Content uploaded by Shihang Lu
Author content
All content in this area was uploaded by Shihang Lu on Feb 21, 2024
Content may be subject to copyright.
1
Integrated Sensing and Communications:
Recent Advances and Ten Open Challenges
Shihang Lu, Fan Liu, Senior Member, IEEE, Yunxin Li, Kecheng Zhang, Hongjia Huang, Jiaqi Zou, Xinyu Li,
Yuxiang Dong, Fuwang Dong, Jia Zhu, Yifeng Xiong, Member, IEEE, Weijie Yuan, Member, IEEE, Yuanhao
Cui, Member, IEEE and Lajos Hanzo, Life Fellow, IEEE
Abstract—It is anticipated that integrated sensing and com-
munications (ISAC) would be one of the key enablers of next-
generation wireless networks (such as beyond 5G (B5G) and
6G) for supporting a variety of emerging applications. In this
paper, we provide a comprehensive review of the recent advances
in ISAC systems, with a particular focus on their foundations,
physical-layer system design, networking aspects and ISAC
applications. Furthermore, we discuss the corresponding open
questions of the above that emerged in each issue. Hence, we
commence with the information theory of sensing and commu-
nications (S&C), followed by the information-theoretic limits of
ISAC systems by shedding light on the fundamental performance
metrics. Next, we discuss their clock synchronization and phase
offset problems, the associated Pareto-optimal signaling strate-
gies, as well as the associated super-resolution physical-layer
ISAC system design. Moreover, we envision that ISAC ushers
in a paradigm shift for the future cellular networks relying on
network sensing, transforming the classic cellular architecture,
cross-layer resource management methods, and transmission
protocols. In ISAC applications, we further highlight the security
and privacy issues of wireless sensing. Finally, we close by
studying the recent advances in a representative ISAC use case,
namely the multi-object multi-task (MOMT) recognition problem
using wireless signals.
Index Terms—Integrated sensing and communications, 6G,
performance limitations, system design, network and application,
Internet of Things (IoT).
I. INTRODUCTION
A. Potential Drive of Integrated Sensing and Communications
GIVEN the rapid roll-out of the 5G network, numer-
ous beyond-5G (B5G) and 6G concepts have emerged
[1]–[6]. There is a flood of emerging applications as well,
such as Internet of Things (IoT) networks [7]–[11], vehicle-
to-everything (V2X) communications [12]–[15], connected
(Corresponding authors: Fan Liu and Yuanhao Cui.)
S. Lu, F. Liu, Y. Li, K. Zhang, Y. Dong, F. Dong, W. Yuan
and Y. Cui are with the School of System Design and Intelligent
Manufacturing (SDIM), Southern University of Science and Technology,
Shenzhen 518055, China (email: {lush2021@mail, liuf6, liyx2022@mail,
zhangkc2022@mail, dongyx2021@mail}.sustech.edu.cn; {dongfw, yuanwj,
cuiyh}@sustech.edu.cn).
H. Huang is with the Department of Electronic and Information Engineer-
ing, The Hong Kong Polytechnic University, Hong Kong 999077, China (e-
mail: hongj.huang@connect.polyu.hk).
X. Li is with the School of Information Science and Engineering, Southeast
University, Nanjing 210096, China (e-mail: xinyuli@seu.edu.cn).
J. Zou, J. Zhu, and Y. Xiong are with the School of Information and Com-
munication Engineering, Beijing University of Posts and Telecommunications,
Beijing 100876, China (e-mail: {jqzou, zhujia, yifengxiong}@bupt.edu.cn).
L. Hanzo is with the School of Electronics and Computer Sci-
ence, University of Southampton, Southampton SO17 1BJ, U.K. (e-
mail:lh@ecs.soton.ac.uk).
Target
Direct link
Comm
users
Clutter
Target
Radar interference
Downlink comm signal
Uplink comm signal Comm
users
Direct
link
Relay
Destination node
Broadcast
Multi-access
Monostatic Sensing
Cooperative Comm
Bistatic Sensing
Single-Cell Comm
Fig. 1. The topology of monostatic/bistatic deployments and
single-cell/cooperative scenarios.
autonomous systems [16], autonomous driving [17], human
activity sensing [18] as well as smart home and unmanned
aerial vehicle (UAV) networks [19], [20] that rely on sensing
functionalities, such as wireless localization in support of
compelling services [21]–[27]. Consequently, having a sensing
functionality is envisioned to become one of the basic services
in the B5G/6G networks as the next evolutionary stage beyond
the existing communication-only scenarios [28]–[35].
In the meantime, given the smooth evolution of wireless
communication systems, the increasingly congested spectral
resources tend to limit the throughput of future wireless
systems [36], [37]. As a remedy, the radar bands set aside for
sensing can be harnessed as one of the potential alternative
future bands. This exciting prospect has promoted the initial
integration of spectral resources of radar and communica-
tion systems, leading to the concept of radar-communication
coexistence (RCC) [26]. On the other hand, the millimeter
wave (mmWave) and terahertz (THz) bands envisaged for
next generation networks could also be exploited for sensing
in future cellular networks [38]–[43]. To this context, as
illustrated in Fig. 1, the generic radar sensing topology can
be categorized into monostatic and bistatic scenarios [44]–
[46], which are similar to the single-cell scenario [47], [48]
and cooperative communication scenarios of wireless systems
[9], [49], respectively.
As a benefit of intense research, sensing and communica-
tions (S&C) tend to become more integrated both in terms of
their hardware architecture and signal processing algorithms,
making the co-design of these systems more realistic, even
though in the past they have been considered as two isolated
arXiv:2305.00179v2 [cs.IT] 17 Dec 2023
2
Sec. V ISAC Applications
How could we provide security and privacy guarantees within ISAC applications?
How could we achieve MOMT recognition by relying on wireless signals?
Sec. II Theoretical Foundations of ISAC Systems
What are the information-theoretic limits of ISAC systems?
How much channel information can be inferred from the sensory data?
How could we quantify the integration and coordination gains?
Sec. III Physical-Layer System Design
How could we deal with the clock synchronization and phase offset both in the
bistatic and distributed deployments?
How far are the existing designs from the Pareto-optimal boundary?
How could we improve the sensing performance by the emerging super-
resolution methods?
Sec. IV ISAC Networks & Cross-Layer Design
What is the potential cellular architecture of future network of sensing?
How could we conceive resource management schemes and protocols tailored
specifically for the ISAC network?
Peeper
Ten Open Challenges & Corresponding Scenarios
2. Channel estimation from sensory data
4. Clock synchronization and phase offset
6. Super-resolution algorithm
1. Information-theoretic
limits of ISAC systems
7. New cellular architecture
3. Integration and coordination
gains quantification
5. Pareto-optimal boundary
8. Resource management
schemes and protocols for
ISAC networks
9. Security and privacy for wireless sensing
10. MOMT recognition by wireless signals
Theory
System
Network
Application
Fig. 2. Structure of this paper and the relevant ISAC scenarios.
fields [50]. Given the shared use of both the hardware and
wireless resources, radar and communication systems evolve
towards integrated sensing and communications (ISAC) sys-
tems [27], [29], [30]. In other words, the joint exploitation
of the limited hardware, spectral as well as energy resources,
results in a beneficial integration gain [27]. Moreover, sharing
resources between S&C leads to the compelling concept
of sensing-assisted communications [14], thus offering an
attractive coordination gain. Inspired by the aforementioned
benefits of integration, sensing-assisted communication and
communication-assisted sensing have attracted tremendous re-
search attention in recent years [51]–[59].
As illustrated in Fig. 2, recent advances of ISAC facilitate
the support of emerging IoT scenarios, such as smart cities
and smart homes [2], [30], [60]. Moreover, embracing ISAC
leads to a transformative shift in IoT architectures, elucidated
in the following two aspects..
1) Leveraging ISAC equips IoT devices with the ability to
“perceive and connect” with the world. This transformation
enables IoT devices to not only sense their environment but
also to establish connections with other devices and networks,
facilitating seamless data sharing and decision-making [3]. As
IoT ecosystems continue to evolve, the ability to “perceive
and connect” becomes a critical enabler for the successful
implementation of IoT applications [30]. To be more spe-
cific, the resilient sensing capacity of the distributed IoT
nodes plays a vital role in improving the performance of
networked sensing. Subsequently, networked collaborative IoT
nodes may be flexibly deployed to enhance environmental
sensing capabilities through decentralized transmission and
processing [27]. The multiple-source data fusion significantly
reduces sensing inaccuracies, thereby elevating the efficiency
of distributed sensing. To this end, sensing assignments can be
effectively distributed across multiple nodes, and the exchange
of “sensing-with-communication” findings promotes efforts to
reach a consensus regarding the surrounding environment [61].
Finally, IoT devices are equipped with the ability to “perceive
and connect” with the physical world within ISAC modes [27].
2) Utilizing ISAC meets the diverse demands of IoT
scenarios to support future networks. In future networks
supporting vertical industries, a multitude of IoT devices like
sensors and robots will be prevalent [7]. These devices rely on
precise sensing and positioning information to make decisions
and enable intelligent operations and management [9]. In a
large-scale connected sensing architecture, it is essential not
only to sense and detect the physical environment information
of target nodes but also to integrate this sensory data with
digital information [6]. Traditional wireless sensor networks
may struggle to meet this demand. Fortunately, ISAC allows
the convergence of sensing and communication functions by
the unified signals, addressing the limitations of conventional
wireless sensor networks and achieving a tight coupling be-
tween the physical and digital realms of network nodes [27]–
[30].
Note that the traditional IoT data processing flow falls
in the “communication-after-sensing” approach, while ISAC-
enabled IoT devices can sense environmental data and reduce
data-exchange time delays as well as the processing cost,
resulting in “sensing-with-communication” [27]. In large-scale
connected IoT scenarios, ISAC-enabled IoT devices surpass
3
the limitations of traditional wireless sensor networks, tightly
coupling the physical and digital aspects of IoT nodes [8],
[16], [20], [61]. However, numerous ISAC challenges systems
still exist, as detailed below.
B. Ten Open Challenges in ISAC Systems
Recently, ISAC has garnered recognition from the Interna-
tional Telecommunication Union (ITU) as a pivotal element
within the 6G landscape, serving as one of its six key usage
scenarios [62]. This acknowledgment underscores the growing
significance of ISAC for the evolution of next-generation wire-
less systems. Although substantial advances have been made,
ranging from determining the degrees of freedom tradeoffs,
the fundamental limits, coding design, resource allocation as
well as signaling strategies of ISAC systems [63]–[73], there
still remains a host of open questions in the fields of the
theoretical foundation of ISAC, physical-layer system design,
ISAC networking and ISAC applications, which motivates the
conception of this paper.
First of all, the fundamental theories of radar sensing and
wireless communication are challenging to be unified under
the joint concept of ISAC systems, since radar and wireless
communication systems glean different information from the
received signals [66], [76]–[79]. Specifically, radar systems
focus on how to minimize the uncertainty concerning the target
environments based on the received echo signals. By contrast,
communication systems concentrate on how to minimize the
uncertainty related to the transmitted random signals in order
to recover useful information [80]–[83]. In this spirit, it is
of pivotal significance to resolve what the prime objective of
ISAC systems is from an information-theoretic perspective.
This will provide reliable theoretical guidance for the design
of ISAC systems. Therefore we commence with the salient
question Challenge 1) What are the information-theoretic
limits of ISAC systems? Then we will discuss the intricate
relationships among those limitations. The ultimate long-term
objective would be to formulate pertinent multi-component
objective functions unifying both communications and radar
metrics and finding the optimal non-dominated Pareto front
[35]. Moreover, the sensory data might also be used for
estimating the channel information to reduce the pilot overhead
so that the ISAC systems can harvest sensing-assisted perfor-
mance gains for communications [38], [87], [89]. As a step
further, we attempt to ask and answer the question Challenge
2) How much channel information can be inferred from
the sensory data? Finally, to quantify the mutual performance
gains of ISAC systems [93], we conclude the theoretical
foundation of ISAC systems part by discussing the question
Challenge 3) How could we quantify the integration and
coordination gains?
Secondly, towards system design, typical methods originally
conceived for cellular networks might be adapted to the
needs of ISAC systems, as exemplified by service requests
and resource scheduling in the physical (PHY) layer. Nev-
ertheless, when the sensing functionality is integrated into
wireless systems as an extra basic service [67], [88], the
need for providing tangible sensing performance guarantees
brings about several emerging challenges. For example, the
synchronization requirement of sensing is significantly tighter
than for communications, because the resultant timing offset
may inflict interference upon the sensing functionality. To
avoid phase offset and hence provide high-quality sensing
services (e.g., super-resolution sensing), ISAC systems require
precise clock synchronization between the transmitter and
receiver [95]–[97]. Furthermore, with the potential integration
of S&C, both sensing-centric as well as communication-
centric and joint designs have been proposed for ISAC signal
processing [30], which can be harnessed for striking a flexible
performance tradeoff between S&C. Even though numerous
tradeoff designs have been proposed for maximizing the
communication (or sensing) performance subject to sensing
(or communication) requirements [14], [41], [51], [54], [63],
it remains unclear, what the Pareto-optimal signaling strategy
is. Moreover, for considering radar sensing as an inherent
service in future networks, it is quite important to improve
the sensing resolution to meet the requirements of emerging
applications such as V2X and IoT [29], [30]. As a result, the
corresponding ISAC system design faces the following open
questions: Challenge 4) How could we deal with the clock
synchronization and phase offset both in the bistatic and
distributed deployments? Challenge 5) How far are the
existing designs from the Pareto-optimal boundary? Chal-
lenge 6) How could we improve the sensing performance
by exploiting the emerging super-resolution methods?
Thirdly, we note that the existing cellular architecture and
the classic resource management schemes have to be fine-
tuned in support of ISAC networks [27], [29]. More specif-
ically, the classical communication-only cellular architecture
treats inter-cell interference as a harmful factor that has to be
reduced. However, the seamless sensing service of the future
might require a rich set of echo signals to fully reconstruct the
surrounding environments. In this context, the inter-cell signals
are considered as sensing-friendly signal contributions rather
than as hostile interference. This motivates the conception of
novel schemes reminiscent of the popular cooperative multi-
cell processing (CoMP) concept for ISAC networks. Further-
more, resource management constitutes a major challenge for
efficiently responding to sudden sensing requests. Accordingly,
one has to address the following pair of open challenges in
ISAC networks: Challenge 7) What is the potential cellular
architecture of the future network of sensing? Challenge
8) How could we conceive resource management schemes
and protocols specifically tailored for the ISAC network?
Finally, it is worth pointing out that ISAC systems and
networks may not be implemented until the security and
privacy issues of sensing have been addressed [18], [129],
[130]. For example, in human action sensing scenarios [131],
the radar/WiFi-based sensing [121] technique tends to incur a
potential illegitimate interception of a target, due to the open
nature of the wireless sensing medium [13], [132]. Moreover,
the CSI might contain certain information of private nature
concerning both the transmitters and targets [133], which has
to be settled to avoid being eavesdropped upon simultaneously.
On the other hand, as a typical use case in ISAC applications,
multi-object multi-task (MOMT) recognition aims to identify
4
TABLE I: Recent advances and ten open questions in ISAC systems.
Open Challenges Recent Advances and Efforts Future Directions & Potential Solutions Scenarios
What are the information-theoretic limits of ISAC systems?
•MMSE and MI for radar sensing [74], [75]
•Wireless localization [76]
•I-MMSE equation [77]
•Distortion-capacity [78], [79]
•CRB-rate region [66], [80], [81]
•MI-based ISAC framework [82]–[85]
•Joint communication and binary state
detection [86]
•The first principle in ISAC information
theory
•Proper strategy and coding approach
•KLD measure vs. mutual information
tradeoff
•Ergodic CRB and PCRB
⊙⋆
♣
How much channel information can be inferred from the sensory data?
•OTFS vs. OFDM in ISAC systems [87], [88]
•EKF and deep learning methods [14], [41]
•Two-stage channel estimation protocol [89]
•Real-world S&C dataset [90]–[92]
•The relationship between S&C
channels
•CSI inferred from the sensory data
•Mutual information method and factor
graph method
⊙⋆
♣▲
How could we quantify the integration and coordination gains?
•Communication-assisted sensing [27]
•Sensing-assisted communications [14], [41], [87]
•Performance gain and S&C subspaces [93]
•Proper measures for integration and
coordination gains
•Geometric measure from the perfor-
mance region
⊙⋆
♣
How could we deal with the clock synchronization and phase
offset both in the bistatic and distributed deployments?
•GPS-based time &phase synchronization [94]
•Reference clock, compensation and etc [95]
•Ultra-wideband technique [96], [97]
•Further reduction of clock synchro-
nization errors
•Security of time information in signals
♣▲
■
How far are the existing designs from the Pareto-optimal boundary?
•Pulse interval modulation method [98]
•OFDM used for target detection [23]
•Waveform Design [51]–[54]
•The essential tradeoff in ISAC
•Pareto boundary-based the unified
waveform
⊙▲
How could we improve the sensing performance by the emerging
super-resolution methods?
•CA technique for angle estimaton [99], [100]
•Sparse array methods and algorithms [101]–[103]
•Sparse signal reconstruction [104]–[106]
•Non uniform sparse arrays [107], [108]
•Carrier aggregation assisted sensing
•Bistatic radar sensing
•Other super-resolution algorithms
⊙▲
■
What is the potential cellular architecture of future network of sensing?
•Soft fractional frequency reuse and etc [109]
•User-centric C-RAN [110]
•Cell-free for wireless network [111], [112]
•Cell-free architecture for ISAC network
•Multi-station multi-cell cooperation
scenario
♣▲
How could we conceive resource management schemes and protocols
specifically tailored for the ISAC network?
•PHY layer resource allocation in ISAC [67]
•Cellular cross-layer optimization [113]–[116]
•Channel access algorithm in MAC
layer
•Queuing theory in Network layer
•Adaptive source data encoding in
Application layer
♣■
How could we provide the security and privacy guarantees within ISAC
applications?
•Secure ISAC [117], [118]
•Radar privacy protection in shared spectrum
scenarios [119], [120]
•WiFi sensing security [121]
•Optimal pilots for anti-eavesdropping [122]
•Actual face protection by Doppler radar [123]
•Information theory perspective
•Leakage reduce in PHY layer
•Access control in MAC layer
⊙⋆
♣▲
How could we achieve MOMT recognition by relying on wireless signals?
•Human identify and recognition [124], [125]
•Radar- and WiFi-based sensing
[95], [126]–[128]
•Improvement on the spatial resolution
of wireless sensing
•Multi-object-multi-task sensing
⊙▲
♣■
1In order to clearly indicate the above questions may arise in which kinds of ISAC scenarios, we artificially categorize them into two generic types, e.g., ⊙:monostatic deployment and ♣:
bistatic /distributed deployments.
2In order to qualitatively express the above questions may fall into which focus of future works, we artificially categorize them into three types, e.g., ⋆:fundamental and theoretical investigation,
▲: signal processing algorithm and ■:practical system design.
multiple targets and recognize their behaviors simultaneously.
Substantial efforts have been dedicated to the MOMT recog-
nition in wireless human sensing relying on WiFi signals
[134], [135], albeit with more emphasis on single-person sens-
ing. Clearly, the more challenging high-accuracy multi-person
sensing problem relying on wireless signals still remains open,
especially when the reflected echoes are buried in clutter and
interference. Based on the above brief discussions, we pose
the final pair of cardinal questions: Challenge 9) How could
we provide security and privacy guarantees within ISAC
applications? Challenge 10) How could we achieve MOMT
recognition by relying on wireless signals?
C. Existing Efforts and the Scope of This Paper
In this paper, we critically appraise the recent advances and
formulate ten open challenges in ISAC systems, some of which
have already had some initial progress, while others are still
in the exploratory phase. Following the narrative of “Theory-
System-Network-Application”, we summarize the structure of
this paper as well as the aforementioned open questions and
the relevant ISAC scenarios in Fig. 2. As a benefit of concerted
community effort, the ISAC philosophy has evolved from
a compelling theoretical concept to a practical engineering
challenge [23], [52], [66], [67], [76]–[79], [92], [95], [121],
[124], [131], [136]–[139]. To further pave the way for its suc-
cessful evaluation, we critically appraise the recent advances
and summarize the above ten questions in TABLE I.
Indeed, there have been several pioneering
tutorial/overview/survey papers on ISAC-related topics
throughout the recent decade, e.g., [23] on the intelligent
design of dual-functional waveforms, [24] on radar signals
embedded into communication signals, [25] on WiFi-
based residential healthcare sensing, [26] on radar and
communication coexistence, [27] on ISAC for IoT scenarios,
[28] on signal processing techniques conceived for joint
communication and radar sensing, [29] on the fundamental
limits of ISAC as well as [30] on the ISAC concepts proposed
for 6G and beyond. In contrast to previous works on the
specifics of ISAC fundamental limits, IoT applications or
dual-functional wireless networks, this paper adopts a broader
perspective on recent advances as well as open questions
in ISAC and serves as a complement of the existing efforts
[22]–[33]. The authors of [29] have discussed the fundamental
limits of ISAC commencing from both device-free and device-
based sensing scenarios, followed by the joint-design-based
ISAC and practical ISAC systems. Moreover, other existing
works such as [31] and [32] have discussed the waveform
design by categorizing it into sensing-centric, communication-
centric, and joint design methods. Moreover, the authors
5
TABLE II: Existing overview papers on ISAC
Existing Works [23] [24] [25] [26] [27] [28] [29] [30] This paper
Year 2011 2016 2018 2019 2021 2021 2022 2022 2023
Type Tutorial Tutorial Overview Survey Survey Overview Survey Survey Survey
Information-theoretic limits ✓ ✓ ✓ "
Channel estimation from sensory data ✓"
Integration &coordination gains ✓"
Clock synchronization &phase offset ✓"
Pareto-optimal boundary ✓ ✓ ✓ ✓ ✓ "
Super-resolution algorithm ✓"
New cellular architecture ✓"
Resource management schemes &protocols ✓ ✓ "
Security and privacy in wireless sensing "
MOMT by wireless signals ✓ ✓ ✓ ✓ ✓ "
of [29-31] generally considered three categories, namely
sensing-only, communication-only, and ISAC scenarios. In
contrast to this, we aim to provide another holistic perspective
spanning from fundamental theory to various applications of
ISAC by categorizing the proposed open challenges into four
parts, as illustrated in Fig. 2. For clarity, we provide a detailed
comparison between the existing contributions and this paper
in terms of the aforementioned ten open questions in TABLE
II, followed by a detailed discussion in the remainder of the
paper.
D. Organization of This Paper
The rest of this paper is organized into five sections as
seen in Fig. 2. In Section II, we introduce the basic perfor-
mance metrics for both S&C, discuss the associated channel
estimation issues and provide a metric for quantifying the
integration gain and coordination gain. Section III discusses
clock synchronization, Pareto-optimal signaling strategies, and
super-resolution sensing in the context of physical-layer sys-
tem design. In Section IV, we briefly discuss the potential cell
architectures and cross-layer protocols of ISAC networks. In
Section V, we shed light on sensing security and privacy issues
as well as MOMT, while relying on wireless signals. Finally,
Section VI concludes the paper. For crisp and convenient
clarity, we organize all sections into three subsections, namely
Background, Existing Literature as well as Future Directions
and Potential Solutions. The related abbreviations of this paper
are given in TABLE III.
II. THEORETICAL FO UNDATIO NS OF ISAC SY STE MS
In this section, we explore Challenge 1-3 of Fig. 2. we
first shed light on a range of measures proposed for ISAC
systems from an information-theoretic perspective. Then, we
highlight the associated channel overlap between S&C and
discuss how much channel information can be inferred from
the sensory data. Finally, we propose a “simple but intuitive”
metric to grasp for assessing the integration and coordination
gains based on the three categories of S&C channels, namely,
uncorrelated, moderately correlated, and strongly correlated
scenarios. This metric is informally defined as the ratio of the
area in the mutual-boost region for moderately correlated S&C
scenarios to the area in the non-boost region for uncorrelated
S&C scenarios.
Challenge 1: What Are the Information-Theoretic Limits of
ISAC Systems?
1) Background: Information theory is critical and funda-
mental for evaluating the performance limits of ISAC systems
[29], [30]. Although the integration of S&C is indeed promis-
ing as a benefit of utilizing a common hardware platform
and a common transmitted signal, their basic information
theory has both connections and distinctions. For example,
both S&C systems focus on the mutual information (MI)
maximization. Specifically, one might always try to increase
the achievable rate or ergodic rate by maximizing the MI
between the transmitted and received signals in wireless com-
munication systems. The sensing tasks, on the other hand,
mainly rely on estimation and detection. As for parameter
estimation, maximizing the MI between the target impulse
response and received echo signals usually leads to minimizing
the minimum mean-square error (MMSE) [74], [77]. For target
detection, maximizing the Kullback-Leibler divergence (KLD)
between the pair of scenarios when the target is present
and absent asymptotically leads to a problem reminiscent of
detection probability maximization [140]–[142].
In what follows, we elaborate on the family of MI-oriented
ISAC systems, and then we introduce other metrics for quan-
tifying the sensing performance, such as the KLD and the
Cramér-Rao bound (CRB). Furthermore, we briefly discuss
several intrinsic connections among these ISAC metrics.
•MI-Oriented ISAC Systems: The classical information
measure in modern wireless systems is the MI [143]. Let us
consider a generic linear Gaussian model for both S&C, which
is formulated as
Yc=HcX+Zc,(1a)
Ys=HsX+Zs,(1b)
where Ycand Ys,Hcand Hs,Zcand Zsrepresent the
received signal, channel/target impulse response (CIR/TIR),
and the additive white Gaussian noise (AWGN) at the commu-
6
TABLE III: List of Abbreviations
Abbreviation Definition
ISAC Integrated sensing and communications
B5G Beyond 5G
S&C Sensing and communications
MOMT Multi-object multi-task
V2X Vehicle-to-everything
UAV Unmanned aerial vehicle
IoT Internet of things
MI Mutual information
KLD Kullback-Leibler Divergence
CRB Cramér-Rao bound
PCRB/BCRB posterior/Bayesian CRB
CIR/TIR Channel/target impulse response
AWGN Additive white Gaussian noise
PDF Probability density functions
SPEB Squared position error bound
MMSE Minimum mean squared error
SNR Signal-to-noise ratio
FIM Fisher information matrix
BS Base station
CSI Channel state information
V2I Vehicle-to-infrastructure
MIMO Multiple-input-multiple output
EKF Extended Kalman Filter
OTFS Orthogonal time frequency space
OFDM Orthogonal frequency division multiplexing
TMO Timing offset
CFO Carrier frequency offset
TOA Time-of-arrival
TDOA Time-difference-of-arrival
GPS Global positioning system
UWB Ultra-wideband
PIM Pulse interval modulation
ASK/PSK Amplitude/phase shift keying
LFM Linear frequency modulation
SINR Signal-to-interference plus noise ratio
MSE Mean square error
DOA Direction of arrival
OF Objective functions
BER Bit error rate
CA Carrier aggregation
ULA Uniform linear array
RAN Radio access networks
C-RAN Cloud RAN
RRHs Remote radio heads
APs Access points
OSI Open system interconnections
MAC Media access control
RTS Request to send
CTS Clear to send
QoS Quality of service
DNN Deep neural network
FMCW Frequency modulation wave
PRI Pulse repetition interval
FFT Fast Fourier transform
3D 3-dimensional
0 5 10 15 20 25 30
0
1
2
3
4
5
6
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Fig. 3. KLD and detection probability versus SNR.
nication and sensing receivers, respectively. Finally, Xdenotes
the transmit signal. On one hand, with the goal of improving
the performance of communication systems, one has to design
the optimal Xin order to maximize the conditional MI
I(X;Yc|Hc)[142], which quantifies the maximum achiev-
able point-to-point rate, in the face of practical constraints
such as the total power budget. On the other hand, radar
systems aim for the maximum conditional MI I(Hs;Ys|X)to
quantify the sensing performance [74], [75], [84]. Explicitly,
we have the following pair of optimization problems toward
communication-only or sensing-only scenarios:
max I(X;Yc|Hc) or I(Hs;Ys|X)(2a)
s.t.Other Specific Practical Constraints.(2b)
By appropriately optimizing the signaling strategies, such
as symbol- and block-level transmit precoding [51], [144],
we notice that the aforementioned optimization problem (2)
leads to an MI-based tradeoff between S&C, leading to
the following weighted MI-oriented maximization problem in
ISAC scenarios:
max ρI(X;Yc|Hc) + (1 −ρ)I(Hs;Ys|X)(3a)
s.t.Other Specific Practical Constraints,(3b)
where the weighting factor ρ∈[0,1] controls the priorities
assigned to S&C functionalities. This means that one can find
the Pareto-optimal MI boundary in ISAC systems to illustrate
the MI tradeoff between S&C [82]–[85].
Indeed, the MI measure is suitable for radar sensing, but
it might not always be precise. This is because the MI does
not have an explicit operational interpretation in radar sensing
systems, which is in stark contrast to wireless communication
systems. More specifically, we can readily quantify how much
useful information can be inferred by using bits in communi-
cation systems, but it remains unclear how many bits one can
obtain by observing the trajectory of the target, including its
velocity and angle estimation. Moreover, it is worth noting that
another drawback of using the MI for sensing is the difficulty
in characterizing it by a closed-form formulation in general.
7
•Target Detection Metric: KLD. The KLD metric, a.k.a.
the relative entropy, can be used for quantifying the differ-
ence between two distributions having the probability density
functions (PDF) of p0and p1, which is defined as
D(p0∥p1) = Zp0(x) log p0(x)
p1(x)dx
=Ep0ïlog p0(X)
p1(X)ò,(4)
where Ep0(·)represents the expectation under distribution
p0. In a statistical sense, the KLD quantifies the distance
between a pair of probability distributions [140]. When the
two distributions are identical, the corresponding KLD is
zero. In particular, for the detection problem (ignoring clutter)
in monostatic scenarios, we seek to choose between two
hypotheses, i.e. H1representing that the target is present, or
H0, the target is absent, which can be formulated as
Ys=ßH1:HsX+Zs,
H0:Zs.(5)
Let us denote the PDF of Ysunder H1and H0as p1(Ys)and
p0(Ys), respectively. Then, upon following the Chernoff-Stein
lemma [142], the KLD can be used for quantifying the target
detection probability, expressed as [140]–[142]
D[p0(Ys)∥p1(Ys)] = lim
N→∞ Å−1
Nlog (1 −PD)ã,(6)
where Nis the number of observations and PDis the target
detection probability. More explicitly, Equation (6) suggests
that upon increasing N, maximizing the KLD is asymptoti-
cally equivalent to maximizing the detection probability.
To elaborate a little further for lifting the KLD into target
detection problem, let us consider the basic example of (5) in
the single-antenna scenario, having the following representa-
tion
y[ℓ] = ßH1:hx[ℓ] + z[ℓ], ℓ = 0,1, . . . , L −1,
H0:z[ℓ], ℓ = 0,1, . . . , L −1,(7)
where h∼ CN(0, σ2
h)is the reflecting coefficient, x[ℓ]is
the transmit signal and z[ℓ]∼ CN(0, σ2
z)is the AWGN,
respectively. By stacking all the received signals y[ℓ]into
y= [y[0], . . . , y[L−1]]T, the PDFs of yunder H0and H1
can be expressed as
p0(y) = 1
πLσ2L
z
exp ï−1
σ2
z
yHyò,(8a)
p1(y) = 1
πLdet(Cs+σ2
zI)exp −yH(Cs+σ2
zI)−1y,
(8b)
where x= [x[0], . . . , x[L−1]]Trepresents the transmitted
signals and Cs=σ2
hxxHdenotes the covariance matrix of
hx, while Iis the L-dimension identity matrix, respectively.
Then the log-likelihood ratio can be derived as
log p1(y)
p0(y)=−yHï(Cs+σ2
zI)−1−1
σ2
z
Iòy
−log det(Cs+σ2
zI) + log det σ2L
z,(9)
Following the Neyman-Person lemma, the detection probabil-
0 5 10 15 20 25 30
-30
-25
-20
-15
-10
-5
0
Fig. 4. CRB and BCRB versus SNR.
ity is [145]
PD=P
1
1+ P σ2
h
σ2
z
F A ,(10)
where P=xHxdenotes the total transmit power and PF A
represents the given false alarm probability. We remark that
it is difficult to design the transmit signal xby (10) in ISAC
scenarios where xmust be tailored to achieve specific com-
munication performance optimization goals, such as reducing
the symbol error rate (SER) [51]. This is primarily due to the
term xHx, which is independent of the power budget P. As
an alternative manner to tackle this issue, one may choose
to maximize the KLD which is a function of the transmitted
signals, as detailed below.
Let us denote the KLD between H0and H1by D(p0∥p1),
which can be calculated by
D(p0∥p1) = Zp0(y) log p0(y)
p1(y)dy
=−Ep0(y)ïlog p1(y)
p0(y)ò
=Ep0(y)ïyHÅ(Cs+σ2
zI)−1−1
σ2
z
Iãyò+ϕ
= tr σ2
z(Cs+σ2
zI)−1−I+ϕ, (11)
where we have ϕ= log det(Cs+σ2
zI)−log det σ2L
z. Then,
the Chernoff-Stein lemma indicates that the missed detection
probability of 1−PDbecomes exponentially small, with an
exponential rate D(p0∥p1)given above. Finally, we identify
the relationship between the KLD and the detection probability
versus transmit SNR = P/σ2
zin Fig. 3. It can be observed
that upon increasing the signal-to-noise ratio (SNR), the KLD
is also increased, which results in improved detection perfor-
mance.
It is worth pointing out that the specific choice of sensing
metrics may be determined by the application scenarios and
design principles. We remark that the detection probability
PDin the binary test (7), as expressed in (10), typically lacks
a straightforward link with the transmit signals, making the
8
design of the transmit signal xa complex task. Another viable
approach is to use the KLD expression, which offers more
flexibility in designing the transmitter’s covariance matrix,
such as Csin (11). This increased flexibility is particularly
valuable in ISAC scenarios, where the transmit waveform
design must satisfy stringent communication requirements.
•Parameter Estimation Metric: CRB. We assume that the
vector θdenotes the parameters to be estimated, and its
estimate is ˆ
θ. Typically, we are concerned with unbiased
estimation scenarios, which indicates E(ˆ
θ) = θ. Then the
estimation performance can be quantified by the well-known
CRB, which serves as the lower bound on the mean square
error (MSE) of any ˆ
θover θ. Let J(θ)denote the Fisher
information matrix (FIM). Then the CRB matrix can be written
as [145]
MCRB(θ) = J−1(θ),(12)
where [J(θ)]ij =−Eh∂2log p(y|θ)
∂θi∂ θjirepresents the i, j-th
element of the FIM J(θ)and log p(y|θ)is the log-likelihood
function of the observed data y, given θ. The above expression
represents the perspective of the Frequentists who assert
that the θto be estimated is an unknown but deterministic
parameter [146]. Accordingly, the CRB of the deterministic
parameter θis
CRB(θ) = tr J−1(θ).(13)
On the other hand, the Bayesians state that θrepresents
unknown random variables having a-priori distribution of
π(θ). Therefore, the joint PDF of θand yis formulated as
p(θ,y) = p(y|θ)π(θ).(14)
According to (14), the a-posteriori FIM of θcan be written
as [76]
Jp(θ) = −EÅ∂2log p(θ,y)
∂θ2ã
=−EÅ∂2log p(y|θ)
∂θ2ã
| {z }
Observed Fisher Information
−EÅ∂2log π(θ)
∂θ2ã
| {z }
Prior Fisher Information
≜J+Jprior,(15)
which leads to the Bayesian/a-posteriori CRB (BCRB/PCRB)
of
BCRB(θ) = tr J−1
p(θ).(16)
To provide further intuition about the estimation perfor-
mance, let us recall the basic linear model of the single-
antenna scenario in (5), having the following representation
y[ℓ] = hx[ℓ] + z[ℓ], ℓ = 0,1, . . . , L −1,(17)
where his the reflecting coefficient having the a-priori distri-
bution CN(0, σ2
h)to be estimated. For ease of illustration, we
assume that the noise is AWGN, where z[ℓ]∼ CN(0, σ2
z)
and σ2
zdenotes the variance. We would like to emphasize
that incorporating colored noise might ameliorate this basic
example, but in a first approximation, one can simply replace
the white noise variance matrix σ2
zIwith the colored noise
variance matrix. Denote the real part and imaginary part of h
MI
MMSE
KLD Detection
Probability
Chernoff-
Stein Lemma
I-MMSE
Equality
Second-order
Taylor Series
Fisher
Information
Differential
Entropy
CRB
CRB
Inequality
de Bruijn
Identity
Definition
Expansion
Fig. 5. The connections among ISAC metrics.
by hRand hI. Then the joint PDF of θ= [hR, hI]Tand y
can be written as
p(θ,y) = 1
πLσ2L
z
exp ï−1
σ2
z
(y−hx)H(y−hx)ò
| {z }
p(y|θ)
·1
πσ2
h
exp ï−1
σ2
h|h|2ò
| {z }
π(θ)
.(18)
Therefore, the a-posteriori FIM of θcan be formulated as
Jp(θ)=2"xHx
σ2
z+1
σ2
h
0
0xHx
σ2
z+1
σ2
h#.(19)
The corresponding BCRB of θis expressed as
BCRB(θ) = σ2
hσ2
z
σ2
hxHx+σ2
z
.(20)
Additionally, if the a-priori distribution π(θ)is not available,
the BCRB degenerates into the classical CRB in (13), given
as [145]
CRB(θ) = σ2
z
xHx.(21)
Notice that the BCRB usually provides additional Fisher
information, which is also independent of the real value of
the parameters to be estimated, resulting in better estimation
performance than the classical CRB. We illustrate this point
in Fig. 4. It is observed that with the a-priori distribution at
hand, the BCRB metric is attainable with better estimation
performance than the classical CRB. We remark that the CRB
and BCRB/PCRB represent different perspectives in statistics,
which constitute fundamental philosophical and cognitive is-
sues [146].
As illustrated in Fig. 5, there are several connections among
the aforementioned metrics in general, which are detailed
below. Furthermore, these metrics may also gradually find
their way into the Pareto-optimization of ISAC systems. To
elaborate briefly, it cannot be expected that an ISAC system
having the lowest MMSE is also the best in terms of any
of the other metrics. Hence if we want to improve any of
the other metrics, we must accept an MMSE degradation for
9
example. This Pareto-optimization effort may commence from
a pair of metrics and then gradually extend to three or more
of them. However, as the number of metrics is increased,
the ‘search-space’ may gradually become excessive hence
requiring substantial future research.
•The I-MMSE equation 1relates the MMSE of communi-
cation signal Xto the mutual information, as [77]
d
dSNRI(X;Y) = 1
2MMSEX(SNR).(22)
•The MI can be expressed also by KLD in the form of
I(X;Yc) = D[p(X,Yc)∥p(X)p(Yc)] [142].
•The Fisher information characterizes the curvature of the
KLD between two distributions [147]. Let us assume that the
log-likelihood function of θ∈Θis given by log p0(y|θ)and
log p1(y|θ1)for a given θ1∈Θ. The second-order Taylor
series expansion with respect to θat θ=θ1may be written
as
D(p0∥p1) = 1
2(θ−θ1)TJ(θ1) (θ−θ1)
+o∥θ−θ1∥2.(23)
•The Fisher information can be linked to the differential
entropy via the de Bruijn identity given by [148]
d
dth(HcX+√tN0) = 1
2tr{J(HcX+√tN0)},(24)
where the parameter tis assumed to be non-negative, N0is a
noise vector having independent standard Gaussian entries and
it is assumed to be independent of X. Here, h(·)represents
the differential entropy and J(·)is the Fisher information
matrix. By following the linear Gaussian model of (1a) and
assuming Nc=tN0, the de Bruijn identity states that the
Fisher information of Yccan be viewed as the curvature of
differential entropy [77], which can be expressed as
d
dth(Yc) = 1
2tr{J(Yc)}.(25)
2) Existing Literature: Given the evolving integration of
S&C, the information theories of S&C may also be expected
to be unified. As for the fundamental limits of sensing, a recent
use case was reported in a wireless localization scenario [76].
By introducing the squared position error bound (SPEB) to
characterize the localization accuracy [76], one can analyze
the fundamental limits of device-based localization. On the
other hand, when considering the communication part, the
early results on the information-theoretic connection between
S&C were disseminated in the well-known I-MMSE equation
[77].
Another information-theoretic perspective is represented
by the capacity-distortion tradeoff model, where the target
response (e.g., radar echo signals) is modeled as a delayed
feedback channel [78]. In this setup, the state-dependent
channel outputs the signal to the receiver and returns the state
feedback to the transmitter for estimating the channel it state.
1Briefly, this equation stated that the MMSE of communication signal X
is equal to the derivative of the MI with respect to the SNR [77]. However,
in monostatic sensing systems, the ISAC base station already has the perfect
knowledge of its own transmit signal X. Then one only has to estimate the
target parameters embedded in Hsinstead of Xin the I-MMSE equation.
Rate
CRB
Communication-optimal point
Sensing-optimal point
max Rate
min CRB
Fig. 6. The pentagon inner bound of the CRB-rate region.
Following the spirit of [78], the authors of [79] extended
to correlated sensing and channel states by considering a
generalized channel model of multiple-access ISAC systems.
More recently, the authors of [66] proposed a pentagon inner
bound of the CRB-rate region to reveal the fundamental
tradeoff between S&C, as illustrated in Fig. 6. This model
first revealed the information-theoretic connections among the
classic communication capacity, the target channel states and
the parameters to be estimated in ISAC systems.
3) Future Directions and Potential Solutions: Clearly, the
information theoretic research of ISAC systems is still in its
infancy. There are numerous open problems to be addressed
in future work, some of which are listed as follows.
•The weighted optimization problem (3) leads to the
Pareto-optimal MI boundary in ISAC systems [82]–[85],
but different design goals and metrics should also con-
sidered in the context of the Pareto boundary, such as
communication rate and the target detection probability.
•Again, substantial community effort is required for quan-
tifying the S&C performance in terms of their MI. In
simple terms, how many bits must be transmitted in
different sensing scenarios, for example, for delay, angle
and Doppler estimation.
•Identifying the most promising techniques (e.g., cod-
ing approach) capable of approaching the Pareto front
remains widely unexplored. Although the authors of
[66] invested substantial efforts into analyzing the two
corner points, i.e., communication- and sensing-optimal
performance in Fig. 6, how to attain the actual boundary
of S&C is still an open issue.
•There are still fundamental issues to be addressed both
in the target detection and in the wireless communica-
tion components of ISAC systems from an information-
theoretic perspective. Pioneering efforts have been made
in joint communication and binary state detection in [86].
As a further step, it is necessary to clarify the connection
between the KLD measure (in both single and multiple
target(s) detection) as well as the mutual information (in
wireless communication), which may help us to reveal
the fundamental tradeoff between the target detection
probability and communication rate.
•To reveal the fundamental limits of ISAC systems, a
10
potential solution is to establish the unified metrics for
both S&C. For example, the authors of [149] defined the
sensing estimation rate (SER) to unify the information-
and estimation-theoretic perspectives of ISAC systems.
However, it is still challenging to carry out a unified
performance evaluation in other ISAC scenarios.
•In ISAC systems, a promising technique is to perform
sensing by utilizing communication signals. However, the
signals dedicated to information transfer are inherently
random. Therefore, the specific sensing metrics, such as
the ergodic CRB or ergodic PCRB have to be tailored
and well-defined for different use cases.
•The fundamental information theory for unveiling the
attainable degrees of freedom, and of massive MIMO
ISAC systems as well as of their networking issues is
still open.
Challenge 2: How Much Channel Information Can be
Inferred From the Sensory Data?
1) Background: In conventional wireless communication
systems, the base station (BS) firstly transmits pilots to all
the users via the downlink channel. The users then estimate
the channel state information (CSI) and transmit their channel
estimates back to the BS in the uplink channel. Naturally,
the pilot symbols result in communication overheads and thus
they limit the useful transmission rate [14]. In the context
of ISAC, the sensory data that contains information about
the surrounding environment can also be exploited to obtain
channel information. It is envisioned that ISAC may offer a
solution that either partially or fully eliminates the need for this
feedback loop and thus boosts the communication performance
[38].
Communication User
Communication User
Sensing Target
(Scatterer)
ISAC User
Communication Signal
ISAC Signal
Sensing Signal
Sensing Target
Uncorrelated
Moderately Correlated
Strongly Correlated
Fig. 7. Three categories of S&C environment with typical
scenarios.
The prerequisite of inferring CSI is that there is a suffi-
ciently high correlation between the sensing and communica-
tion channels. Therefore, we firstly classify the correlation of
S&C channels into three categories, as shown in Fig. 7.
•Uncorrelated: S&C take place in different spatial envi-
ronments and the channels of sensing and communication
are completely uncorrelated or independent of each other.
For example, the BS is delivering information to a UAV
and simultaneously sensing a passing-by vehicle. There-
fore, it is difficult to infer any useful channel information
from the sensory data in an uncorrelated scenario.
•Moderately correlated: In this category, the channels
of sensing and communication are partially correlated,
when, for instance, the BS is communicating with a
pedestrian and sensing a passing-by vehicle at the same
time. The latter happens to be a scatterer of the communi-
cation signal, in which case the sensing path contributes
partly to the communication channel. Accordingly, the
S&C channels are moderately coupled. Therefore, the
ISAC BS may infer partial CSI from the strong echoes
[89].
•Strongly correlated: The strongly coupled scenario refers
to the case when the sensing target is exactly the com-
munication target and the channels of S&C are strongly
correlated. A typical example can be found in vehicle-to-
infrastructure (V2I) communication systems [14], where
the roadside unit (RSU) can exploit the echo signals to
predict and track the vehicular target, so as to assist
the downlink communication and to reduce the pilot
overhead.
2) Existing Literature: In massive multiple-input-multiple-
output (mMIMO) systems, the transmitted beampattern is
chosen by tentatively harnessing all the beams from a pre-
defined codebook, which leads to high beam training overhead
in practical scenarios [150], [151]. The sensory data containing
specific features such as velocity, location and angle could
provide useful information both for beam tracking and beam
prediction, hence reducing the pilot and feedback overheads
to a certain extent. Some Bayesian filtering methods, such
as extended Kalman Filters (EKF) [14] can be employed for
improving the tracking accuracy and reducing the pilot over-
head. Furthermore, deep learning algorithms can be leveraged
to map the features from the sensory data to the optimal beam,
thus further reducing the communication overheads [90]–[92].
3) Future Directions and Potential Solutions: The corre-
lation of S&C channels can also be considered as a parame-
terization model, where the three categories of Fig. 7can be
modeled by the correlation between channel parameters. To
better motivate the technical discussions in this part, we first
recall the general linear Gaussian model of (1) and add the
parameters ηcand ηc, yielding
Yc=Hc(ηc)X+Zc,(26a)
Ys=Hs(ηs)X+Zs,(26b)
where Hc(ηc)and Hs(ηs)denote the communication and
sensing channels that fluctuate with the variation of the chan-
nel parameters ηcand ηs, respectively.
The relationship between the parameters of interest in sens-
ing and communication can be illustrated in the Venn diagram
of Fig. 8, where ηcs represents the correlation parameters
in both the S&C channels. Following the three categories
of ISAC channels in Fig. 7,ηcs may be a null set in the
uncorrelated scenario. Otherwise, ηcs will reflect the corre-
lation of both S&C channels. The factor graph of Fig. 8
encapsulates the relationship between the parameters, intimat-
ing how we can infer channel information from sensing data.
11
Factor node Variable node
c
H
Communication-only
Parameters
Sensing-only
Parameters
cs
η
Overlap Parameters
c
η
s
η
s
H
c
η
s
η
cs
η
Fig. 8. Channel parameters between S&C with the correspond-
ing factor graph.
More particularly, the mutual information I(Hc;Hs|Ys,X)
between Hcand Hscan be applied to estimate how much
information can be inferred from the sensory data Ys. Gen-
erally, I(Hc;Hs|Ys,X)will be higher in strongly cou-
pled scenarios than that in the uncorrelated scenario. This
information-theoretic perspective provides some theoretical
insights concerning how to glean some channel information
from the sensory data.
Finally, to harness the sensory data to infer CSI, one has to
concentrate on the modeling and algorithmic design methods
in the future. Specifically, one has to precisely model the
relationships between the communication channels Hcand the
sensing channels Hsthrough factor graph relationships, espe-
cially in complex wireless environments, while considering
spectrum interference, the presence of clutter sources and so
on. Meanwhile, meeting the real-time requirements of sensing
presents a significant challenge in designing efficient practical
algorithms suitable for diverse scenarios, such as dynamic
V2I networks. To this end, it would be promising to establish
corresponding data sets of complex environments and explore
machine learning methods for utilizing huge sensory data to
infer the CSI.
Challenge 3: How Could We Quantify the Integration and
Coordination Gains?
1) Background: Sharing the same spectrum is only the
first step in moving from complete separation toward the
integration of sensing and wireless communication systems.
As a benefit of the commonalities between the S&C systems
in terms of their hardware architecture and signal processing
algorithms, ISAC systems aim for a higher degree of inte-
gration, which we refer to as the integration gain [27]. On
the other hand, with the aid of mutual assistance, we can
attain beneficial coordination gain, hence boosting the ISAC
performance [27]. The stylized relationship of the integration
gain and coordination gain is portrayed in Fig. 9, which man-
ifest themselves in terms of spectrum sharing [26], waveform
design [57], interference management [58], [59].
To illustrate the main ideas, let us consider a sensing-
assisted V2I communication scenario reported in [14]. More
specifically, the radar echos received at the RSU are leveraged
to predict a vehicle’s trajectory for beam alignment in the
next round of communication. Therefore, downlink commu-
nications can benefit in terms of reduced pilot overhead [14].
In this setup, by utilizing both a common platform as well
Spectrum Application
Hardware
Waveform
Interference
Cooperation
Mutual Boost
Integration Gain
Coordination Gain
Fig. 9. The stylized interplay of integration and coordination
gains in ISAC.
Sensing Performance (e.g., Detection Probability )
Communication Performance (e.g., Rate or SINR)
OA
BC
Boundary
AB
Boundary
Boundary AC-CB
1
S
2
S
AB
Fig. 10. Graphical illustration of various integration and coor-
dination scenarios.
as the same waveform, the ISAC system can perform dual-
functional S&C for improving the hardware-, spectrum-, and
energy-efficiencies. This naturally attains both integration and
coordination gains in ISAC systems at the same time.
2) Future Directions and Potential Solutions:
On one hand, if orthogonal resources are scheduled for S&C
functionalities, in the context of either time- or frequency-
division schemes, this implies that no integration gain is
achieved, since no resources are shared between S&C. On the
other hand, a beneficial integration gain can be acquired when
the sensing and communication channels are highly overlapped
[93]. For example, if the communication user is also the target
to be sensed, the frequency- or time-domain resources are
shared and the channels are correlated, as illustrated in Fig.
7. In such a case, both the S&C functionalities glean benefits.
Therefore, we conclude that the integration gain depends on
the correlation of the S&C channels (i.e., uncorrelated and
strongly correlated as the boundary AC-CB shows in Fig. 10).
These scenarios have to be discussed on a case-by-case basis.
Before exploiting the benefits of integrated S&C, we have
to define metrics for quantifying the performance gains of
ISAC systems [93]. We attempt to highlight the integration
and coordination gains in Fig. 10, where points A, B, and
C stand for the sensing-optimal, communication-optimal, and
12
S&C-optimal performance, respectively. In what follows, we
first elaborate further on the different degrees of the channel
overlap and the corresponding performance bounds. Then we
discuss how to quantify the performance gains in ISAC.
•Uncorrelated: The inner bound, i.e., boundary AB of Fig.
10, refers to the scenario where the S&C fail to boost
each other. This is the case upon allocating orthogonal
resources between S&C, hence resulting in no integration
gain. A basic example can be found in Fig. 7, where the
BS is sensing a passing-by vehicle while simultaneously
communicating with a UAV. This case indicates that
S&C take place in different spatial environments and the
wireless resources are allocated in an orthogonal spatial
region.
•Moderately correlated: In most practical scenarios, the
S&C components may benefit from each other. Then the
achievable performance can be illustrated by the arc-
shaped boundary ˜
AB of Fig. 10. Again, as illustrated
in Fig. 7, the passing-by vehicle to be sensed is also a
scatterer of the communication signal. This case means
that some of wireless resources, e.g., the transmit power,
can be jointly reused by S&C systems.
•Strongly correlated: The upper bound, i.e., the boundary
AC-CB of Fig. 10 represents the scenario, where the wire-
less channels of S&C are fully aligned. In other words,
the system can always operate at both the sensing- and
communication-optimal points without any performance
erosion on either side 2. As illustrated in Fig. 7, when the
sensing target is also playing the role of a communication
user, the wireless resources are jointly exploited by S&C,
leading to high integration and coordination gains.
It is plausible to find that the blue area S2represents
the performance gain, when both S&C are mutually rein-
forced by each other, while the green area S1denotes the
non-cooperative case. Following the graphical illustration of
various possible integration and coordination gains, we can
visualize both the integration and coordination gains in the
ISAC systems by the ratio between the areas of the above
performance regions, which can be informally expressed as
Integration & Coordination Gains ∝Area ÅS2
S1ã.(27)
We remark that this is merely an intuitive representation for
the sake of illustration, rather than a rigorous mathematical
description. Usually, it is challenging to quantify S1and S2,
which has to be analyzed based on the specific conditions. As
a preliminary work, the authors of [93] attempted to quantify
the integration gain of ISAC systems in a single-target single-
antenna communication user scenario. This was attempted by
defining the subspace “correlation coefficient” to attain S2 to
investigate the coupling effect between S&C channels. The
2It is worth pointing out that there is a two-fold tradeoff in ISAC systems,
namely, the subspace tradeoff (ST) and deterministic-random tradeoff (DRT)
[66]. In this paper, we mainly focus on the ST from the perspective of S&C
channels, while we refer readers to [66] for more details on DRT. To avoid
out-of-scope digression, we discuss the S&C performance tradeoffs in the
specific context where the data frame length of ISAC signals is sufficiently
large and the impact of DRT on the S&C system is negligible [66], [152],
[153].
insight is that a larger “correlation coefficient” of S&C may
lead to a larger performance region S2, resulting in higher
performance gain.
III. PHYSICAL-LAYER SYS TEM DESIGN
In this section, we tackle Challenges 4-6 of Fig. 2and
present three key aspects of physical-layer system design, with
an emphasis on clock synchronization, Pareto-optimal-oriented
signaling strategies, and super-resolution methods conceived
for the Sub-6G bands. In particular, we commence by briefly
discussing the synchronization and phase offset issues of
both bistatic and distributed scenarios. Then, inspired by the
joint design of the ISAC signal, how to attain Pareto-optimal
signaling strategies is discussed as well. Finally, we briefly
introduce the family of super-resolution methods designed
for wireless sensing networks and propose potential research
directions for achieving super-resolution sensing.
Challenge 4: How Could We Deal With the Clock Synchro-
nization and Phase Offset in the Bistatic and Distributed
Deployments?
1) Background: Asynchronous operation imposes widely
recognized challenges on collaborative communication net-
works. Hence sophisticated techniques have been proposed
for the conventional communication systems to tackle this
problem, e.g., broadcast-based infrastructure synchronization
[154]. However, these methods may not be applicable to ISAC
systems, since their synchronization requirement related to
the sensing function is generally more tight than that of the
communication component. For example, the synchronization
requirement between BSs relying on different techniques typ-
ically ranges from 0.2µs to 12.8µs [155]. By contrast, for
meter-level positioning accuracy, the synchronization require-
ment is at the nanosecond/sub-nanosecond level, since the
time-of-arrival difference at 0.3×109m/s and 0.3mdistance
is 1µs. Thus, the clock synchronization problem should be
reconsidered in ISAC systems, which will be discussed in this
subsection.
2) Influence of Asynchronous Clock: There are sev-
eral issues related to asynchronous clocks, such as timing
offset (TMO), carrier frequency offset (CFO), and random
phase shift. Specifically, TMO mainly influences the range
estimation accuracy by incurring range bias in time-related
estimation algorithms, such as the family of time-of-arrival
(TOA) and time-difference-of-arrival (TDOA) based schemes.
As shown in Fig. 11, due to the existence of TMO, an extra
error is imposed on estimating the propagation delay, and
accordingly a range bias is introduced. In the case of TDOA,
tight clock synchronism between the receivers is also required.
Given the fact that the signal travels at the speed of light,
even a tiny estimation error in signal delay, e.g., at the level of
nanoseconds, could lead to a range error of meters, which may
prevent accurate navigation. The CFO brings about ambiguity
when estimating the Doppler frequency, thereby exacerbating
the error imposed on the estimation of the target speed.
Apart from TMO and CFO, the presence of random phase
shifts prevents us from coherently aggregating measurements
at different timeslots/packets [95]. Since clock asynchronism
13
Range Ambiguity
Target
Fig. 11. Range bias caused by TMO.
may gravely degrade a system’s sensing performance, one
of the major tasks in designing the ISAC system is that of
ensuring tight clock synchronization between the active nodes
or, alternatively compensating for the synchronization errors.
To elaborate briefly, communication receivers harness chan-
nel estimation which is also capable of simultaneously can-
celing out both the propagation delay and clock offset. By
contrast, an ISAC receiver should perform compensation for
different functions in different ways, as shown in Fig. 12,
where the sensor should extract the target information from
the received signal. Therefore, only the clock offset can be re-
moved. This leads to a difference between the synchronization
of ISAC and communication-only systems.
3) Existing Literature: A whole suite of sophisticated
methods capable of achieving high-accuracy synchronization
have been proposed, for example, by applying a common
reference clock [94]. The common reference clock may be
broadcast by a dominant node of the ISAC system or by the
global positioning system (GPS). Then the active nodes of
the system may simply lock on to this reference clock to
synchronize with each other. As a result, the hardware cost
will be significantly reduced, since only a few active nodes
should have a high-precision clock. Specifically, by extracting
the time information from the GPS signal, a high clock
synchronization accuracy can be obtained at the nanosecond
level [94]. In [96], a clock synchronization scheme based
on the ultra-wideband (UWB) technique has been proposed.
The theoretical results show that the method of [96] achieves
a high clock accuracy (less than 3ns) for 100 nodes. The
synchronization is realized by exchanging packets having
timestamps among the active nodes. With the help of the UWB
technique, high-resolution timestamps can be obtained, and
accordingly, high-precision synchronization between the active
nodes is achieved.
For an ISAC receiver equipped with multiple antennas, the
clock asynchronism can be canceled out with the aid of signal
processing, since the asynchronous clock could be regarded as
adding some phase-shifted terms to the expression of channel
states. Since the phase-shifted terms imposed by the asyn-
chronous clock across different antennas are approximately
the same, mathematical manipulations can be performed to
remove these terms. Hence synchronization between active
s(t-τ-Δt)
Channel
s(t)
s(t-τ)
s(t)
s(t): transmitted signal
τ: transmitted delay
Δt: clock asynchronism
Fig. 12. Different compensation techniques in ISAC systems.
nodes can be readily achieved. We refer readers to [95] for
further details concerning the above solutions.
4) Future Directions and Potential Solutions: Again,
although numerous methods have been proposed to address the
clock synchronization issues in various application scenarios,
they may not be directly applicable to ISAC systems. The GPS
signal-based methods require a long synchronization time,
which limits their application in high-mobility scenarios. As
for signal processing aided techniques, they generally require
a multiple antenna receiver, and the synchronization accuracy
is lower than that of the GPS-based method. Therefore, how
to adapt these schemes to the ISAC system is still an open
question.
Applying UWB techniques in ISAC systems is indeed a
possible solution. Firstly, UWB schemes can help the ISAC
signal obtain high-resolution time stamps to realize a high-
precision measurement. Moreover, the UWB technique fits
naturally into ISAC systems, since a high carrier frequency
and a wide operating bandwidth will be used in future ISAC
systems to increase the data rate. However, although UWB
ISAC obtains excellent clock synchronization performance, it
prevents the application of algorithms based on the narrow-
band assumption. Furthermore, the synchronization scheme
based on UWB requires two rounds of signal trips between
transceivers [96], which is also unsuitable for high-mobility
scenarios. These issues still have to be circumvented.
The security of time information is also an open research
topic. As mentioned above, even a tiny synchronization error
will cause a considerable range estimation perturbation. In the
case of tracking automotive vehicles, the ghost targets caused
by asynchronous timing may lead to accidents. Therefore,
further research is required to secure the timing information in
the transmitted signals. We remark that in the near-field model,
the TOA estimation may be eliminated by exploiting the
distance information embedded in the array response [156]–
[158]. In ISAC systems, if the targets are also communication
users in location division multiple access (LDMA) [159], their
location can also be extracted from the communication unit.
It is envisaged that near-field ISAC has promising prospects
[160], [161].
Challenge 5: How Far Are the Existing Designs From the
Pareto-Optimal Boundary?
1) Background: In general, the Pareto-optimal boundary
can be interpreted as the optimal case in the “moderately cor-
related” scenarios of S&C channels of Fig. 10. Finding suitable
14
waveform designs and signaling strategies for approaching
the Pareto-optimal boundary is of considerable importance in
ISAC systems. At the time of writing, research has primarily
been focused on optimizing either radar performance metrics
under communication constraints or vice versa [162]. Never-
theless, due to the absence of a Pareto optimization framework
for ISAC systems, attaining an optimal performance trade-
off between the two functionalities remains challenging. The
underlying problem is to find the Pareto-optimal boundary [53]
and to measure the distance between the existing designs and
the Pareto-optimal boundary.
2) Existing Literature: Generally, the waveform design
in the existing ISAC research can be split into three cat-
egories: radar-centric design [163], communication-centric
design [164] and joint waveform design [165]. The radar-
centric approaches are implemented based on the radar probing
signals, such as that of pulse interval modulation (PIM) and
index modulation to map the communication symbols onto
the radar pulses [55], [98]. Similar methodologies, such as the
combination of amplitude/phase shift keying (ASK/PSK) and
linear frequency modulation (LFM) signals were also proved
to be effective for carrying information based on radar probing
signals [166], [167]. As for the communication-centric ap-
proaches, they are designed relying on existing communication
signals and protocols. One can extract the target information
of Doppler and delay parameters from the transmitted-and-
received signals by classical signal processing techniques, such
as the fast Fourier transform (FFT) and the fractional Fourier
transform (FrFT). In addition, the IEEE 802.11ad protocol
has also been adopted for radar sensing in vehicular networks
[136].
Compared to the approaches based on existing radar or com-
munication waveforms, the joint waveform design takes both
the S&C performance into consideration, which is deemed to
be a promising direction in ISAC systems as a benefit of its
capability of striking a favorable tradeoff between S&C. We
remark that sensing MI does not have an exact operational
interpretation and MI-oriented S&C tradeoffs based on (2)
may not be suitable for investigating Pareto-optimal design. In
contrast to the MI-oriented S&C tradeoffs in (2), we highlight
another representative formulation as [52]
min
Xρ∥HcX−S∥2
F+ (1 −ρ)∥X−X0∥2
F,(28a)
s.t.Specific Practical Constraints,(28b)
where Hc,X,S, and X0denote the communication channel
matrix, the transmitted signal matrix, the desired constellation
symbol matrix and the ideal radar waveform, respectively.
Thus, the first term in (28a) represents the multi-user in-
terference encountered in downlink communication, and the
second term forces Xto approach a well-designed pure
sensing waveform. The weighting factor ρ∈[0,1] controls
the priorities assigned to S&C functionalities.
3) Future Directions and Potential Solutions: While the
ISAC waveform design has been extensively studied in the
recent literature [163]–[165], the Pareto-optimal signaling
strategies still remain open challenges. First of all, it is
unclear how to define the joint performance metrics for the
Radar Constrained
Communication Constrained
Radar Performance
Communication Performance
ISAC Constrained
Fig. 13. Illustration of the Pareto boundary of achievable
performance region.
ISAC system, despite the abundance of existing radar-only
and communication-only metrics [35], [168]. For instance,
the signal-to-interference plus noise ratio (SINR), the achiev-
able sum rate, etc, are widely used in communication-only
waveform design. There are also popular metrics such as the
CRB, MMSE, and MI for radar-only design. Secondly, the
optimal waveform design is rather challenging due to the
design conflicts between radar and communication waveform
optimization, especially in light of the increased complexity
introduced by both the communication and radar constraints.
The Pareto boundary of the achievable performance region is
shown in the stylized Fig. 13 [53]. In a nutshell, there are
still several open questions to be explored, some of which are
closely related to the fundamental limits of ISAC, such as,
•Where is the Pareto boundary and how far are the existing
designs from the boundary of optimal design?
•Is there any opportunity to attain the Pareto-optimal
performance between sensing and communication based on
a unified waveform?
These questions may only be answered on a step-by-step
basis, commencing from simple twin-component objective
functions (OF) based on the bit error rate (BER) of com-
munication and the target detection probability of radar for
example. As a next step further, metrics may be added to the
OF one-by-one, which of course, expands the search-space.
Hence sophisticated reduced-scope search techniques must be
conceived.
Challenge 6: How Could We Improve the Sensing Perfor-
mance by the Emerging Super-Resolution Methods?
1) Background: Resolution is one of the most important
factors determining the sensing performance, especially for the
tasks of localization, imaging, and recognition. Most current
cellular networks (e.g., 4G and 5G) operating in sub-6G bands
can only provide meter-level accuracy sensing due to the low
range and angle resolutions, which are unable to meet the
requirements of demanding applications, such as V2X, IoT,
etc [30].
Although the recently proposed millimeter wave and
THz communication systems can potentially provide high-
resolution sensing service within their limited coverage area,
achieving super-resolution sensing in the sub-6G bands is
still of significance to reuse the current cellular network for
15
Fig. 14. Harnessing carrier aggregation.
sensing. It is widely exploited in radar signal processing that
the distance resolution is determined by the bandwidth Bof
the transmitted waveforms [169], i.e. ∆R=c/2B, where
cis the speed of light, and the angular resolution depends
on the size of the array aperture WNN [170], i.e., ∆θ=
2 arcsin(0.446λ/WNN), where λrepresents the wavelength of
the transmitted signal. Therefore, it is an appealing idea for
super-resolution methods to extend both the signal bandwidth
and the array aperture in the sub-6G bands.
2) Existing Literature:
•Carrier Aggregation Assisted Sensing: Carrier aggregation
(CA) is a powerful technique of harnessing multiple compo-
nent carriers across the available spectrum bands. As illus-
trated in Fig. 14, CA can be mainly classified into three types,
i.e., intra-band contiguous CA, intra-band non-contiguous CA,
and inter-band non-contiguous CA [99]. On the one hand,
the communication throughput and peak data rate can be
significantly improved through CA techniques [100]. As a
further benefit, the bandwidth aggregating several discrete
carriers is expected to achieve an improved distance resolution.
However, there are also several challenges to be addressed.
For instance, the propagation path loss and Doppler shift will
be quite different for the non-contiguous carrier components,
hence a well-designed carrier selection scheme and a matching
resource allocation method are required for circumventing
the degradation in S&C performance. Nevertheless, although
the CA technique has already been applied in operational
communication systems, there are numerous new challenges
when considering sensing functionality. For instance, the non-
contiguous CA may cause high sidelobes in the ambiguity
function of sensing, since the initial phases of discrete carriers
are not necessarily continuous, which leads to significant
sensing performance degradation.
•Sparse Array Based ISAC Platform: Compared to the
conventional uniform linear array (ULA), the core idea of
sparse arrays is converting the sample covariance matrix
of the sensor outputs into the so-called difference coarray
domain by combing two or more ULAs, which have in-
creased inter-sensor spacing, hence enlarging the virtual array
aperture [171]. Specifically, let Sbe an integer set of the
sensor locations. Then the difference set can be defined as
D={n1−n2|n1, n2∈S}. One of the typical sparse array
geometries is the nested array [101], which consists of a dense
1 2 3 4 812
02 3 4 6 9
0 1 2 3 4 5 6 7 9
Dense ULA Sparse ULA
Sparse ULA with separation 2
Sparse ULA with separation 3
Physical array
Effective array
0 1 2 3 4 5 6 7 8 9 10 11
Physical array
Nested array
Coprime array
Effective array
Antenna location
Array hole
Fig. 15. Array geometry of the 6-antenna nested array and
coprime array.
ULA with separation 1 (in unit of λ/2), and a sparse ULA
with sensor separation of (N1+ 1). The associated sensor
locations can be expressed by S={n|n= 1,··· , N1} ∪
{m(N1+ 1)|m= 1,··· , N2}. As illustrated in Fig. 15,
there are 6 physical sensors, with N1=N2= 3 forming
a difference coarray associated with integers spanning from 0
to 1, which significantly enlarges the array aperture compared
to the conventional ULA. Another popular array geometry is
coprime array [172], which consists of two sparse ULAs with
sensor separation determined by a coprime pair of integers N
and M. The sensor location set is defined as S={nM|n=
0,··· , N1} ∪ {mN |m= 1,··· ,2M−1}. Fig. 15 shows a
coprime array with M= 2 and N= 3 forming a difference
coarray containing consecutive lags from 0 to 7. Although
lag 8 is missing (usually called a ‘hole’), coprime arrays are
capable of reducing the mutual coupling effect compared to
nested arrays. By using the above sparse arrays, the angular
resolution is expected to be substantially improved thanks to
the increased virtual array aperture size. Moreover, from the
perspective of communication systems, using a sparse array
will not bring about intractable challenges for the transmission
and reception of communication symbols. Therefore, using a
sparse array constitutes a promising technique of acquiring
super-resolution at sub-6G bands for ISAC systems.
•Super-resolution Algorithms for ISAC: There are two
main categories of super-resolution algorithms for array signal
processing, namely the subspace-based angle estimation al-
gorithms, such as the multiple signal classification (MUSIC)
[173] and the estimation of signal parameters via rotational
invariance techniques (ESPRIT) [174]. These kinds of methods
are based on the data signal’s covariance and hence they
are very sensitive to noise, data snapshots, and source cor-
relations. Another family is constituted by the sparse signal
reconstruction frameworks, including on-grid [104], off-grid
[105], and gridless [106] algorithms. These methods exploit
the spatially sparse nature of the received signals rather than
their second-order statistics, where the latter is applicable for
limited snapshots, for an unknown number of sources, etc.
Furthermore, combining the sparse array geometry with the
super-resolution algorithms is also a promising scheme for
16
improving the resolution of ISAC systems.
3) Future Directions and Potential Solutions: The afore-
mentioned solutions rely on state-of-the-art techniques for
improving both the distance and angular resolutions of ISAC
systems. However, the unique challenges and opportunities
introduced by the ISAC techniques are still widely unexplored.
For example, the well-designed dual-functional radar and com-
munication signal is expected to improve joint bandwidth uti-
lization. Furthermore, high-resolution sensing can be achieved
by establishing a communication-assisted sensing framework
for cooperative networks.
IV. ISAC NETWOR KS & CROSS-LAYER DESIGN
In this section, we mainly explore Challenges 7-8 of Fig.
2and elaborate on a pair of key concepts in networked
sensing, including the cellular architecture as well as cross-
layer resource management and protocols. We remark that
there is a paucity of research focusing on these two concepts,
hence this section is more of a philosophical discussion of
potential questions and an outlook on future directions.
Challenge 7:What is the Potential Cellular Architecture of
Future Network of Sensing?
1) Background: For future radio access networks (RAN),
sensing is envisioned as a key functionality to be integrated
into the dense cellular architecture, which facilitates the
construction of network sensing based on sharing the dual-
functional transmitted signals and hardware implementation.
In this way, both the user equipment and cellular network can
sense the surroundings and enable a variety of applications
such as environmental monitoring and target detection.
2) Existing Literature: In the 5G cellular network, multi-
station cooperation based on the cloud RAN (C-RAN) concept
is widely used [110]. For communication-only systems, the
inter-cell interference of remote radio heads (RRHs) can be
mitigated by specifically designed interference management
strategies, such as soft fractional frequency reuse or coordi-
nated multi-point techniques [109]. Furthermore, cooperative
transmission in the scenario of user-centric C-RAN has also
been investigated, where usually global channel state infor-
mation is required [110]. In general, inter-cell interference is
decked to be a harmful factor that has to be reduced.
Central Processing Unit Interference
ISAC AP ISAC Signal Reflected Signal
Fig. 16. Potential cellular architecture in ISAC systems.
Multi-station cooperation is potentially capable of expand-
ing the detection range and/or improving the SNR as well
as the detection probability. A typical multi-station-multi-cell
cooperation scenario is illustrated in Fig. 16. Specifically, the
single target (e.g., UAV [175]) can be cooperatively detected
by multiple stations from different viewing angles, which
can provide accurate overall sensing of the detected target.
In this scenario, the cooperation of cells and stations is of
vital importance. However, the inter-cell interference in ISAC
networks may be rather different from that of communication-
only networks. Inter-station interference may contain useful
information with respect to the targets of interest, which has
to be exploited for enhancing the sensing performance, rather
than being canceled. In addition to receiving the echo signal
originating from the monostatic sensing operation, each station
may also receive ISAC interference transmitted by other BSs
or UEs.
3) Future Directions and Potential Solutions: Both the cell
architecture and the station location are important factors. The
cellular architecture of ISAC maybe expected to rely on the
existing station location resources, which may be a convenient
way of realizing the integration of communication and sensing
networks. However, the communication-only design principle
may not be appropriate for frequency reuse in terms of
functionalities and locations.
The cell-free massive MIMO architecture [111], [112],
[176], on the other hand, may be a promising one in support of
ISAC systems. In cell-free massive MIMO, the access points
(APs) can be distributed over a larger area and provide flexible
load-balancing, where a group of users may be supported by
a group of APs. A larger number of communication users and
targets can be cooperatively served or detected simultaneously
at a high S&C performance.
It is important to note that networked sensing holds sig-
nificant promise in terms of achieving seamless sensing in
the future, particularly with the proliferation of ISAC base
stations. In this context, these distributed base stations have
the capability to accumulate vast amounts of sensory data,
which is then relayed via backhaul links to the computational
processing unit for extensive data analysis and for the subse-
quent extraction of sensing information. This integration often
demands the seamless amalgamation of multistatic sensing,
cooperative communications, and computational elements [6].
However, networked sensing also has its own challenges that
require immediate attention. For example, the collaboration in
multistatic systems necessitates precise clock synchronization,
and the presence of clutter interference in complex envi-
ronments gravely impacts the overall efficacy of networked
sensing.
Challenge 8: How Could We Conceive Resource Manage-
ment Schemes and Protocols Specifically Tailored for the
ISAC Network?
1) Background: In ISAC network of the future, it can be
predicted that the request for sensing services arises usually
randomly and unexpectedly. Based on Fig. 16, we have to
conceive effective resource management schemes in response
to these sensing requests. In this spirit, the frame structure and
17
TABLE IV: Resource management schemes and protocols for
ISAC network.
Layers
Open Problems
PHY Layer
•
Adaptive power allocation
•
Adaptive frequency bandwidth allocation
MAC Layer
•
Sensing request priority declaration
•
Media access control algorithm
Network Layer
•
Queuing theory for efficient S&C services
•
Clustering algorithm
Application Layer
•
Source data encoding
•
Adaptive rate adjustment
the resource scheduling algorithms have to be tailor-made in
support of flawless ISAC services.
2) Existing Literature: The authors of [67] designed a
sophisticated PHY layer as well as a resource allocation
strategy and then evaluated the performance tradeoffs between
S&C. However, if the sensing services are heavily congested,
the ISAC-BS cannot respond to all the requests in time, since
the PHY layer resources, such as the bandwidth or time slots,
are limited. To satisfy the emerging burst of sensing requests
in ISAC networks, efficient cross-layer resource management
techniques have to be conceived. The cross-layer optimization
of traditional wireless communication has been widely inves-
tigated [113]–[116], but the cross-layer optimization of ISAC
networks is in its infancy, which may involve several open
system interconnections (OSI) layers [114]. We commence our
discourse with the PHY, followed by the media access control
(MAC), network, and application layer.
We briefly highlight the open problems in the cross-layer
optimization of ISAC networks in TABLE IV, and expound
on them in the following.
•PHY layer: Although noise and interference reduction
have been widely studied in the context of dedicated com-
munication or sensing functionality, handling the interference
caused by echo waves has become a challenging new problem
in the ISAC scenarios. Due to the shared use of limited
resources, such as spectrum and hardware, ISAC is expected to
enhance both the S&C performance. However, the interference
caused by echo waves plays different roles in communication
and sensing services. For instance, in wireless communication
scenarios, all propagation paths can be leveraged to improve
the communications performance, as valuable signal energy
may be gleaned from every path. Conversely, for sensing
services, only the specific paths reflected by the targets of
interest are desired. Therefore, to support ISAC, it is essential
to distinguish sensing echoes from the received multipath
signal, which remains an open issue that needs to be resolved.
•MAC layer: To handle the burst of sensing requests, a
feasible strategy is to prioritize S&C requests based on their
specific requirements. For instance, the sensing requests of
high-velocity vehicles should receive higher sensing priority
and better Quality of Service (QoS) than those of stationary
objects. For sensing requests with a higher priority, the BS
should respond promptly and allocate more resources to de-
liver high-specification services. To implement this feature,
the control frame structure of the Media Access Control
(MAC) layer has to be designed for low latency. As the
ISAC network simultaneously provides both S&C services, the
channel access procedure has evolved from that of traditional
wireless networks. For example, the Request To Send / Clear
To Send (RTS/CTS) channel access mechanism is commonly
used in 802.11 wireless networks to avoid frame collisions.
However, in the ISAC network, the control frames can also
serve sensing tasks. In a scenario where there is a bistatic
sensing link in the ISAC network, the sensing initiator sends
an RTS frame that has both S&C functions. Subsequently,
the sensing receiver has to send the sensing result back,
while the communication receiver sends the CTS back to the
sensing initiator. To avoid a collision between the feedback of
sensing results and wireless communication control frames, the
sequence of control frame interactions in the channel access
procedure has to be redesigned.
•Network layer: Compared to traditional wireless commu-
nication networks, the PHY layer of the ISAC network has
additional sensing tasks. To evaluate the QoS of both sensing
and communication, additional metrics such as sensing delay,
bandwidth utilization, and sensing accuracy must be consid-
ered. In traditional wireless networks, typically queuing theory
is used for efficiently scheduling communication. However, the
challenge in ISAC is how to incorporate sensing metrics into
queuing theory. In addition to queuing theory, clustering algo-
rithms can also be used to support S&C services. In contrast to
traditional sensing networks, users in ISAC networks may have
both S&C service requests, and may only be in communication
with a sensing target for a short period. Therefore, the sensing
clusters in ISAC evolve rapidly, making traditional clustering
algorithms unsuitable. To address this problem, agile adaptive
clustering algorithms must be developed. For example, the BS
and users that have sensed the same target can automatically
be grouped into a cluster, expediting the clustering process.
•Application layer: In the ISAC network, transmitting
results sensed by the BS to the user is a critical operation.
To improve transmission efficiency, effective source encod-
ing mechanisms can be harnessed for reducing the physical
resource requirements. As a benefit, more sensory data can
be transmitted within a given bandwidth. When a user sends
a sensing request to the BS, it can also provide the target
class, indicating whether it is a vehicle or a pedestrian.
This is advantageous because different types of sensing tar-
gets require varying amounts of sensory data. Vehicles or
pedestrians have a higher priority in sensing services, since
they have to signal their velocity, movement direction, and
location to describe their states. In contrast, stationary objects
such as trees or buildings require less data, perhaps only
their location. Based on this difference, we can employ a
more complex source data encoding scheme for high-priority
objects to improve the sensory data transmission efficiency,
while simultaneously giving cognizance to the compression
delay and complexity. By contrast, low-priority objects can be
encoded by low-complexity source encoding schemes, thereby
conserving computational resources.
3) Future Directions and Potential Solutions: The above
discussions highlighted the potential problems in the resource
18
Downlink bistatic
sensing signal
Sensory data leakage
TX 1 Pep
Target 2
Tx 3
Uplink bistatic
sensing signal
TX 2
TX 4
Target 1
Fig. 17. Sensing security and privacy scenarios.
management of ISAC network. The optimization strategies
were introduced layer-by-layer. However, since every opti-
mization action has the potential to increase the cross-layer
interaction, it is important to consider their long-term effects
right across the entire framework [177]. Moreover, in future
research on ISAC techniques, we have to formulate bespoke
optimization problems. We conclude by highlighting some of
the open problems at a glance in TABLE IV again.
V. ISAC APPLICATIONS
In this section, we critically appraise Challenges 9-10 of
Fig. 2and first highlight that the security and privacy issues
of wireless sensing systems have to be addressed before inte-
grating sensing into existing cellular networks and supporting
ISAC applications. Then, we consider a representative use case
related to the sensing of human activities by utilizing wireless
signals. Besides, we elaborate on a suite of open questions
and their potential solutions.
Challenge 9: How Could We Provide Security and Privacy
Guarantees Within ISAC Applications?
1) Background: In the next-generation era, sensing may
be viewed as a compelling service in support of emerging
applications, such as the smart home and V2I networks [30],
[67]. However, due to the open nature of the wireless sensing
medium, the sensing services incur potential security risks
and privacy issues [132]. Specifically, malicious entities may
overhear the CSI of the targets that itself may be confidential
target-related information [123]. Malicious agents may also
misuse localization information [119], [120] as well as the
information they inferred concerning target movements [123].
As a result, those malicious entities may become capable of
breaching the users’ privacy, or of contaminating the legitimate
reception. We refer to contamination of the legitimate signal as
a security problem and to privacy breaches as privacy problems
(mainly relevant to the legitimate target), respectively.
2) Existing Literature: In radar-communication spectrum
sharing scenarios, one has to guarantee sensing security for
the transmitter, since the transmit precoding matrix assigned to
the communication system contains implicit information about
the radar [119], [120]. Hence, the authors of [119] investigated
different precoder matrices to simulate an adversary inference
attack and characterized the associated risks. As a further
advance, the authors of [120] proposed another precoder
design by considering the tradeoff between the interference
power and radar security. The authors of [121] proposed to
apply a so-called CSI “fuzzer” to enhance the privacy of WiFi
CSI applications. On the other hand, although wireless sensing
is capable of potentially avoiding the direct leakage of the
targets’ images relying on visual acquisition [123], the chan-
nel impulse response might also contain private information
concerning the targets, such as peoples’ movement in human
action sensing [131]. Recent advances have revealed for exam-
ple optimal pilot designs conceived for eavesdropping-resistant
channel estimation [122], as a potential solution for protecting
the security of sensing services.
To provide further intuitions about the issues of sensing
security and privacy, let us elaborate further with the aid of Fig.
17. For example, in a monostatic downlink sensing scenario,
the transmitter (Tx 1) wishes to sense a legitimate target
(Target 1). It was reported that the precoding matrix might
contain the location of the radar transmitter [119]. Due to
the leakage of the standardized transmit waveform, a potential
unauthorized entity, a.k.a. a peeper (Pep) wishes to intercept
the designed waveform and contaminate it by malicious spatial
interference. Meanwhile, the sensing information that contains
the presence, the location, and the behavior of the targets may
be reflected by Target 2 and intercepted by Pep, in which case
the target’s privacy is leaked (see the dashed line in Fig. 17).
It may be anticipated that more and more sensing security and
privacy problems will occur in wireless sensing, as illustrated
for downlink/uplink bistatic sensing in Fig. 17.
3) Future Directions and Potential Solutions: In contrast
to classical pure communication security, the sensing systems
seek to convert the TIR Hsinstead of the transmit signals
X. While some of the existing communication security ap-
proaches may be borrowed to secure the sensing network
[178], there are still quite a lot of open issues to be addressed.
In the sequel, we briefly discuss potential solutions from
different perspectives.
•Information theoretic perspective: The leakage of target-
related information, which is usually carried by the
transmitted waveform, can be mitigated by optimiz-
ing the transmit signals in order to maximize the MI
I(Y;Hs|X)of the transmitter while minimizing the MI
at Pep.
•Physical layer perspective: To reduce sidelobe leakage of
the sensing beam, one can minimize the energy radiation
in the Pep’s direction in the spatial/angular domain by
utilizing transmit beamforming techniques, so that the
Pep cannot intercept the waveform leaked. If the Pep
is located within the spatial/angular beam of the target,
another potential solution to avoid interception is to add
artificial noise. More explicitly, one may specifically
condition the artificial noise that is only known by the
transmitter, while contaminating the Pep’s reception.
•MAC layer perspective: In typical wireless standards,
both the payload waveform and the pilot organization
are openly accessible, which may however cause privacy
19
Fig. 18. Radio signal processing pipeline for human sensing.
leakage in sensing systems. Hence innovative techniques
are required for protecting the S&C privacy and security
by conceiving sophisticated authentication and access
control solutions. By contrast, the legitimate transmit-
ters/receivers can be identified through location-, angle-,
Doppler- and even channel-aware secret key generation,
while protecting them from unauthorized parties.
Challenge 10: How Could We Achieve MOMT Recognition
by Wireless Signals?
1) Background: Identifying human targets and recognizing
their actions is essential for numerous context-aware applica-
tions, such as assisted living, health monitoring, and intelligent
transportation. Compared to the commonly used camera-based
human sensing solutions, the wireless signal-based approach is
more robust to environmental variations and its coverage tends
to be more ubiquitous. Camera-based and beamforming-aided
techniques may also be beneficially combined [179].
There are two main categories of wireless human sensing
methods, i.e. radar-based and WiFi-based human sensing. As a
benefit of its ability to estimate speed and range (time delay),
radar has natural advantages in terms of near-field human
sensing and has been increasingly employed in miniaturized
portable form. Given a common high-accuracy clock signal,
radar systems are capable of achieving precise transceiver
synchronization even under bistatic and networked deploy-
ments, facilitating accurate human-related feature extraction.
Furthermore, compared to WiFi systems, radar has higher
signal bandwidth and hence has finer range resolution in
separating multiple targets or components. Radar-based human
sensing is achieved by measuring the movement parameters of
human targets with the aid of the received echo signals, such as
range, Doppler speed, and angle. The general radar signal pro-
cessing pipeline of human sensing is illustrated in Fig. 18. As
shown in this figure, when considering a continuous frequency
modulation wave (FMCW) based radar as an example, the
reflected echo signals can be transformed into a time-varying
sequence of “fast time-slow time” snapshots. By performing
the fast Fourier transform (FFT) along the fast-time and the
slow-time dimensions, respectively, the 2D radar snapshots can
be converted into a “range-Doppler frequency” map. In this
way, a series of “fast time-slow time” data matrices can be
transformed into a 3D "time-range-Doppler frequency" data
cube, as shown in Fig. 19(a). The time-varying range and
Fig. 19. Sensing measurements that contain different informa-
tion [124]. (a) 3D time-Range-Doppler map, (b) 2-dimensional
(2D) time-Doppler map, (c) 2D time-range map, and (d) 2D
range-Doppler map.
Fig. 20. Time-Doppler frequency maps from a WiFi system
and a 5G NR system, respectively.
Doppler speed parameters of human targets can be estimated
With the aid of the 3D data cube. In addition, by compressing
one of the three dimensions, three types of radar maps can be
obtained. For instance, by summing the 3D data cube along
the range domain, a 2D “time-Doppler frequency” spectrogram
(see 19(b)) is acquired, which depicts the variations of human
moving speed versus time. By compressing the 3D cube
along the time domain, a “range-Doppler frequency” map
(see 19(d)) could be obtained, which can be employed for
distinguishing different targets according to their range and
speed information. Furthermore, with the aid of an antenna
array at the receiver, angular human target information can be
estimated using algorithms like multiple signal classification
(MUSIC) [180].
2) Existing Literature: Again, in ISAC systems, both the
pilot signals and data frames can be employed for sensing.
For sensing relying on their data frames, the transmitted
communication information has to be first demodulated from
the received signals, and then the demodulated data frame can
be exploited for sensing. Another option is to exploit the CSI
estimated with the aid of pilot symbols, which is by far the
most appealing solution. The CSI captures how human targets
and other objects in the environment affect the propagation of
ambient wireless signals. In this context, the classic MIMO-
20
aided OFDM technology and its diverse multicarrier relatives
can provide a 3-dimensional (3D) matrix of complex values,
showing the variations of multipath channels in the presence
of moving human targets in the time, frequency, and spatial
domains. Due to the low distance-range resolution caused by
the narrow bandwidth, ISAC systems cannot directly estimate
the target range. Hence the existing solutions tend to generally
estimate the relative range of the target [126]. Furthermore,
since no common clock signal is available in the current
communication systems, and the WiFi systems are mostly
bistatic, there is undesirable timing offset, carrier frequency
offset, and sampling offset in the received CSI measurements,
substantially degrading the performance of human sensing
tasks [95]. As a result, phase offset removal is a vital step for
human sensing. In addition to constructing a reference signal,
the commonly utilized single-node solutions can be classified
into cross-antenna cross-correlation (CACC) [181] and cross-
antenna signal ratio (CASR) aided approaches [182]. Based on
these phase removal solutions, both modeling-assisted [127]
and learning-enhanced techniques [128] have been proposed
for estimating human-related parameters and used for different
human sensing tasks.
In the realm of human activity recognition (HAR) using
radio signals, the process can be broadly categorized into fea-
ture extraction and activity identification, both of which benefit
from machine learning (ML) techniques used for enhancing
the sensing performance. In what follows, we briefly discuss
the ML methods of HAR:
•Feature Extraction: Radio features are obtained through
manual feature engineering or automatic machine learning
(ML) algorithms [18].Manual feature engineering typically
relies on the amplitude and phase of received signals, as
they are influenced by human activity [18], [124]. Time-
varying Doppler/micro-Doppler frequency shifts are effective
for single-person activity recognition, representing radial ve-
locities of various body parts [183]. Additionally, the domain-
independent body-coordinate velocity profile (BVP) combines
Doppler frequency, orientation, and location to characterize
human gestures [184]. In multi-person scenarios, the spatial
parameters like range and angle are essential for distinguishing
signals from different targets. ML, especially deep learn-
ing (DL), automates HAR-related feature extraction [185].
While early research combined convolutional neural networks
(CNNs) with radio-based HAR feature extraction, this ap-
proach can cause temporal information loss. To address this
issue, memory-enabled recurrent neural networks (RNNs) are
often adopted, despite their high computational complexity,
resulting in significantly improved performance [124].
•Activity Recognition: HAR algorithms fall into two cate-
gories: model-based and learning-based [186]. Model-based
approaches, like the Fresnel Zone and Angle of Arrival
models, mathematically define the relationship between hu-
man activity and signal variations. These approaches quantify
human movement-related parameters through signal dynamics
and hold the potential for fine-grained activity recognition
and exploring sensing limits. Learning-based algorithms focus
on mapping sensing measurements to human activity labels
using pre-extracted features [131], [187]. Both manually and
automatically extracted features can be used, but learning-
based methods often combine ML-based feature extraction and
classification for end-to-end activity classification. Common
ML classifiers include Hidden Markov Models (HMM), K-
nearest neighbors (kNN), and support vector machines (SVM),
while DL classifiers encompass fully connected neural net-
works, CNNs, RNNs, and more.
3) Future Directions and Potential Solutions: In practice,
both WiFi and cellular systems can be utilized for sensing.
The former is more suitable for indoor human sensing, since
the hardware is simple and the system’s power is lower, while
the latter is more promising for outdoor sensing tasks. We
used a pair of wireless communication systems, including i) a
WiFi system associated with 40 MHz bandwidth, 5.0 ms pulse
repetition interval (PRI), and 5.8 GHz central frequency and
ii) a 5G new radio (NR) BS system with 100 MHz bandwidth,
10.0 ms PRI, and 3.6 GHz central frequency, to collect sensing
measurements. Both the WiFi and NR systems collect signal
reflections from separately deployed receiver antennas, so they
are working in the bi-static sensing mode. Then, the time-
Doppler maps obtained from a WiFi system and a 5G NR
system are shown in Fig. 20 [18], respectively. The maps
describe the variations of Doppler frequencies, in which a
human jumps twice consecutively. It can be seen that the
intensity of the NR spectrogram is strong, indicating that the
NR system may be deemed more robust to interference and
has a wider coverage range for human sensing. Furthermore,
the lower PRI of WiFi results in a finer time-resolution at the
left of Fig. 20 [18].
Although a wide variety of solutions have been proposed
for wireless human sensing [18], [127], [128], [181], [182],
there are still numerous open problems to be solved. For
instance, most existing work merely focuses on single-person
sensing [180]. Initial attempts concerning multi-person sensing
tasks were carried out under the assumption that the signals
reflected from different persons are independent and separable
[131]. In cellular networks, beamforming is capable of sup-
pressing clutter [63], while carrier aggregation technology is
promising in terms of increasing the range resolution, hence
it is promising for improving the performance of multi-person
sensing. Additionally, device-free and device-based solutions
3can be combined for sensing multiple persons. Specifically,
by associating the device-free reflected signals and the device-
based tag information (e.g., the ID of mobile phones), multiple
persons can be separated more easily. Furthermore, the low
spatial resolution of the current ISAC devices (e.g., WiFi and
portable radars) limits the multi-target separation capability
of these wireless sensing systems. In future research, we
need to improve the spatial resolution of wireless sensing
systems and devise solutions for MOMT sensing. For example,
we have to investigate the potential communication-assisted
sensing by utilizing CSI to boost the sensing performance,
where the receivers may analyze a variety of channel features
3Device-free sensing involves detecting changes in the environment caused
by the presence of people or objects without the use of dedicated sensing
devices. On the other hand, device-based sensing involves using dedicated
sensing devices, such as cameras or sensors, to directly measure the properties
of the environment.
21
Reduction of Clock Synchronization Errors & Phase Offset
Pareto Boundary-Based Waveform Designs
CA-Assisted Sensing &Super-Resolution Algorithms
Exploration of Cell-Free Sensing & Network Sensing Performance
Resource Management & Protocols in Different OSI Layers
Proper Schemes for Sensing Security & Privacy Protection
Resolution Improvement for Multi-Object Multi-Task Sensing
MI-based S&C Tradeoffs
MI & Factor Graph Methods CSI Estimation From Sensory Data
Proper Measures for Integration & Coordination Gains
Hardware Testbed Standardization Commercialization
Theoretical Research
Infancy Investigation for ISAC
Fig. 21. Design guidelines and future research directions for ISAC systems.
to detect the presence and location of multiple targets. In
some applications, combining these solutions with cameras are
expected to be beneficial [179]. In B5G and 6G, one needs to
investigate more advanced technologies to enable the emerging
sensing scenarios where extracting human-related features and
performing activity recognition are difficult, such as integrated
human sensing and communications, ultra-reliable and low-
latency human sensing [124], and non-line-of-sight human
sensing [18].
VI. DESIGN GUIDELINES AND A BR IEF SUMMARY
In this section, we first provide some general design guide-
lines and future directions for ISAC systems based on the ten
challenges discussed in Fig. 2, which are summarized in Fig.
21. Then, we conclude by highlighting a range of take-home
messages.
A. Design Guidelines
As evidenced by more and more emerging 6G white papers
and WiFi 7 papers, ISAC has drawn significant attention from
major industrial enterprises, and its standardization is also
under discussion in 3GPP [188], [189]. As shown in Fig. 21,
we touched upon the most influential design factors based on
the discussions in Section II-V, ranging from the theoretical
foundations of ISAC, ISAC system designs, ISAC networks,
and ISAC applications.
Detection Probability
CRB/BCRB
Mutual Information
MMSE
SNR/SINR
Beampattern
Power Consumption
Throughput
Cost Functions
BER
AI
Testbed
Computation
RIS
RSMA
UAVs
Index Modulation
Techniques
Holographic Surfaces
mmWave/THz
Methodologies
Pareto-optimal
Design
Convex Optimization
Stochastic Geometry
Machine Learning
Information Theory
Functional
Optimization
Sensing
Both S&C
Communi-
cation
Fig. 22. Stylized factors affecting the Pareto-optimal design of
ISAC systems.
•Theoretical Foundations of ISAC: The holistic design of
ISAC systems has to take the theoretical foundations into
consideration. The role of MI in radar sensing systems has to
be clarified first, so that one may quantify the MI-based S&C
tradeoffs in the design of ISAC systems. As a further step,
it is pivotal to infer the inner linkage between sensory data
and communication CSI by relying on the MI-based approach.
Furthermore, it is important to characterize both the integration
and coordination gains by exploiting “simple but intuitive”
metrics.
•Physical-Layer System Design: As presented in Section
22
III, we have discussed three key aspects of ISAC system
designs, including clock synchronization, Pareto-optimal sig-
naling strategies and super-resolution methods. In order to
reliably integrate the S&C components, it is essential to
reduce the clock synchronization errors and phase offset in
distributed deployment scenarios in support of high-precision
ranging [95]. It is also important to explore the entire Pareto-
front of optimal ISAC solutions, while gradually including an
increasing number of S&C metrics, as the technology evolves
[53]. Finally, super-resolution algorithms, such as CA-assisted
radar sensing, may be exploited for improving the sensing
performance in a variety of emerging applications, such as
V2X and IoT scenarios [2], [30].
•ISAC Networks & Cross-Layer Design: To incorporate
an ISAC capability into the existing infrastructures, cell-
free sensing has to be explored [190], in which the sensing
capability may be provided as a basic service. In contrast to the
ubiquitous communication service, the sensing service tends
to be performed in a more random bursty manner. Hence, the
joint resource management schemes and protocols tailored for
ISAC networks are highly desirable and are expected to lead
to compelling services.
•ISAC Applications: The ISAC network of the future has
to be rolled out with security and privacy guarantees [191],
[192]. Some potential directions relying on the optimization of
both the PHY and MAC layers have been discussed in Section
V. Furthermore, to realize practical ISAC services relying
on commercial wireless devices, it is critical to improve the
sensing resolution for supporting MOMT sensing, in both the
smart home and in smart city applications [2], [124], [125].
In addition to the proposed design guidelines and directions
of ISAC systems, substantial efforts are required for promot-
ing the standardization and commercialization of ISAC. As
illustrated in Fig. 22, we highlight the general philosophy of
the Pareto-optimal design for ISAC systems, with a glimpse
of its pivotal factors. Specifically, we should choose multi-
component cost functions bearing in mind the salient de-
sign perspectives as well as methodologies, including convex
optimization [54], [162], [192], stochastic geometry [193],
machine learning and so on [90]–[92]. Furthermore, artificial
intelligence (AI) and testbed based on practical data sets are
required [2], [6], [91], [92]. Furthermore, in-depth over-the-
air ISAC and computation investigations are necessitated [60],
while relying on the latest advances both in S&C, such as
reflecting intelligent surface (RIS)-assisted ISAC [34], rate-
splitting multiple access (RSMA)-aided ISAC [48], UAVs
equipped with ISAC in the sky [175], and holographic ISAC
[194], just to name a few.
B. Summary
In this paper, we critically appraised the recent advances
and formulated ten open challenges in ISAC systems, some of
which have already made initial progress, while others are still
in the exploratory phase. Firstly, we introduced the theoretical
foundations of ISAC systems, starting with an introduction
to Challenge 1 on fundamental theory that concerns the
limitations of S&C performance. Then in Challenge 2, we
discussed how to infer the CSI from the sensory data and
presented several potential solutions and future directions.
In addition, we presented Challenge 3, which focuses on
integration and coordination gains, and proposed an informal
metric to quantify them. Furthermore, we have addressed the
design issues of ISAC systems in the context of Challenges 4-
6. More specifically, we have elaborated on the open problems
of clock synchronization, Pareto-optimal signaling strategies,
and super-resolution methods, along with their potential so-
lutions and future directions. We then continued by shifting
our focus to ISAC networks. By considering sensing as a
service in the cellular network of the future, we investigated
the potential cellular architectures as well as the cross-layer
resource management and the protocol design of networked
sensing in Challenge 7 and Challenge 8, respectively. Next,
we concentrated our attention on attractive ISAC applications.
We highlighted the associated sensing security and privacy
issues, proceeding by presenting the corresponding future
directions from the perspective of the PHY and MAC layer
in Challenge 9. Finally, we touched upon a human activity
sensing scenario relying on wireless signals and discussed the
open issues of MOMT sensing in Challenge 10.
REFERENCES
[1] F. Liu, C. Masouros, A. P. Petropulu, H. Griffiths, and L. Hanzo, “Joint
radar and communication design: Applications, state-of-the-art, and the
road ahead,” IEEE Trans. Commun., vol. 68, no. 6, pp. 3834–3862,
2020.
[2] J. Wang, N. Varshney, C. Gentile, S. Blandino, J. Chuang, and
N. Golmie, “Integrated sensing and communication: Enabling tech-
niques, applications, tools and data sets, standardization, and future
directions,” IEEE Internet Things J., vol. 9, no. 23, pp. 23 416–23440,
2022.
[3] D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li,
D. Niyato, O. Dobre, and H. V. Poor, “6G Internet of things: A
comprehensive survey,” IEEE Internet Things J., vol. 9, no. 1, pp. 359–
383, 2021.
[4] L. Leyva, D. Castanheira, A. Silva, A. Gameiro, and L. Hanzo,
“Cooperative multiterminal radar and communication: A new paradigm
for 6G mobile networks,” IEEE Veh. Technol. Mag., vol. 16, no. 4, pp.
38–47, 2021.
[5] S. Dang, O. Amin, B. Shihada, and M.-S. Alouini, “What should 6G
be?” Nature Electronics, vol. 3, no. 1, pp. 20–29, 2020.
[6] G. Zhu, Z. Lyu, X. Jiao, P. Liu, M. Chen, J. Xu, S. Cui, and P. Zhang,
“Pushing AI to wireless network edge: An overview on integrated
sensing, communication, and computation towards 6G,” Sci. China Inf.
Sci., vol. 66, no. 3, pp. 130–301, 2023.
[7] J. A. Stankovic, “Research directions for the Internet of Things,” IEEE
Internet Things J., vol. 1, no. 1, pp. 3–9, 2014.
[8] L. Ma, J. Lu, C. Gu, and J. Mao, “A wideband dual-circularly polarized,
simultaneous transmit and receive (star) antenna array for integrated
sensing and communication in IoT,” IEEE Internet Things J., vol. 10,
no. 7, pp. 6367–6376, 2023.
[9] M. Liu, M. Yang, H. Li, K. Zeng, Z. Zhang, A. Nallanathan, G. Wang,
and L. Hanzo, “Performance analysis and power allocation for coop-
erative ISAC networks,” IEEE Internet Things J., vol. 10, no. 7, pp.
6336–6351, 2023.
[10] C. Ren and L. Liu, “Towards full passive Internet of things: Symbiotic
localization and ambient backscatter communication,” IEEE Internet
Things J., doi=10.1109/JIOT.2023.3262779, 2023.
[11] S. Ma, Y. Zhang, H. Li, J. Sun, J. Shi, H. Zhang, C. Shen, and S. Li,
“Covert beamforming design for intelligent-reflecting-surface-assisted
IoT networks,” IEEE Internet Things J., vol. 9, no. 7, pp. 5489–5501,
2021.
[12] I. Bilik, O. Longman, S. Villeval, and J. Tabrikian, “The rise of radar for
autonomous vehicles: Signal processing solutions and future research
directions,” IEEE Signal Process. Mag., vol. 36, no. 5, pp. 20–31, 2019.
23
[13] D. P. M. Osorio, I. Ahmad, J. D. V. Sánchez, A. Gurtov, J. Scholliers,
M. Kutila, and P. Porambage, “Towards 6G-enabled internet of vehi-
cles: Security and privacy,” IEEE Open J. Commun. Soc., vol. 3, pp.
82–105, 2022.
[14] F. Liu, W. Yuan, C. Masouros, and J. Yuan, “Radar-assisted predictive
beamforming for vehicular links: Communication served by sensing,”
IEEE Trans. Wireless Commun., vol. 19, no. 11, pp. 7704–7719, 2020.
[15] X. Wang, Z. Fei, and Q. Wu, “Integrated sensing and communica-
tion for RIS assisted backscatter systems,” IEEE Internet Things J.,
doi=10.1109/JIOT.2023.3262867, 2023.
[16] H. Xu, Y. Fan, W. Li, and L. Zhang, “Wireless distributed consensus
for connected autonomous systems,” IEEE Internet Things J., vol. 10,
no. 9, pp. 7786–7799, 2023.
[17] K. Zheng, Q. Zheng, H. Yang, L. Zhao, L. Hou, and P. Chatzimisios,
“Reliable and efficient autonomous driving: The need for heteroge-
neous vehicular networks,” IEEE Commun. Mag., vol. 53, no. 12, pp.
72–79, 2015.
[18] X. Li, Y. Cui, J. A. Zhang, F. Liu, D. Zhang, and L. Hanzo, “Integrated
human activity sensing and communications,” IEEE Commun. Mag.,
2022.
[19] Q. Wu, J. Xu, Y. Zeng, D. W. K. Ng, N. Al-Dhahir, R. Schober,
and A. L. Swindlehurst, “A comprehensive overview on 5G-and-
beyond networks with UAVs: From communications to sensing and
intelligence,” IEEE J. Sel. Areas Commun., 2021.
[20] Z. Jing, J. Mu, X. Li, Q. Zhou, and Q. Tian, “Efficient fusion and
reconstruction for communication and sensing signals in green IoT
networks,” IEEE Internet Things J., vol. 10, no. 11, pp. 9319–9328,
2023.
[21] L. Liu, S. Zhang, R. Du, T. X. Han, and S. Cui, “Networked
sensing in 6G cellular networks: Opportunities and challenges,”
arXiv:2206.00493, 2022.
[22] Z. Feng, Z. Fang, Z. Wei, X. Chen, Z. Quan, and D. Ji, “Joint radar
and communication: A survey,” China Commun., vol. 17, no. 1, pp.
1–27, 2020.
[23] C. Sturm and W. Wiesbeck, “Waveform design and signal processing
aspects for fusion of wireless communications and radar sensing,” Proc.
IEEE, vol. 99, no. 7, pp. 1236–1259, 2011.
[24] A. Hassanien, M. G. Amin, Y. D. Zhang, and F. Ahmad, “Signaling
strategies for dual-function radar communications: An overview,” IEEE
Aerospace and Electronic Systems Magazine, vol. 31, no. 10, pp. 36–
45, 2016.
[25] B. Tan, Q. Chen, K. Chetty, K. Woodbridge, W. Li, and R. Piechocki,
“Exploiting WiFi channel state information for residential healthcare
informatics,” IEEE Commun. Mag., vol. 56, no. 5, pp. 130–137, 2018.
[26] L. Zheng, M. Lops, Y. C. Eldar, and X. Wang, “Radar and communi-
cation coexistence: An overview: A review of recent methods,” IEEE
Signal Process. Mag., vol. 36, no. 5, pp. 85–99, 2019.
[27] Y. Cui, F. Liu, X. Jing, and J. Mu, “Integrating sensing and communi-
cations for ubiquitous IoT: Applications, trends, and challenges,” IEEE
Network, vol. 35, no. 5, pp. 158–167, 2021.
[28] J. A. Zhang, F. Liu, C. Masouros, R. W. Heath, Z. Feng, L. Zheng,
and A. Petropulu, “An overview of signal processing techniques for
joint communication and radar sensing,” IEEE J. Sel. Topics Signal
Process., vol. 15, no. 6, pp. 1295–1315, 2021.
[29] A. Liu, Z. Huang, M. Li, Y. Wan, W. Li, T. X. Han, C. Liu, R. Du,
D. K. P. Tan, J. Lu, Y. Shen, F. Colone, and K. Chetty, “A survey on
fundamental limits of integrated sensing and communication,” IEEE
Commun. Surveys Tuts., vol. 24, no. 2, pp. 994–1034, 2nd quarter 2022.
[30] F. Liu, Y. Cui, C. Masouros, J. Xu, T. X. Han, Y. C. Eldar, and
S. Buzzi, “Integrated sensing and communications: Towards dual-
functional wireless networks for 6G and beyond,” IEEE J. Sel. Areas
Commun., vol. 40, no. 6, pp. 1728–1767, 2022.
[31] W. Zhou, R. Zhang, G. Chen, and W. Wu, “Integrated sensing and
communication waveform design: A survey,” IEEE Open J. Commun.
Soc., vol. 3, pp. 1930–1949, 2022.
[32] K. V. Mishra, M. B. Shankar, V. Koivunen, B. Ottersten, and S. A.
Vorobyov, “Toward millimeter-wave joint radar communications: A
signal processing perspective,” IEEE Signal Process. Mag., vol. 36,
no. 5, pp. 100–114, 2019.
[33] O. B. Akan and M. Arik, “Internet of radars: Sensing versus sending
with joint radar-communications,” IEEE Commun. Mag., vol. 58, no. 9,
pp. 13–19, 2020.
[34] R. Liu, M. Li, H. Luo, Q. Liu, and A. L. Swindlehurst, “Integrated
sensing and communication with reconfigurable intelligent surfaces:
Opportunities, applications, and future directions,” IEEE Wireless Com-
mun., vol. 30, no. 1, pp. 50–57, 2023.
[35] J. Wang, C. Jiang, H. Zhang, Y. Ren, K.-C. Chen, and L. Hanzo,
“Thirty years of machine learning: The road to Pareto-optimal wireless
networks,” IEEE Commun. Surveys Tuts., vol. 22, no. 3, pp. 1472–1514,
2020.
[36] W. Saad, M. Bennis, and M. Chen, “A vision of 6G wireless systems:
Applications, trends, technologies, and open research problems,” IEEE
Network, vol. 34, no. 3, pp. 134–142, 2019.
[37] K. Chen, C. Qi, O. A. Dobre, and G. Y. Li, “Simultaneous beam
training and target sensing in ISAC systems with RIS,” IEEE Trans.
Wireless Commun., 2023.
[38] Z. Du, F. Liu, W. Yuan, C. Masouros, Z. Zhang, S. Xia, and G. Caire,
“Integrated sensing and communications for V2I networks: Dynamic
predictive beamforming for extended vehicle targets,” IEEE Trans.
Wireless Commun. doi: 10.1109/TWC.2022.3219890, 2022.
[39] W. Chen, L. Li, Z. Chen, T. Quek, and S. Li, “Enhancing THz/mmWave
network beam alignment with integrated sensing and communication,”
IEEE Commun. Lett., 2022.
[40] Y. Sun, J. Li, T. Zhang, R. Wang, X. Peng, X. Han, and H. Tan,
“An indoor environment sensing and localization system via mmWave
phased array,” J. Commun. Inf. Netw., vol. 7, no. 4, pp. 383–393, 2022.
[41] S. Huang, M. Zhang, Y. Gao, and Z. Feng, “MIMO radar aided
mmwave time-varying channel estimation in MU-MIMO V2X com-
munications,” IEEE Trans. Wireless Commun., vol. 20, no. 11, pp.
7581–7594, 2021.
[42] H. Sarieddeen, N. Saeed, T. Y. Al-Naffouri, and M.-S. Alouini, “Next
generation TeraHertz communications: A rendezvous of sensing, imag-
ing, and localization,” IEEE Commun. Mag., vol. 58, no. 5, pp. 69–75,
2020.
[43] Z. Chen, C. Han, Y. Wu, L. Li, C. Huang, Z. Zhang, G. Wang, and
W. Tong, “Terahertz wireless communications for 2030 and beyond:
A cutting-edge frontier,” IEEE Commun. Mag., vol. 59, no. 11, pp.
66–72, 2021.
[44] X. Yu, K. Alhujaili, G. Cui, and V. Monga, “MIMO radar waveform
design in the presence of multiple targets and practical constraints,”
IEEE Trans. Signal Process., vol. 68, pp. 1974–1989, 2020.
[45] G. Cui, J. Liu, H. Li, and B. Himed, “Signal detection with noisy
reference for passive sensing,” Signal Process., vol. 108, pp. 389–399,
2015.
[46] T. Zhang, S. Wang, G. Li, F. Liu, G. Zhu, and R. Wang, “Accelerating
edge intelligence via integrated sensing and communication,” in ICC
2022-IEEE Int. Conf. on Commun. IEEE, 2022, pp. 1586–1592.
[47] G. Zhou, Y. Mao, and B. Clerckx, “Rate-splitting multiple access for
multi-antenna downlink communication systems: Spectral and energy
efficiency tradeoff,” IEEE Trans. Wireless Commun., 2021.
[48] Y. Mao, O. Dizdar, B. Clerckx, R. Schober, P. Popovski, and H. V.
Poor, “Rate-splitting multiple access: Fundamentals, survey, and future
research trends,” IEEE Commun. Surveys Tuts., 2022.
[49] X. Chen, H.-H. Chen, and W. Meng, “Cooperative communications
for cognitive radio networks—From theory to applications,” IEEE
Commun. Surveys Tuts., vol. 16, no. 3, pp. 1180–1192, 2014.
[50] F. Liu, L. Zheng, Y. Cui, C. Masouros, A. P. Petropulu, H. Griffiths,
and Y. C. Eldar, “Seventy years of radar and communications: The road
from separation to integration,” IEEE Signal Process. Mag., vol. 40,
no. 5, pp. 106–121, 2023.
[51] R. Liu, M. Li, Q. Liu, and A. L. Swindlehurst, “Dual-functional radar-
communication waveform design: A symbol-level precoding approach,”
IEEE J. Sel. Topics Signal Process., vol. 15, no. 6, pp. 1316–1331,
2021.
[52] F. Liu, L. Zhou, C. Masouros, A. Li, W. Luo, and A. Petropulu, “To-
ward dual-functional radar-communication systems: Optimal waveform
design,” IEEE Trans. Signal Process., vol. 66, no. 16, pp. 4264–4279,
2018.
[53] L. Chen, F. Liu, W. Wang, and C. Masouros, “Joint radar-
communication transmission: A generalized Pareto optimization frame-
work,” IEEE Trans. Signal Process., vol. 69, pp. 2752–2765, 2021.
[54] X. Liu, T. Huang, N. Shlezinger, Y. Liu, J. Zhou, and Y. C. Eldar,
“Joint transmit beamforming for multiuser MIMO communications and
MIMO radar,” IEEE Trans. Signal Process., vol. 68, pp. 3929–3944,
2020.
[55] T. Huang, N. Shlezinger, X. Xu, Y. Liu, and Y. C. Eldar, “MAJoRCom:
A dual-function radar communication system using index modulation,”
IEEE Trans. Signal Process., vol. 68, pp. 3423–3438, 2020.
[56] X. Yu, X. Yao, J. Yang, L. Zhang, L. Kong, and G. Cui, “Integrated
waveform design for MIMO radar and communication via spatio-
spectral modulation,” IEEE Trans. Signal Process., vol. 70, pp. 2293–
2305, 2022.
24
[57] Z. Xiao and Y. Zeng, “Waveform design and performance analysis for
full-duplex integrated sensing and communication,” IEEE J. Sel. Areas
Commun., vol. 40, no. 6, pp. 1823–1837, 2022.
[58] C. Xu, B. Clerckx, S. Chen, Y. Mao, and J. Zhang, “Rate-splitting
multiple access for multi-antenna joint radar and communications,”
IEEE J. Sel. Topics Signal Process., vol. 15, no. 6, pp. 1332–1347,
2021.
[59] J. Qian, P. Xu, Z. Liu, N. Fu, and S. Wang, “Enhancement for spectrum
compatibility by mutual interference management,” IEEE Wireless
Commun. Lett., 2023.
[60] X. Li, Y. Gong, K. Huang, and Z. Niu, “Over-the-air integrated sensing,
communication, and computation in IoT networks,” IEEE Wireless
Communications, vol. 30, no. 1, pp. 32–38, 2023.
[61] Z. Wei, H. Qu, Y. Wang, X. Yuan, H. Wu, Y. Du, K. Han, N. Zhang,
and Z. Feng, “Integrated sensing and communication signals toward
5G-A and 6G: A survey,” IEEE Internet Things J., vol. 10, no. 13, pp.
11 068–11 092, 2023.
[62] ITU-R WP5D, “Draft New Recommendation ITU-R M. [IMT.
FRAMEWORK FOR 2030 AND BEYOND],” 2023.
[63] S. Lu, F. Liu, and L. Hanzo, “The degrees-of-freedom in monostatic
ISAC channels: NLoS exploitation vs. reduction,” IEEE Trans. Veh.
Technol., vol. 72, no. 2, pp. 2643–2648, 2023.
[64] Y. Liu, G. Liao, J. Xu, Z. Yang, and Y. Zhang, “Adaptive OFDM
integrated radar and communications waveform design based on infor-
mation theory,” IEEE Commun. Lett., vol. 21, no. 10, pp. 2174–2177,
2017.
[65] A. Pezeshki, A. R. Calderbank, W. Moran, and S. D. Howard, “Doppler
resilient Golay complementary waveforms,” IEEE Trans. Inf. Theory,
vol. 54, no. 9, pp. 4254–4266, 2008.
[66] Y. Xiong, F. Liu, Y. Cui, W. Yuan, T. X. Han, and G. Caire, “On the
fundamental tradeoff of integrated sensing and communications under
Gaussian channels,” IEEE Trans. Inf. Theory, vol. 69, no. 9, pp. 5723–
5751, 2023.
[67] F. Dong, F. Liu, Y. Cui, W. Wang, K. Han, and Z. Wang, “Sensing
as a service in 6G perceptive networks: A unified framework for isac
resource allocation,” IEEE Trans. Wireless Commun., vol. 22, no. 5,
pp. 3522–3536, 2023.
[68] C. Shi, F. Wang, S. Salous, and J. Zhou, “Joint subcarrier assignment
and power allocation strategy for integrated radar and communications
system based on power minimization,” IEEE Sensors J., vol. 19, no. 23,
pp. 11 167–11 179, 2019.
[69] X. Wang, Z. Fei, J. A. Zhang, and J. Xu, “Partially-connected hybrid
beamforming design for integrated sensing and communication sys-
tems,” IEEE Trans. Commun., vol. 70, no. 10, pp. 6648–6660, 2022.
[70] Y. Liu, G. Liao, J. Xu, Z. Yang, and Y. Zhang, “Adaptive OFDM
integrated radar and communications waveform design based on infor-
mation theory,” IEEE Commun. Lett., vol. 21, no. 10, pp. 2174–2177,
2017.
[71] K. Meng, Q. Wu, S. Ma, W. Chen, and T. Q. S. Quek, “UAV trajectory
and beamforming optimization for integrated periodic sensing and
communication,” IEEE Wireless Commun. Lett., vol. 11, no. 6, pp.
1211–1215, 2022.
[72] K. Meng, X. He, Q. Wu, and D. Li, “Multi-UAV collaborative sensing
and communication: Joint task allocation and power optimization,”
IEEE Trans. Wireless Commun., vol. 22, no. 6, pp. 4232–4246, Jun.
2023.
[73] K. Meng, Q. Wu, S. Ma, W. Chen, K. Wang, and J. Li, “Throughput
maximization for UAV-enabled integrated periodic sensing and com-
munication,” IEEE Trans. Wireless Commun., vol. 22, no. 1, Jan. 2023.
[74] Y. Yang and R. S. Blum, “MIMO radar waveform design based on
mutual information and minimum mean-square error estimation,” IEEE
Trans. Aerosp. Electron. Syst., vol. 43, no. 1, pp. 330–343, 2007.
[75] B. Tang, J. Tang, and Y. Peng, “MIMO radar waveform design
in colored noise based on information theory,” IEEE Trans. Signal
Process., vol. 58, no. 9, pp. 4684–4697, 2010.
[76] Y. Shen and M. Z. Win, “Fundamental limits of wideband localization –
Part I: A general framework,” IEEE Trans. Inf. Theory, vol. 56, no. 10,
pp. 4956–4980, 2010.
[77] D. Guo, S. Shamai, and S. Verdú, “Mutual information and minimum
mean-square error in Gaussian channels,” IEEE Trans. Inf. Theory,
vol. 51, no. 4, pp. 1261–1282, 2005.
[78] M. Kobayashi, G. Caire, and G. Kramer, “Joint state sensing and
communication: Optimal tradeoff for a memoryless case,” in Proc.
IEEE Int. Symp. Inf. Theory, 2018, pp. 111–115.
[79] Y. Liu, M. Li, A. Liu, J. Lu, and T. X. Han, “Generalized modeling
and fundamental limits for multiple-access integrated sensing and
communication systems,” arXiv:2205.05328, 2022.
[80] H. Hua, T. X. Han, and J. Xu, “MIMO integrated sensing and
communication: CRB-rate tradeoff,” IEEE Trans. Wireless Commun.,
doi=10.1109/TWC.2023.3303326, pp. 1–1, 2023.
[81] Z. Ren, Y. Peng, X. Song, Y. Fang, L. Qiu, L. Liu, D. W. K. Ng, and
J. Xu, “Fundamental CRB-rate tradeoff in multi-antenna ISAC systems
with information multicasting and multi-target sensing,” IEEE Trans.
Wireless Commun., doi=10.1109/TWC.2023.3312723, pp. 1–1, 2023.
[82] C. Ouyang, Y. Liu, and H. Yang, “On the performance of uplink ISAC
systems,” IEEE Commun. Lett., vol. 26, no. 8, pp. 1769–1773, 2022.
[83] C. Ouyang, Y. Liu, H. Yang, and N. Al-Dhahir, “Integrated sensing
and communications: A mutual information-based framework,” IEEE
Commun. Mag., vol. 61, no. 5, pp. 26–32, 2023.
[84] X. Yuan, Z. Feng, J. A. Zhang, W. Ni, R. P. Liu, Z. Wei, and C. Xu,
“Spatio-temporal power optimization for MIMO joint communication
and radio sensing systems with training overhead,” IEEE Trans. Veh.
Technol., vol. 70, no. 1, pp. 514–528, 2020.
[85] J. Li, G. Zhou, T. Gong, and N. Liu, “A framework for mutual
information-based MIMO integrated sensing and communication beam-
forming design,” arXiv:2211.07887, 2022.
[86] H. Joudeh and F. M. Willems, “Joint communication and binary state
detection,” IEEE J. Sel. Areas Inf. Theory, vol. 3, no. 1, pp. 113–124,
2022.
[87] W. Yuan, Z. Wei, S. Li, J. Yuan, and D. W. K. Ng, “Integrated
sensing and communication-assisted orthogonal time frequency space
transmission for vehicular networks,” IEEE J. Sel. Topics Signal
Process., vol. 15, no. 6, pp. 1515–1528, 2021.
[88] K. Wu, J. A. Zhang, X. Huang, and Y. J. Guo, “Integrating low-
complexity and flexible sensing into communication systems,” IEEE
J. Sel. Areas Commun., vol. 40, no. 6, pp. 1873–1889, 2022.
[89] Z. Huang, K. Wang, A. Liu, Y. Cai, R. Du, and T. X. Han, “Joint pilot
optimization, target detection and channel estimation for integrated
sensing and communication systems,” IEEE Trans. Wireless Commun.,
vol. 21, no. 12, pp. 10 351–10 365, 2022.
[90] U. Demirhan and A. Alkhateeb, “Radar aided 6G beam prediction:
Deep learning algorithms and real-world demonstration,” in Proc. IEEE
Wireless Commun. Netw. Conf. (WCNC), 2022, pp. 2655–2660.
[91] S. Wu, M. Alrabeiah, C. Chakrabarti, and A. Alkhateeb, “Blockage
prediction using wireless signatures: Deep learning enables real-world
demonstration,” IEEE Open J. Commun. Soc., vol. 3, pp. 776–796,
2022.
[92] A. Alkhateeb, G. Charan, T. Osman, A. Hredzak, J. Morais,
U. Demirhan, and N. Srinivas, “Deepsense 6G: A large-scale real-world
multi-modal sensing and communication dataset,” IEEE Commun.
Mag., vol. 61, no. 9, pp. 122–128, 2023.
[93] S. Lu, X. Meng, Z. Du, Y. Xiong, and F. Liu, “On the performance
gain of integrated sensing and communications: A subspace correlation
perspective,” in Proc. IEEE ICC, Jul. 2023, pp. 1–6.
[94] W.-Q. Wang, “GPS-based time & phase synchronization processing for
distributed SAR,” IEEE Trans. Aerosp. Electron. Syst., vol. 45, no. 3,
pp. 1040–1051, 2009.
[95] J. A. Zhang, K. Wu, X. Huang, Y. J. Guo, D. Zhang, and R. W. Heath,
“Integration of radar sensing into communications with asynchronous
transceivers,” IEEE Commun. Mag., vol. 60, no. 11, pp. 106–112, 2022.
[96] B. Xue, Z. Li, P. Lei, Y. Wang, and X. Zou, “Wicsync: A wireless
multi-node clock synchronization solution based on optimized UWB
two-way clock synchronization protocol,” Measurement, vol. 183, p.
109760, 2021.
[97] J. J. Pérez-Solano, S. Felici-Castell, A. Soriano-Asensi, and J. Segura-
Garcia, “Time synchronization enhancements in wireless networks with
ultra wide band communications,” Comput. Commun., vol. 186, pp.
80–89, 2022.
[98] R. M. Mealey, “A method for calculating error probabilities in a radar
communication system,” IEEE Trans. Space Electron. Telemetry, vol. 9,
no. 2, pp. 37–42, 1963.
[99] H. Lee, S. Vahid, and K. Moessner, “A survey of radio resource man-
agement for spectrum aggregation in LTE-Advanced,” IEEE Commun.
Surveys Tuts., vol. 16, no. 2, pp. 745–760, 2014.
[100] H. Al-Hraishawi, N. Maturo, E. Lagunas, and S. Chatzinotas, “Schedul-
ing design and performance analysis of carrier aggregation in satellite
communication systems,” IEEE Trans. Veh. Technol., vol. 70, no. 8,
pp. 7845–7857, 2021.
[101] P. Pal and P. P. Vaidyanathan, “Nested arrays: A novel approach to array
processing with enhanced degrees of freedom,” IEEE Trans. Signal
Process., vol. 58, no. 8, pp. 4167–4181, 2010.
[102] S. Qin, Y. D. Zhang, and M. G. Amin, “Generalized coprime array
configurations for direction-of-arrival estimation,” IEEE Trans. Signal
Process., vol. 63, no. 6, pp. 1377–1390, 2015.
25
[103] Z. Ni, J. A. Zhang, and R. P. Liu, “Waveform optimizations using
virtual arrays in broadband radar communications,” IEEE Wireless
Commun. Lett., 2023.
[104] D. Malioutov, M. Cetin, and A. Willsky, “A sparse signal reconstruction
perspective for source localization with sensor arrays,” IEEE Trans.
Signal Process., vol. 53, no. 8, pp. 3010–3022, 2005.
[105] Y. Chi, L. L. Scharf, A. Pezeshki, and A. R. Calderbank, “Sensitivity to
basis mismatch in compressed sensing,” IEEE Trans. Signal Process.,
vol. 59, no. 5, pp. 2182–2195, 2011.
[106] Z. Yang and L. Xie, “On gridless sparse methods for line spectral
estimation from complete and incomplete data,” IEEE Trans. Signal
Process., vol. 63, no. 12, pp. 3139–3153, 2015.
[107] P. Sarangi, M. C. Hücümeno ˘
glu, and P. Pal, “Single-snapshot nested
virtual array completion: Necessary and sufficient conditions,” IEEE
Signal Process. Lett., vol. 29, pp. 2113–2117, 2022.
[108] P. Sarangi, M. C. Hucumenoglu, R. Rajamaki, and P. Pal, “Super-
resolution with sparse arrays: A non-asymptotic analysis of spatio-
temporal trade-offs,” arXiv preprint arXiv:2301.01734, 2023.
[109] M. Peng, C. Wang, V. Lau, and H. V. Poor, “Fronthaul-constrained
cloud radio access networks: Insights and challenges,” IEEE Wireless
Commun., vol. 22, no. 2, pp. 152–160, 2015.
[110] C. Pan, M. Elkashlan, J. Wang, J. Yuan, and L. Hanzo, “User-centric
C-RAN architecture for ultra-dense 5G networks: Challenges and
methodologies,” IEEE Commun. Mag., vol. 56, no. 6, pp. 14–20, 2018.
[111] H. Q. Ngo, A. Ashikhmin, H. Yang, E. G. Larsson, and T. L. Marzetta,
“Cell-free massive MIMO versus small cells,” IEEE Trans. Wireless
Commun., vol. 16, no. 3, pp. 1834–1850, 2017.
[112] E. Björnson and E. Jorswieck, Optimal resource allocation in coordi-
nated multi-cell systems. Now Publishers Inc, 2013.
[113] T. Kaur and D. Kumar, “QoS mechanisms for MAC protocols in
wireless sensor networks: A survey,” IET Commun., vol. 13, no. 14,
pp. 2045–2062, 2019.
[114] W. Zhang, X. Wei, G. Han, and X. Tan, “An energy-efficient ring
cross-layer optimization algorithm for wireless sensor networks,” IEEE
Access, vol. 6, pp. 16 588–16 598, 2018.
[115] H. A. Khattak, Z. Ameer, U. I. Din, and M. K. Khan, “Cross-layer
design and optimization techniques in wireless multimedia sensor
networks for smart cities,” Comput. Sci. Inf. Syst., vol. 16, no. 1, pp.
1–17, 2019.
[116] G. Callebaut, G. Ottoy, and L. Van der Perre, “Cross-layer framework
and optimization for efficient use of the energy budget of iot nodes,”
in Proc. IEEE Wireless Commun. Netw. Conf. (WCNC), Apr. 2019, pp.
1–6.
[117] O. Günlü, M. Bloch, R. F. Schaefer, and A. Yener, “Secure joint
communication and sensing,” in 2022 IEEE International Symposium
on Information Theory (ISIT), 2022, pp. 844–849.
[118] O. Günlü, M. R. Bloch, R. F. Schaefer, and A. Yener, “Secure integrated
sensing and communication,” IEEE J. Sel. Areas Inf. Theory, vol. 4,
pp. 40–53, 2023.
[119] A. Dimas, B. Li, M. Clark, K. Psounis, and A. Petropulu, “Spectrum
sharing between radar and communication systems: Can the privacy of
the radar be preserved?” in Proc. 51st Asilomar Conf. Signals, Syst.,
Comput. IEEE, 2017, pp. 1285–1289.
[120] A. Al Hilli, A. Petropulu, and K. Psounis, “MIMO radar privacy
protection through gradient enforcement in shared spectrum scenarios,”
in 2019 IEEE International Symposium on Dynamic Spectrum Access
Networks (DySPAN). IEEE, 2019, pp. 1–5.
[121] X. Jiao, M. Mehari, W. Liu, M. Aslam, and I. Moerman, “OpenWiFi
CSI fuzzer for authorized sensing and covert channels,” in Proc. 14th
ACM Conf. Secur. Privacy Wireless Mobile Netw., New York, NY, USA,
Jun. 2021, pp. 377–379.
[122] Q. Zhu, S. Wu, and Y. Hua, “Optimal pilots for anti-eavesdropping
channel estimation,” IEEE Trans. Signal Process., vol. 68, pp. 2629–
2644, 2020.
[123] Z. Wang, Z. Meng, K. Saho, K. Uemura, N. Nojiri, and L. Meng, “Deep
learning-based elderly gender classification using Doppler radar,” Per-
sonal and Ubiquitous Computing, vol. 26, no. 4, pp. 1067–1079, 2022.
[124] X. Li, Y. He, and X. Jing, “A survey of deep learning-based human
activity recognition in radar,” Remote Sens., vol. 11, no. 9, p. 1068,
2019.
[125] P. Zhao, C. X. Lu, J. Wang, C. Chen, W. Wang, N. Trigoni, and
A. Markham, “Human tracking and identification through a millimeter
wave radar,” Ad Hoc Networks, vol. 116, p. 102475, 2021.
[126] K. Qian, C. Wu, Z. Yang, Y. Liu, and K. Jamieson, “Widar: Decimeter-
level passive tracking via velocity monitoring with commodity Wi-Fi,”
in Proc. 18th ACM Int. Symp. Mobile Ad Hoc Netw. Comput., 2017,
pp. 1–10.
[127] D. Wu, Y. Zeng, F. Zhang, and D. Zhang, “WiFi CSI-based device-
free sensing: From fresnel zone model to CSI-ratio model,” CCF Trans.
Pervasive Comput. Interact., pp. 1–15, 2021.
[128] Z. Chen, L. Zhang, C. Jiang, Z. Cao, and W. Cui, “WiFi CSI based
passive human activity recognition using attention based BLSTM,”
IEEE Trans. Mobile Comput., vol. 18, no. 11, pp. 2714–2724, 2018.
[129] X. Li, Y. He, F. Fioranelli, and X. Jing, “Semisupervised human activity
recognition with radar micro-Doppler signatures,” IEEE Trans. Geosci.
Remote Sens., 2021.
[130] S. Di Domenico, M. De Sanctis, E. Cianca, and M. Ruggieri, “WiFi-
based through-the-wall presence detection of stationary and moving
humans analyzing the Doppler spectrum,” IEEE Aerosp. Electron. Syst.
Mag., vol. 33, no. 5-6, pp. 14–19, 2018.
[131] Y. Zeng, D. Wu, J. Xiong, J. Liu, Z. Liu, and D. Zhang, “Multisense:
Enabling multi-person respiration sensing with commodity WiFi,”
Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., vol. 4, no. 3,
pp. 1–29, 2020.
[132] H. M. Furqan, M. S. J. Solaija, H. Türkmen, and H. Arslan, “Wireless
communication, sensing, and REM: A security perspective,” IEEE
Open J. Commun. Soc., vol. 2, pp. 287–321, 2021.
[133] D. Zhang, D. Wu, K. Niu, X. Wang, F. Zhang, J. Yao, D. Jiang, and
F. Qin, “Practical issues and challenges in CSI-based integrated sensing
and communication,” in Proc. IEEE Int. Conf. Commun. Workshops
(ICC Workshops). IEEE, 2022, pp. 836–841.
[134] R. Gao, M. Zhang, J. Zhang, Y. Li, E. Yi, D. Wu, L. Wang, and
D. Zhang, “Towards position-independent sensing for gesture recog-
nition with Wi-Fi,” Proc. ACM Interact. Mob. Wearable Ubiquitous
Technol., vol. 5, no. 2, pp. 1–28, 2021.
[135] B. Yu, Y. Wang, K. Niu, Y. Zeng, T. Gu, L. Wang, C. Guan, and
D. Zhang, “WiFi-Sleep: sleep stage monitoring using commodity Wi-
Fi devices,” IEEE Internet Things J., vol. 8, no. 18, pp. 13 900–13913,
2021.
[136] P. Kumari, J. Choi, N. González-Prelcic, and R. W. Heath, “IEEE
802.11ad-based radar: An approach to joint vehicular communication-
radar system,” IEEE Trans. Veh. Technol., vol. 67, no. 4, pp. 3012–
3027, 2018.
[137] J. Wang, X.-D. Liang, L.-Y. Chen, L.-N. Wang, and K. Li, “First
demonstration of joint wireless communication and high-resolution
SAR imaging using airborne MIMO radar system,” IEEE Trans.
Geosci. Remote Sens., vol. 57, no. 9, pp. 6619–6632, 2019.
[138] K. Niu, X. Wang, F. Zhang, R. Zheng, Z. Yao, and D. Zhang, “Rethink-
ing Doppler effect for accurate velocity estimation with commodity
WiFi devices,” IEEE J. Sel. Areas Commun., 2022.
[139] Q. Zhang and X. Gao, “Joint communication and sensing enabled
cooperative perception testbed for connected automated vehicles,” in
IEEE Conf. Comput. Commun. Workshops (INFOCOM WKSHPS).
IEEE, 2022, pp. 1–2.
[140] T. Kailath, “The divergence and Bhattacharyya distance measures in
signal selection,” IEEE Trans. Commun., vol. 15, no. 1, pp. 52–60,
1967.
[141] B. Tang, M. M. Naghsh, and J. Tang, “Relative entropy-based wave-
form design for MIMO radar detection in the presence of clutter and
interference,” IEEE Trans. Signal Process., vol. 63, no. 14, pp. 3783–
3796, 2015.
[142] M. Thomas and A. T. Joy, Elements of information theory. Wiley-
Interscience, 2006.
[143] A. Goldsmith, Wireless communications. Cambridge University press,
2005.
[144] A. Li, C. Shen, X. Liao, C. Masouros, and A. L. Swindlehurst, “Prac-
tical interference exploitation precoding without symbol-by-symbol
optimization: A block-level approach,” IEEE Transactions on Wireless
Communications, vol. 22, no. 6, pp. 3982–3996, 2023.
[145] S. M. Kay, Fundamentals of statistical signal processing: estimation
theory. Prentice-Hall, Inc., 1993.
[146] M. J. Bayarri and J. O. Berger, “The interplay of Bayesian and
frequentist analysis,” Stat. Sci., vol. 19, no. 1, pp. 58–80, 2004.
[147] I. Valero-Toranzo, S. Zozor, and J.-M. Brossier, “Generalization of the
de Bruijn Identity to general phi-entropies and phi-Fisher informa-
tions,” IEEE Trans. Inf. Theory, vol. 64, no. 10, pp. 6743–6758, 2017.
[148] A. J. Stam, “Some inequalities satisfied by the quantities of information
of Fisher and Shannon,” Inf. Contr., vol. 2, no. 2, pp. 101–112, 1959.
[149] F. Dong, F. Liu, S. Lu, and Y. Xiong, “Rethinking estimation rate
for wireless sensing: A rate-distortion perspective,” IEEE Trans. Veh.
Technol.,doi=10.1109/TVT.2023.3298005, pp. 1–6, 2023.
[150] J. Zhang, Y. Huang, Y. Zhou, and X. You, “Beam alignment and
tracking for millimeter wave communications via bandit learning,”
IEEE Trans. Commun., vol. 68, no. 9, pp. 5519–5533, 2020.
26
[151] D. Zhang, A. Li, M. Shirvanimoghaddam, P. Cheng, Y. Li, and
B. Vucetic, “Codebook-based training beam sequence design for
millimeter-wave tracking systems,” IEEE Trans. Wireless Commun.,
vol. 18, no. 11, pp. 5333–5349, 2019.
[152] S. Lu, F. Liu, F. Dong, Y. Xiong, J. Xu, Y.-F. Liu, and S. Jin,
“Random ISAC signals deserve dedicated precoding,” arXiv preprint
arXiv:2311.01822, 2023.
[153] Y. Xiong, F. Liu, K. Wan, W. Yuan, Y. Cui, and G. Caire, “From
torch to projector: Fundamental tradeoff of integrated sensing and
communications,” arXiv preprint arXiv:2310.09749, 2023.
[154] G. Cena, S. Scanzio, A. Valenzano, and C. Zunino, “Implementation
and evaluation of the reference broadcast infrastructure synchronization
protocol,” IEEE Trans. Ind. Inform., vol. 11, no. 3, pp. 801–811, 2015.
[155] M. Lévesque and D. Tipper, “A survey of clock synchronization over
packet-switched networks,” IEEE Commun. Surveys Tuts., vol. 18,
no. 4, pp. 2926–2947, 2016.
[156] Y. Liu, Z. Wang, J. Xu, C. Ouyang, X. Mu, and R. Schober, “Near-
field communications: A tutorial review,” IEEE Open J. Commun. Soc.,
vol. 4, pp. 1999–2049, 2023.
[157] H. Chen, M. F. Keskin, A. Sakhnini, N. Decarli, S. Pollin, D. Dar-
dari, and H. Wymeersch, “6G localization and sensing in the near
field: Fundamentals, opportunities, and challenges,” arXiv preprint
arXiv:2308.15799, 2023.
[158] J. An, C. Yuen, L. Dai, M. Di Renzo, M. Debbah, and L. Hanzo,
“Toward beamfocusing-aided near-field communications: Research ad-
vances, potential, and challenges,” arXiv preprint arXiv:2309.09242,
2023.
[159] Z. Wu and L. Dai, “Multiple access for near-field communications:
SDMA or LDMA?” IEEE J. Sel. Areas Commun., vol. 41, no. 6, pp.
1918–1935, 2023.
[160] Z. Wang, X. Mu, and Y. Liu, “Near-field integrated sensing and
communications,” IEEE Commun. Lett., vol. 27, no. 8, pp. 2048–2052,
2023.
[161] S. Yang, W. Lyu, Z. Zhang, and C. Yuen, “Enhancing near-field sensing
and communications with sparse arrays: Potentials, challenges, and
emerging trends,” arXiv preprint arXiv:2309.08681, 2023.
[162] F. Liu, Y.-F. Liu, A. Li, C. Masouros, and Y. C. Eldar, “Cramér-Rao
bound optimization for joint radar-communication beamforming,” IEEE
Trans. Signal Process., vol. 70, pp. 240–253, 2021.
[163] S. H. Dokhanchi, B. S. Mysore, K. V. Mishra, and B. Ottersten, “A
mmWave automotive joint radar-communications system,” IEEE Trans.
Aerosp. Electron. Syst., vol. 55, no. 3, pp. 1241–1260, 2019.
[164] B. Donnet and I. Longstaff, “Combining MIMO radar with OFDM
communications,” in Proc. IEEE European Radar Conf. IEEE, 2006,
pp. 37–40.
[165] P. Kumari, D. H. Nguyen, and R. W. Heath, “Performance trade-off in
an adaptive IEEE 802.11 ad waveform design for a joint automotive
radar and communication system,” in Proc. IEEE Int. Conf. Acoustic
Speech Signal Processing. IEEE, 2017, pp. 4281–4285.
[166] M. Nowak, M. Wicks, Z. Zhang, and Z. Wu, “Co-designed radar-
communication using linear frequency modulation waveform,” IEEE
Aerosp. Electron. Syst. Mag., vol. 31, no. 10, pp. 28–35, 2016.
[167] L. Zhipeng, C. Xingbo, W. Xiaomo, S. Xu, and F. Yuan, “Communica-
tion analysis of integrated waveform based on LFM and MSK,” Proc.
IET Int. Radar Conf, pp. 1–5, 2015.
[168] C. Xu, N. Ishikawa, R. Rajashekar, S. Sugiura, R. G. Maunder,
Z. Wang, L.-L. Yang, and L. Hanzo, “Sixty years of coherent versus
non-coherent tradeoffs and the road from 5G to wireless futures,” IEEE
Access, vol. 7, pp. 178 246–178 299, 2019.
[169] M. A. Richards, Fundamentals of radar signal processing. McGraw-
Hill Education, 2014.
[170] H. L. Van Trees, Optimum array processing: Part IV of detection,
estimation, and modulation theory. John Wiley & Sons, 2002.
[171] C.-L. Liu and P. Vaidyanathan, “Cramér–rao bounds for coprime and
other sparse arrays, which find more sources than sensors,” Digital
Signal Processing, vol. 61, pp. 43–61, 2017.
[172] P. P. Vaidyanathan and P. Pal, “Sparse sensing with co-prime samplers
and arrays,” IEEE Transactions on Signal Processing, vol. 59, no. 2,
pp. 573–586, 2011.
[173] G. Bienvenu and L. Kopp, “Optimality of high resolution array pro-
cessing using the eigensystem approach,” IEEE Trans. Acoust., Speech,
Signal Process., vol. 31, no. 5, pp. 1235–1248, 1983.
[174] R. Roy and T. Kailath, “Esprit-estimation of signal parameters via
rotational invariance techniques,” IEEE Transactions on acoustics,
speech, and signal processing, vol. 37, no. 7, pp. 984–995, 1989.
[175] K. Meng, Q. Wu, J. Xu, W. Chen, Z. Feng, R. Schober, and
A. L. Swindlehurst, “UAV-enabled integrated sensing and communi-
cation: Opportunities and challenges,” IEEE Wireless Commun.,doi =
10.1109/MWC.131.2200442, 2023.
[176] C. D’Andrea, A. Garcia-Rodriguez, G. Geraci, L. G. Giordano, and
S. Buzzi, “Analysis of UAV communications in cell-free massive
MIMO systems,” IEEE Open J. Commun. Soc., vol. 1, pp. 133–147,
2020.
[177] V. Kawadia and P. R. Kumar, “A cautionary perspective on cross-layer
design,” IEEE Wireless Commun., vol. 12, no. 1, pp. 3–11, 2005.
[178] Y. Zou, J. Zhu, X. Wang, and L. Hanzo, “A survey on wireless security:
Technical challenges, recent advances, and future trends,” Proc. IEEE,
vol. 104, no. 9, pp. 1727–1765, 2016.
[179] K. Zheng, H. Yang, Z. Ying, P. Wang, and L. Hanzo, “Vision-assisted
millimeter-wave beam management for next-generation wireless sys-
tems: Concepts, solutions, and open challenges,” IEEE Veh. Technol.
Mag., 2023.
[180] Z. Wang, J. A. Zhang, M. Xu, and J. Guo, “Single-target real-time
passive WiFi tracking,” IEEE Trans. Mobile Comput., 2022.
[181] K. Qian, C. Wu, Y. Zhang, G. Zhang, Z. Yang, and Y. Liu, “Widar2.0:
Passive human tracking with a single Wi-Fi link.” New York, NY,
USA: Association for Computing Machinery, 2018.
[182] Y. Zeng, D. Wu, J. Xiong, E. Yi, R. Gao, and D. Zhang, “Farsense:
Pushing the range limit of WiFi-based respiration sensing with csi ratio
of two antennas,” vol. 3, no. 3, 2019.
[183] J. Pegoraro, F. Meneghello, and M. Rossi, “Multiperson continuous
tracking and identification from mm-wave micro-Doppler signatures,”
IEEE Trans. Geosci. Remote Sens., vol. 59, no. 4, pp. 2994–3009, 2020.
[184] Y. Zheng, Y. Zhang, K. Qian, G. Zhang, Y. Liu, C. Wu, and Z. Yang,
“Zero-effort cross-domain gesture recognition with Wi-Fi,” in Proc.
17th Annu. Int. Conf. Mobile Syst., Appl., Services, 2019, pp. 313–
325.
[185] T. Zheng, Z. Chen, S. Ding, and J. Luo, “Enhancing RF sensing with
deep learning: A layered approach,” IEEE Commun. Mag., vol. 59,
no. 2, pp. 70–76, 2021.
[186] D. Wu, D. Zhang, C. Xu, H. Wang, and X. Li, “Device-free WiFi
human sensing: From pattern-based to model-based approaches,” IEEE
Commun. Mag., vol. 55, no. 10, pp. 91–97, 2017.
[187] B. Jokanovi´
c and M. Amin, “Fall detection using deep learning in
range-Doppler radars,” IEEE Trans. Aerosp. Electron. Syst., vol. 54,
no. 1, pp. 180–189, 2017.
[188] D. K. P. Tan, J. He, Y. Li, A. Bayesteh, Y. Chen, P. Zhu, and W. Tong,
“Integrated sensing and communication in 6G: Motivations, use cases,
requirements, challenges and future directions,” in Proc. 1st IEEE Int.
Online Symp. Joint Commun. Sens. (JC&S). IEEE, 2021, pp. 1–6.
[189] “6G–Connecting a cyber-physical world". Ericsson, Stockholm,
Sweden, White Paper GFTL-20:001402. [Online]. Avail-
able: https://www.ericsson.com/4927de/assets/local/reports-papers/
white-papers/6gâ ˘
A¸Sconnecting-a-cyber-physical-world.pdf
[190] Z. Behdad, Ö. T. Demir, K. W. Sung, E. Björnson, and C. Cavdar,
“Power allocation for joint communication and sensing in cell-free
massive MIMO,” in IEEE Global Commun. Conf. (GlobeCom). IEEE,
2022, pp. 4081–4086.
[191] Z. Wei, F. Liu, C. Masouros, N. Su, and A. P. Petropulu, “Toward multi-
functional 6G wireless networks: Integrating sensing, communication,
and security,” IEEE Commun. Mag., vol. 60, no. 4, pp. 65–71, 2022.
[192] N. Su, F. Liu, and C. Masouros, “Secure radar-communication systems
with malicious targets: Integrating radar, communications and jamming
functionalities,” IEEE Trans. Wireless Commun., vol. 20, no. 1, pp. 83–
95, 2020.
[193] W. Chen, L. Li, Z. Chen, B. Ning, G. Wang, and T. Quek, “An
ISAC-based beam alignment approach for enhancing Terahertz network
coverage,” arXiv:2212.01728, 2022.
[194] H. Zhang, H. Zhang, B. Di, and L. Song, “Holographic integrated sens-
ing and communications: Principles, technology, and implementation,”
IEEE Commun. Mag., 2023.
