessary for Algorithm 2 to achieve
reflection coefficients of each STAR-RIS element can
convergence.
be optimized, and a high degree of flexibility for system
design is enabled. It is the same as the UED mode in
IV. N UMERICAL R ESULTS active STAR-RIS. Therefore, we adopt the ES mode for
In this section, simulation results are conducted to evalu- passive STAR-RIS.
ate the performance of the proposed schemes. As depicted Fig. 4 shows the convergence of the proposed algorithms.
in Fig. 3, in order to accurately depict the locations of DFBS In order to ensure a fair comparison with the case of the

November 15, 2024 DRAFT

 12

Fig. 4. Achievable sum-rate versus the number of iterations. total .

Fig. 6. Achievable sum-rate versus total power budget Pmax

Fig. 5. Achievable sum-rate versus the different power allocations between Fig. 7. Achievable sum-rate versus Rader SINR Γt with different power
DFBS and active STAR-RIS. allocations between DFBS and active STAR-RIS.

total addition, one can observe that the sum-rate with UED mode
passive STAR-RIS, we define the total power budget Pmax =
B R is the highest, while the passive STAR-RIS owns the lowest
Pmax + Pmax . Thus, for passive scheme, the total power
total B sum-rate.
denotes the the DFBS transmit power, i.e., Pmax = Pmax .
For our proposed schemes, we set the power budget at We plot Fig. 5 to show the relationship between achievable
B total
DFBS and active STAR-RIS to Pmax = 0.4 × Pmax and sum-rate and different power allocations between DFBS and
R total
Pmax = 0.6×Pmax , respectively. Note that this kind of power active STAR-RIS. The number on x axis, for example, 0.4
total
allocation is just a case to verify the convergence and effi- represents that under a certain total power budget Pmax , the
R total
ciency of our proposed algorithms. Actually, the achievable power allocated to active STAR-RIS is Pmax = 0.4 × Pmax ,
B total
sum-rate will be different for different allocations, but this and the power allocated to DFBS is Pmax = 0.6 × Pmax .
does not affect the effectiveness of our proposed algorithms. One can observe that when the system’s total power budget
We notice in Fig. 4 that the sum-rate first experiences an is given, different power allocations will lead to different
upward trend and subsequently keeps stabilization as the performance, and there exists an optimal power allocation
number of iterations increases under all schemes, which for each mode. Meanwhile, we can also find that the power
validates the effectiveness of our proposed algorithms. In allocation with middle ratio can obtain a better performance

November 15, 2024 DRAFT

 13

Fig. 8. Achievable sum-rate versus active STAR-RIS elements N . Fig. 9. Achievable sum-rate versus the horizontal distance of the active
STAR-RIS from the DFBS.

under UED and EED modes, because this ratio can not
only satisfy the detection requirements of DFBS but also the sum rate under all schemes decreases with radar SINR
fully leverage the amplification characteristics of the active increases. This can be explained, as more power will be
STAR-RIS and its ability to counteract the multiplicative used for target sensing under a higher radar SINR, and thus,
fading effect. For SD mode, due to the low DoFs of active the power used for communication will be reduced, leading
STAR-RIS elements, it needs more power consumption at to a decrease in the sum-rate. This observation highlights
the DFBS to overcome the double fading. Additionally, it the tradeoff between communication and sensing within
can be observed that the performance of the UED mode is the limited resources, and the realization of high sensing
optimal, which once again confirms the UED mode owns performance comes at the cost of diminished communication
the highest DoFs. performance. In addition, the active STAR-RIS can always
Fig. 6 shows the achievable sum-rate versus total power achieve a performance improvement compared to the passive
budget Pmaxtotal
. We can note that the sum-rate exhibits a one, demonstrating the significant role of deploying active
gradual rise with the total power budget increasing un- STAR-RIS in ISAC systems and the effectiveness of the
der all schemes, which is easy to understand. Obviously, proposed algorithms.
the UED scheme owns the highest sum-rate compared to Fig. 8 shows the sum-rate versus the number of STAR-
other schemes. This can be explained as follows: When RIS elements. It is obvious that the increase in the number of
active STAR-RIS operates in UED mode, each element STAR-RIS elements leads to an enhancement in the sum-rate
can be dynamically adjusted based on the given power under all schemes. This is easy to understand, more STAR-
of the active STAR-RIS. This allows active STAR-RIS to RIS elements can obtain a higher beamforming gain, en-
divide the incident signals upon each element flexibly into hancing the sum-rate. One can also observe that the sum-rate
transmission and reflection parts with different amplitudes improvement by the active schemes is much higher than that
based on actual conditions, thus giving this mode the highest for the passive scheme. This indicates that, unlike passive
DoFs. Meanwhile, due to the multiplicative fading effect, the scheme where all transmit power suffers from multiplicative
performance of the passive scheme is the worst compared to fading of the reflected/transmitted links, the power of active
other schemes, demonstrating the effectiveness of deploying STAR-RIS only experiences fading of the active STAR-RIS-
the active STAR-RIS in ISAC systems. user link, thus the power attenuation is much smaller and can
Fig. 7 shows the achievable sum rate versus radar SINR effectively compensate for multiplicative fading effects. This
under different power allocations between DFBS and active also confirms that active STAR-RIS can achieve the same
STAR-RIS. The number “4:6” indicates that with the given performance as the passive one with much fewer elements,
total which agrees with our analysis in Section I.
total power budget Pmax , the power allocated to DFBS is
B total
Pmax = ×Pmax and the power allocated to active STAR- Finally, in order to investigate the influence of the STAR-
R total
RIS is Pmax = ×Pmax , and the same goes for “2:8”. As RIS location on the sum-rate, we analyze the achievable
clearly seen, with the same radar SINR constraint, the “4:6” sum-rate in relation to the horizontal distance between the
allocation method obtained a better performance, consistent STAR-RIS and the DFBS in Fig. 9. In our considered sce-
with our previous analysis. Moreover, one can find that nario, it can be found that when the distance between DFBS

November 15, 2024 DRAFT

 14

and STAR-RIS is increased, the system is able to function [4] R. Liu, M. Liu, H. Luo, Q. Liu, and A. L. Swindlehurst, “Integrated
normally. However, in practical systems, we need to consider sensing and communication with reconfigurable intelligent surfaces:
Opportunities, applications, and future directions,” IEEE Wireless
the requirements of the actual situation. If the system’s Commun., vol. 30, no. 1, pp. 50–57, Feb. 2023.
requirements for sum-rate are relatively low, then STAR- [5] Z. Zhu, Z. Li, Z. Chu, G. Sun, W. Hao, P. Liu, and I. Lee, “Resource
RIS deployment can be far from DFBS. Conversely, if the allocation for intelligent reflecting surface assisted wireless powered
IoT systems with power splitting,” IEEE Trans. Wireless Commun.,
requirements are high, it needs to be closer to DFBS. Taking vol. 21, no. 5, pp. 29878–2998, May. 2022.
our simulation as an example, if the required communication [6] C. Huang, Z. Yang, G. C. Alexandropoulos, K. Xiong, L. Wei, C.
rate of the system cannot be lower than 45bps/Hz, then only Yuen, Z. Zhang and M. Debbah, “Multi-hop RIS-empowered terahertz
communications: a DRL-based hybrid beamforming design,” IEEE J.
when the distance between DFBS and STAR-RIS is less than Sel. Area Comm., vol. 39, no. 6, pp. 1663–1677, Jun. 2021.
20m, the three modes of active STAR-RIS can be effective. [7] W. Yan, W. Hao, C. Huang, G. Sun, O. Muta, H. Gacanin, C. Yuen,
In addition, as clearly observed, the sum-rate decreases as “Beamforming analysis and design for wideband THz reconfigurable
Intelligent surface communications,” IEEE J. Sel. Area Comm., vol.
the distance between DFBS and STAR-RIS increases. This 41, no. 8, pp. 2306–2320, Aug. 2023.
can be attributed to the increased path loss experienced by [8] Z. Zhou, N. Ge, Z. Wang, and L. Hanzo “Joint transmit precoding and
the DFBS-RIS link as the distance between STAR-RIS and reconfigurable intelligent surface phase adjustment: a decomposition-
aided channel estimation approach,” IEEE Trans. Commun., vol. 69,
DFBS increases. For passive STAR-RIS, it will be affected no. 2, pp. 1228–1243, Feb. 2021.
by significant path loss when the STAR-RIS is located at a [9] Z. Wang, L. Liu, and S. Cui, “Channel estimation for intelligent
considerable distance from the DFBS. However, in the case reflecting surface assisted multiuser communications: Framework,
algorithms, and analysis,” IEEE Trans. Wireless Commun., vol. 19,
of active STAR-RIS, a portion of the total power budget is no. 10, pp. 6607–6620, Oct. 2020.
assigned to the active STAR-RIS that is only impacted by [10] N. Li, W. Hao, F. Zhou, Z. Chu, S. Yang, O. Muta, and H. Gacanin,
the slight path loss between the active STAR-RIS and user “Min-max latency optimization for IRS-aided cell-free mobile edge
computing systems,” IEEE Internet Things J., early access, Oct. 2023.
link, which can compensate for significant path loss between [11] W. Hao, G. Sun, M. Zeng, Z. Chu, Z. Zhen, O. A. Dobre, and P.
DFBS and STAR-RIS. Thus, the active STAR-RIS exhibits Xiao, “Robust design for intelligent reflecting surface-assisted MIMO-
the capability to achieve a higher sum-rate compared to the OFDMA terahertz IoT networks,” IEEE Internet Things J., vol. 8, no.
16, pp. 13052–13064, Aug. 2021.
passive STAR-RIS. [12] W. Hao, J. Li, G. Sun, M. Zeng, and O. A. Dobre, “Securing
reconfigurable intelligent surface-aided cell-free networks,” IEEE T.
Inf. Foren. Sec., vol. 17, pp. 3720–3733, Oct. 2022.
V. C ONCLUSION [13] Z. Zhu, Z. Li, Z. Chu, Q. Wu, J. Liang, Y. Xiao, P. Liu, and I.
In this paper, we have investigated an ISAC system Lee, “Intelligent reflecting surface-assisted wireless powered hetero-
geneous networks,” IEEE Trans. Wireless Commun., vol. 22, no. 12,
that utilizes active STAR-RIS to overcome the challenges pp. 9881–9892, Dec. 2023.
of multiplicative fading and limited coverage. Meanwhile, [14] Z. M. Jiang, M. Rihan, P. Zhang, L. Huang, Q. Deng, J. Zhang, and
we have considered multiple targets and multiple users E. M. Mohamed, “Intelligent reflecting surface aided dual-function
radar and communication system,” IEEE Syst. J., vol. 16, no. 1, pp.
scenario and assumed the existence of SI. In particular, we 475–486, Mar. 2022.
have formulated an optimization problem with the goal of [15] R. Liu, M. Liu, and A. L. Swindlehurst, “Joint beamforming and
maximizing the achievable communication sum-rate, while reflection design for RIS-assisted ISAC systems,” in Proc. of the 2022
30th European Signal Processing Conf. (EUSIPCO), Belgrade, Serbia,
also considering the requirement of sensing, the power Oct. pp. 1–5, 2022.
budget and the active STAR-RIS hardware constraints. An [16] R. Liu, M. Li, Y. Liu, Q. Wu, and Q. Liu, “Joint transmit waveform
AO algorithm have been proposed to efficiently solve the and passive beamforming design for RIS-aided DFRC systems,” IEEE
J. Sel. Topics Signal Process., vol. 16, no. 5, pp. 995–1010, Aug. 2022.
formulated problem by decomposing it into several tractable [17] X. Song, D. Zhao, H. Hua, T. Xiao Han, X. Yang, and J. Xu,
subproblems. Through fixing variables, these subproblems “Joint transmit and reflective beamforming for IRS-assisted integrated
can be easily tackled with standard convex optimization sensing and communication,” in Proc. of the 2022 IEEE Wireless
Commun. Netw. Conf. (WCNC), Austin, TX, USA, May. pp. 1–6,
algorithms and state-of-the-art algorithms. Simulation results 2022.
have shown the potential of deploying active STAR-RIS in [18] H. Luo, R. Liu, M. Li, and Q. Liu, “RIS-aided integrated sensing
ISAC system and the efficiency of the proposed joint beam- and communication: joint beamforming and reflection design,” IEEE
Trans. Veh. Technol., vol. 72, no. 7, pp. 9626–9630, Mar. 2022.
forming optimization algorithms. In future, we will explore [19] K. Zhong, J. Hu, C. Pan, M. Deng and J. Fang, “Joint waveform
the scenario based on imperfect channel state information to and beamforming design for RIS-aided ISAC systems,” IEEE Signal
enhance the practicality of our research. Processing Lett., vol. 30, pp. 165–169, Feb. 2023.
[20] Y. Mai, and H. Du, “Joint beamforming and phase shift design for
RIS-aided ISAC system,” in Proc. of the 2022 IEEE 8th International
R EFERENCES Conference on Computer and Commun. (ICCC), Chengdu, China,
Mar. pp. 1–6, 2023.
[1] Z. Feng, Z. Fang, Z. Wei, X. Chen, Z. Quan, and D. Ji. Sira-Ramirez, [21] M. Hua, Q. Wu, C. He, S. Ma, and W. Chen, “Joint active and
“Joint radar and communication: A survey,” China Communications, passive beamforming design for IRS-aided radar-communication,”
vol. 17, no. 1, pp. 1–27, Jan. 2020. IEEE Trans. Wireless Commun., vol. 22, no. 4, pp. 2278–2294, Apr.
[2] X. Tong, Z. Zhang, J. Wang, C. Huang and M. Debbah, “Joint 2023.
multi-user communication and sensing exploiting both signal and [22] Z. Xing, R. Wang, and X. Yuan, “Joint active and passive beamform-
environment sparsity,” IEEE J. Sel. Topics Signal Process, vol. 15, ing design for reconfigurable intelligent surface enabled integrated
no. 6, pp. 1409–1422, Nov. 2021. sensing and communication,” IEEE Trans. Commun., vol. 71, no. 4,
[3] F. Liu, C. Masouros, A. P. Petropulu, H. Griffiths, and L. Hanzo, pp. 2457–2474, Apr. 2023.
“Joint radar and communication design: Applications, state-of-the- [23] Y. Liu et al., “STAR: Simultaneous transmission and reflection for
art, and the road ahead,” IEEE Trans. Commun., vol. 68, no. 6, pp. 360◦ coverage by intelligent surfaces,” IEEE Wireless Commun., vol.
3834–3862, Jun. 2020. 28, no. 6, pp. 102–109, Dec. 2021.

November 15, 2024 DRAFT

 15

[24] X. Mu, Y. Liu, L. Guo, J. Lin, and R. Schober, “Simultaneously trans- Proc. of the GLOBECOM 2022 - 2022 IEEE Global Communications
mitting and reflecting (STAR) RIS aided wireless communications,” Conference, Rio de Janeiro, Brazil, pp. 1–6, Jan. 2023.
IEEE Trans. Wireless Commun., vol. 21, no. 5, pp. 3083–3098, May. [46] Z. Xiao, and Y. Zeng, “Waveform design and performance analysis
2022. for full-duplex integrated sensing and communication,” IEEE J. Select.
[25] S. Zhang et al., “Intelligent omni-surfaces: Ubiquitous wireless trans- Areas Commun., vol. 40 no. 9, pp. 1823–1837, Jun. 2022.
mission by reflective-refractive metasurfaces,” IEEE Trans. Wireless [47] Z. He, W. Xu, H. Shen, D. Ng, Y. C. Eldar, and X. You, “Full-
Commun., vol. 21, no. 1, pp. 219–233, Jan. 2022. duplex communication for ISAC: joint beamforming and power
[26] W. Cai, M. Li, Y. Liu, L. Q. Wu, and Q. Liu, “Joint beamforming optimization,” IEEE J. Select. Areas Commun., vol. 41, no. 9, pp.
design for intelligent omni surface assisted wireless communication 2920–2936, Sep. 2023.
systems,” IEEE Trans. Wireless Commun., vol. 22, no. 2, pp. 1281– [48] K. Shen and W. Yu, “Fractional programming for communication
1297, Feb. 2023. systems-Part I: Power control and beamforming,” IEEE Trans. Signal
[27] Z. Wang, X. Mu, and Y. Liu, “STARS enabled integrated sensing and Process., vol. 66, no. 10, pp. 2616–2630, May. 2018.
communications,” IEEE Trans. Wireless Commun., vol. 22, no. 10, [49] K. Shen and W. Yu, “Fractional programming for communication
pp. 6750–6765, Feb. 2023. systems-Part II: Uplink scheduling via matching,” IEEE Trans. Signal
[28] H. Luo, R. Liu, M. Li, and Q. Liu, “Toward STAR-RIS-empowered Process., vol. 66, no. 10, pp. 2631–2644, May. 2018.
integrated sensing and communications: Joint active and passive [50] Y. Sun, P. Babu, and D. P. Palomar, “Majorization-minimization algo-
beamforming design,” IEEE Trans. Veh. Technol., vol. 72, no. 12, rithms in signal processing, communications, and machine learning,”
pp. 15991–16005, Dec. 2023. IEEE Trans. Signal Process., vol. 65, no. 3, pp. 794–816, Feb. 2017.
[29] Z. Liu, X. Li, H. Ji, and H. Zhang, “Exploiting STAR-RIS for physical [51] A. Ben-Tal and A. Nemirovski, Lectures on Modern Convex Optimiza-
layer security in integrated sensing and communication networks,” in tion: Analysis, Algorithms, and Engineering Applications, Philadel-
Proc. of the IEEE 34th Annual International Symposium on Personal, phia, PA, USA: for Industrial and Applied Mathematics, 2001.
Indoor and Mobile Radio Commun., Jun. 2023, pp. 1–8. [52] C. Pan et al., “Multicell MIMO communications relying on intelligent
[30] W. Chao, C. Wang, Z. Li, D. Ng, G, and K. Wong, “PHY se- reflecting surfaces,” IEEE Trans. Wireless Commun., vol. 19, no. 8,
curity enhancement exploiting STAR-RIS for dual-functional radar- pp. 5218–5233, Aug. 2020.
communication,” in Proc. of 2023 IEEE International Conference on
Communications Workshops (ICC Workshops), Oct. 2023, pp. 1–5.
[31] Z. Zhu, M. Gong, Y. Liu, Z. Chu, P. Xiao, G. Sun, D. Mi, Z. He,
and F. Tong, “DRL-based STAR-RIS-assisted ISAC secure commu-
nications,” in Proc. of the International Conference on Ubiquitous
Commun. 2023 (Ucom 2023), Xi’an, China, Jun. 2023, pp. 1–6.
Shuang Zhang received the M.S. degree in Com-
[32] Z. Zhang, L. Dai, X. Chen, C. Liu, F. Yang, R. Schober, and H. V.
munication and Information System from Xidian
Poor, “Active RIS vs. passive RIS: Which will prevail in 6G?,” IEEE
University, Xi’an, China, in 2017. She is cur-
Trans. Commun., vol. 71, no. 3, pp. 1707–1725, Mar. 2023.
rently pursuing the Ph.D. degree in information
[33] Q. Zhu, M. Li, R. Liu, and Q. Liu, “Joint transceiver beamforming and communication engineering with Zhengzhou
and reflecting design for active RIS-aided ISAC systems,” IEEE Trans. University, Zhengzhou, China. Her research inter-
Veh. Technol., vol. 72, no. 7, pp. 9636–9640, Jul. 2023. ests include integrated sensing and communica-
[34] Y. Zhang, J. Chen, C. Zhong, H. Peng, and W. Lu, “Active IRS- tion (ISAC) and reconfigurable intelligent surface
assisted integrated sensing and communication in C-RAN,” IEEE (RIS).
Wireless Commun. Lett., vol. 12, no. 3, pp. 411–415, Mar. 2023.
[35] A. A. Salem, M. H. Ismail, and A. S. Lbrahim, “Active reconfigurable
intelligent surface-assisted MISO integrated sensing and communica-
tion systems for secure operation,” IEEE Trans. Veh. Technol., vol.
72, no. 4, pp. 4919–4931, Apr. 2023.
[36] Y. Chen, Y. Wang, Z. Wang, and P. Zhang, “Robust beamforming for
active reconfigurable intelligent omni-surface in vehicular communi-
cations,” IEEE J. Sel. Areas Commun., vol. 40, no. 10, pp. 3086–3103, Wanming Hao (Senior Member, IEEE) received
Oct. 2022. the Ph.D. degree from School of Electrical and
[37] H. Luo, L. Lv, Q. Wu, Z. Ding, N. A. Dhahir, and J Chen, “Beam- Electronic Engineering, Kyushu University, Japan,
forming design for active IOS aided NOMA networks,” IEEE Wireless in 2018. He worked as a Research Fellow at the
Commun. Lett., vol. 12, no. 2, pp. 282–286, Feb. 2023. 5G Innovation Center, Institute of Communication
[38] Y. Ma, M. Li, Y. Liu, Q. Wu, and Q. Liu, “Optimization for reflection Systems, University of Surrey, U.K. Now, he is an
and transmission dual-functional active RIS-assisted systems,” IEEE Associate Professor at the School of Electrical and
Trans. Commun., vol. 71, no. 9, pp. 5534–5548, Jun. 2023. Information Engineering, Zhengzhou University,
[39] W. Wang, W. Ni, and H. Tian, “Multi-functional RIS-aided wireless China. His research interests include millimeter-
communications,” IEEE Internet Things J., vol.10, no, 23, pp.21133– wave and RIS. He is the editor of IEEE O-JCS.
21134, Dec. 2023.
[40] Y. Guo, Y. Liu, Q. Wu, Q. Shi, and Y. Zhao, “Enhanced secure
communication via novel double-faced active RIS,” IEEE Trans.
Commun., vol. 71, no. 6, pp. 3497–3512, Jun. 2023.
[41] Y. Liu, Y. Ma, M. Li, Q. Wu, and Q. Shi, “Spectral efficiency max-
imization for double-faced active reconfigurable intelligent surface,”
IEEE Trans. Signal Process., vol. 70, pp. 5397–5412, Nov. 2022.
Gangcan Sun received his Ph.D. degree in Com-
[42] A. Zheng, W. Ni, W. Wang, and H. Tian, “Enhancing NOMA networks
munication and Information System from Beijing
via reconfigurable multi-functional surface,” IEEE Commun. Lett.,
Institute of Technology, China, in 2009. He was
vol.27, no, 4, pp.1195–1199, Apr. 2023.
promoted to an associate professor in School of
[43] H. Luo, R. Liu, M. Li, Y. Liu, and Q. Liu, “Joint beamforming design Information Engineering, Zhengzhou University,
for RIS-assisted integrated sensing and communication systems,” China, in 2013, and now he has a full professor.
IEEE Trans. Veh. Technol., vol. 71, no. 12, pp. 13393–13397, Dec. His research interests include communication sig-
2022. nal processing, communication modulation recog-
[44] S. Buzzi, E. Grossi, M. Lops, and L. Venturino, “Foundations of nition and blind estimation of key parameters of
MIMO radar detection aided by reconfigurable intelligent surfaces,” communication signals.
IEEE Trans. Signal Process., vol.70, pp.1749–1763, Mar. 2022.
[45] E. Balti, C. Dick, and B.L. Evans, “Low complexity hybrid beam-
forming for mmWave full-duplex integrated access and backhaul,” in

November 15, 2024 DRAFT

 16

Chongwen Huang (Member, IEEE) received his Chau Yuen (Fellow, IEEE) received the [Link].
[Link]. degree in 2010 from Nankai University, and and Ph.D. degrees from Nanyang Technological
the [Link]. degree from the University of Electronic University, Singapore, in 2000 and 2004, respec-
Science and Technology of China in 2013, and tively. He was a Post-Doctoral Fellow with Lucent
PhD degree from Singapore University of Tech- Technologies Bell Labs, Murray Hill, in 2005.
nology and Design (SUTD) in 2019. From Oct. From 2006 to 2010, he was with the Institute
2019 to Sep. 2020, he is a Postdoc in SUTD. Since for Infocomm Research, Singapore. From 2010
Sep. 2020, he joined into Zhejiang University to 2023, he was with the Engineering Product
as a tenure-track young professor. Dr. Huang is Development Pillar, Singapore University of Tech-
the recipient of 2021 IEEE Marconi Prize Paper nology and Design. Since 2023, he has been with
Award, 2023 IEEE Fred W. Ellersick Prize Paper the School of Electrical and Electronic Engineer-
Award and 2021 IEEE ComSoc Asia-Pacific Outstanding Young Researcher ing, Nanyang Technological University, currently he is Provost’s Chair in
Award. He has served as an Editor of IEEE Communications Letter, Elsevier Wireless Communications, and Assistant Dean in Graduate College.
Signal Processing, EURASIP Journal on Wireless Communications and Dr. Yuen received IEEE Communications Society Leonard G. Abraham
Networking and Physical Communication since 2021. His main research Prize (2024), IEEE Communications Society Best Tutorial Paper Award
interests are focused on Holographic MIMO Surface/Reconfigurable Intel- (2024), IEEE Communications Society Fred W. Ellersick Prize (2023),
ligent Surface, B5G/6G Wireless Communications, mmWave/THz Commu- IEEE Marconi Prize Paper Award in Wireless Communications (2021),
nications, Deep Learning technologies for Wireless communications, etc. IEEE APB Outstanding Paper Award (2023), and EURASIP Best Paper
Award for JOURNAL ON WIRELESS COMMUNICATIONS AND NET-
WORKING (2021).
Dr. Yuen current serves as an Editor-in-Chief for Springer Nature
Computer Science, Editor for IEEE TRANSACTIONS ON VEHICULAR
TECHNOLOGY, IEEE SYSTEM JOURNAL, and IEEE TRANSACTIONS
ON NETWORK SCIENCE AND ENGINEERING, where he was awarded
as IEEE TNSE Excellent Editor Award and Top Associate Editor for
Zhengyu Zhu (Senior Member, IEEE) received
TVT from 2009 to 2015. He also served as the guest editor for several
the Ph.D. degree in information engineering
special issues, including IEEE JOURNAL ON SELECTED AREAS IN
from Zhengzhou University, Zhengzhou, China,
COMMUNICATIONS, IEEE WIRELESS COMMUNICATIONS MAG-
in 2017. He is currently an Associate Professor
AZINE, IEEE COMMUNICATIONS MAGAZINE, IEEE VEHICULAR
with Zhengzhou University. From October 2013 to
TECHNOLOGY MAGAZINE, IEEE TRANSACTIONS ON COGNITIVE
October 2015, he visited the Communication and
COMMUNICATIONS AND NETWORKING, and ELSEVIER APPLIED
Intelligent System Laboratory, Korea University,
ENERGY.
Seoul, South Korea, to conduct a collaborative re-
He is a Distinguished Lecturer of IEEE Vehicular Technology Society,
search as a Visiting Student. His research interests
Top 2% Scientists by Stanford University, and also a Highly Cited Re-
include in formation theory and signal processing
searcher by Clarivate Web of Science. He has 4 US patents and published
for wireless communications, such as B5G/6G,
over 400 research papers at international journals.
Intelligent reflecting surface, the Internet of Things, machine learning,
millimeter wave communication, UAV communication, physical layer secu-
rity, convex optimization techniques, and energy harvesting communication
systems. He was an Associate Editor for the Journal of Communications
and Networks, the Wireless Communications and Mobile Computing, and
the Physical Communications from 2021.

Xingwang Li (Senior Member, IEEE) received his

[Link]. and Ph.D. degrees from University of Elec-
tronic Science and Technology of China and Bei-
jing University of Posts and Telecommunications
in 2010 and 2015. From 2010 to 2012, he was
working with Comba Telecom Ltd. in Guangzhou
China, as an engineer. He spent one year from
2017 to 2018 as a visiting scholar at Queen’s
University Belfast, Belfast, UK. He is currently an
Associated Professor with the School of Physics
and Electronic Information Engineering, Henan
Polytechnic University, Jiaozuo China. His research interests span wireless
communication, intelligent transport system, artificial intelligence, Internet
of things. He was a recipient of exemplary Reviewer for IEEE Transactions
on Communications and Journal of Electronics & Information Technology
in 2022. He is on the editorial board of IEEE Transactions on Intelligent
Transportation Systems, IEEE Transactions on Vehicular Technology, IEEE
Systems Journal, IEEE Sensors Journal, Physical Communication, etc. He
has serviced as the Guest Editor for the special issue on Integrated Sensing
and Communications (ISAC) for 6G IoE of IEEE Internet of Things Journal,
Computational Intelligence and Advanced Learning for Next-Generation
Industrial IoT of IEEE Transactions on Network Science and Engineering,
“AI driven Internet of Medical Things for Smart Healthcare Applications:
Challenges, and Future Trends” of the IEEE Journal of Biomedical and
Health Informatics, etc.

November 15, 2024 DRAFT