Outlook

Physical-Biological World Cyber World

Virtual-X
Tactile
Inferencing

Sensing
Learning

Neural Edge Neural Center

Integrated Sensing and Communication

(ISAC) — From Concept to Practice
1 2 1 1 1 1 2 2
Alireza Bayesteh , Jia He , Yan Chen , Peiying Zhu , Jianglei Ma , Ahmed Wagdy Shaban , Ziming Yu , Yunhao Zhang ,
2 2
Zhi Zhou , Guangjian Wang
1
Ottawa Wireless Advanced System Competency Centre
2
Wireless Technology Lab

Abstract

6G will serve as a distributed neural network for the future Intelligence of Everything. Network Sensing and Native AI will
become two new usage scenarios in the era of connected intelligence. 6G will integrate sensing with communication in a
single system. Radio waves can be exploited to "see" the physical world and make a digital twin in the cyber world. This
article introduces the concept of integrated sensing and communication (ISAC) and typical use cases, and provides two case
studies of how to use 6G ISAC to improve localization accuracy and perform millimeter level imaging using future portable
devices. The research challenges to implementing ISAC in practice are discussed.

Keywords

integrated sensing and communication (ISAC), localization, THz imaging, sensing accuracy, sensing resolution, prototype

4 | Communications of HUAWEI RESEARCH | September 2022

1 Introduction localization, imaging, and environment reconstruction
obtained from sensing can improve communication

In 6G mobile communication systems, the use of higher performance — for example, more accurate beamforming,

frequency bands (from mmWave up to THz), wider faster beam failure recovery, and less overhead when

bandwidth, and massive antenna arrays will enable high- tracking the channel state information (CSI) [2–3]. This is

accuracy and high-resolution sensing, which can help called "sensing-assisted communication". Moreover, sensing

implement the integration of wireless signal sensing and is a "new channel" that observes, samples, and links the

communication (ISAC) in a single system for their mutual physical and biological world to the cyber world. Real-time

benefit. On the one hand, the entire communications sensing is therefore essential to make the concept of the

network can serve as a sensor. The radio signals transmitted digital twin — a true and real-time replica of the physical

and received by network elements and the radio wave world — a reality in the future.

transmissions, reflections, and scattering can be used

3GPP has initiated some preliminary study on use cases
to sense and better understand the physical world.
and potential ISAC requirements using the air interface
The capabilities to obtain range, velocity, and angle
of 5G advanced [1]. 6G ISAC systems will, however,
information from the radio signals can provide a broad
be further optimized, fully integrated, and will not be
range of new services, such as high accuracy localization,
constrained by the limitations of the current 5G system.
gesture capturing and activity recognition, passive object
The sensing use cases offered by future 6G ISAC systems
detection and tracking, as well as imaging and environment
will most likely include ultra-high accuracy localization and
reconstruction [1]. This is called "network as a sensor".
tracking, simultaneous imaging, mapping, and localization,
On the other hand, the capabilities of high-accuracy

Table 1 ISAC use cases as new services in 6G according to different categories

Use Case Category

High-Accuracy Simultaneous Imaging, Gesture and Activity
Localization and Mapping, and Localization Augmented Human Sense Recognition
Application Category Tracking

• Surgery with cooperative • Gesture-controlled

robots • Sensing glasses with ultra- • Tele-surgery smart operation theater
Vertical Industry high resolution imagery • Pollution and air quality
• Docking drone on a • In-cabin monitoring
moving vehicle • 3D road environment detection and contactless control
Intelligent healthcare mapping • Automatic flaw detection on
• Device/module • Contactless control for
Intelligent transportation • Warehouse robotics products
placement and intelligent manufacturing
Intelligent factory/ automation system • Intelligent crop monitoring
installation system
manufacturing • Crop production and crop for nutrients, water stress and • Gesture-based robots and
• Livestock movement
Smart agriculture physiology disease
and animal migration machinery control for
monitoring precision agriculture

• Collaborative robots for

Consumer • Imaging of water pipes
household chores • Virtual piano
behind walls
• Precise localization of • Touchless home appliances
Smart home & • Close-in scene and object • Calories count
small objects (tag or • Contactless control on
entertainment imaging • Contaminated ingredients
active objects) using intelligent screen
Smart mobile devices detection
mobile phones

• Crack detections in
• Drone as robotic waiter
Public Service buildings, bridges and • Gesture-based
• Hydrological monitoring • Wireless SLAM
man-made structures appliances for enhanced
{e.g., precipitation, water • Drone base stations swarm
Smart city • Fine particulate matter accessibility for seniors
flow/level} SAR imaging
Smart environment detection (PM10, PM2.5) and differently abled people
• Crowd management and • In-car sensing for driver and
Smart security and • Explosive detection and gas • Panic and terrifying emotion
emergency evacuation passenger monitoring
public safety sensing recognition
for major events
• Security scans on packages

September 2022 | Communications of HUAWEI RESEARCH | 5

Outlook
augmented human sensing, gesture and activity recognition, integrated into the 6G mobile communication system. This
as illustrated in Table 1 [1]. The use cases and performance capability will open up brand new services for 6G. These
requirements will be further discussed in Section 2. services are currently provided by various dedicated sensing
equipment, such as radar, light detection and ranging
The integration of sensing and communication functions
(LIDAR), and professional CT and MRI equipment.
can happen at three different levels, from loosely coupled to
fully integrated. At the lowest integration level, sensing and The ISAC capability will thus enable many new services that
communication capabilities can co-exist on hardware by mobile communication system operators can offer. These
sharing the spectrum, which is more efficient than dedicated include very high accuracy positioning, localization and
spectrum usage. Sensing can benefit from the economies of tracking, imaging for biomedical and security applications,
scale in the mobile communication network, where shared simultaneous localization and mapping to automatically
hardware will be cost effective and eases deployment and construct maps of complex indoor or outdoor environments,
maintenance issues. The second level of integration calls pollution or natural disaster monitoring, gesture and
for the integration of waveform and signal processing activity recognition, flaw and material detection and many
where the time, frequency, and spatial domain processing other services. These services will in turn enable application
techniques have a common objective and can be combined scenarios in all kinds of business for future consumers and
to serve both sensing and communication functions. A vertical industries. The potential new services that could be
fully integrated system with cross-layer, cross-module, and supported by future ISAC systems are listed in Table 1. In
cross-node information sharing is expected to significantly the table, the use cases are categorized into four functional
enhance the mutual performance of both sensing and categories across different applications/industries (vertical
communication, as well as reduce the overall cost, size and industry, consumer and public services):
power consumption of the network system.

· High-accuracy localization and tracking

In addition to the wider spectrum and the larger number
of antennas, the sensing functionality and performance will · Simultaneous imaging, mapping and localization
be further enabled by other technology innovations such as
the larger scale of cooperation between base stations and · Augmented human sensing
user equipment (UE), joint design of communication and
· Gesture and activity recognition
sensing waveforms, advanced techniques for interference
cancellation, and the native AI capability to better deal with
It is also worth mentioning that, in addition to the preceding
the sensed data.
services, sensing can also be used to assist communications

Next, we will discuss typical ISAC use cases and then and positioning, more details of which can be found in

elaborate on examples of ISAC application in enhanced Section 5.4.

localization and millimeter level resolution. Design
challenges will be discussed thereafter, followed by the 2.2 High-Accuracy Localization and
conclusions. Tracking
Low-latency high-accuracy localization and tracking enable
2 ISAC Use Cases meaningful association between cyber information and
the locations of physical entities in multiple scenarios from

2.1 Overview factories to warehouses, hospitals to retail shops, and

agriculture to mining.

Wireless sensing has long been a separate technology

The 6G network will provide services for both device-
developed in parallel with the mobile communication
based and device-free object localization. For 6G device-
systems. Positioning is the only sensing service that mobile
based localization, the target is a connected device in the
communication systems (until 5G) could offer. General
network, and the location information is derived from
sensing rather than positioning will become a new function
the reference signals or measurement feedback from the

6 | Communications of HUAWEI RESEARCH | September 2022

device. Localization for 6G device-free objects, on the other 2.3 Imaging, Mapping, and Environment
hand, does not need the object to be a connected device in
Reconstruction
the network. The estimation of delay, Doppler, and angle
In simultaneous imaging, mapping, and localization, the
spectrum information (corresponding to the distance,
sensing capabilities from these three perspectives are
velocity, and angle of the objects) are obtained from the
mutually enhanced. Particularly, the imaging function is
scattered and reflected wireless signals either through
used to capture the images of the surrounding environment,
monostatic sensing (receiver is the same as transmitter)
and the localization function is used to obtain the locations
or bistatic sensing (receiver is another node or device in
of surrounding objects. These images and/or locations are
the network). By processing these wireless signals further,
then used by the mapping function to construct a map.
the locations, orientations, velocity, and other geometric
The mapping function helps the localization function
information of the objects in a physical 3D space can be
improve the inference of locations. ISAC will leverage on
extracted. With higher bandwidth and increased antenna
advanced algorithms, edge computing, and AI to produce
aperture, the 6G ISAC system can have strong capabilities to
super-resolution and highly recognizable images and maps
separate multipaths, through which better localization and
in which the vast network of objects, including vehicles
tracking performance can be achieved, and the localization
and base stations, act as sensors to provide a remarkably
accuracy for outdoor use cases can be up to the centimeter
extended imaging area. Moreover, performance will
level.
significantly improve due to the ease of fusing results that
Having high-accuracy relative localization is important are shared with cloud-based services across the entire
when two or more entities exist and they are approaching network.
one another, or the entities have coordinated moving
6G-based super-resolution and high accuracy sensing
direction and speed. In automatic warehousing applications,
applications open up a range of possibilities in 3D indoor
centimeter-level accuracy enables device-level placement,
imaging and mapping, which in turn enable various
and the near-millimeter-level accuracy can further enable
applications, such as scene reconstruction, spatial
module-level installation and placement in tight spaces,
localization, and navigation for indoor scenarios, and help
allowing for efficient storage of components that have a
provide the most up-to-date knowledge of an environment
small form factor. Relative localization is necessary as a
for networks and devices. The accurate mapping information
viable alternative for close-in maneuvering owing to the
can then be applied to determine the multipath reflection
fact that complexity, physical limitations, and external
points. Owing to the fact that scattered signals bounce
infrastructure are mission critical for each robot to
multiple times where the LOS surfaces act as mirrors,
accurately determine its location in relation to a common
compensated images of NLOS objects can be reconstructed
datum. An example will be a drone docking onto a moving
by applying mirroring techniques. Once the environment
vehicle with an extremely small margin for landing, due to
is reconstructed, the next step will be the localization and
the limited area of the moving vehicle's cargo platform.
imaging of the NLOS targets. Target locations can then
Future ISAC systems that are empowered by native AI be detected with good accuracy when prior information
can provide semantic localization capability with context regarding the scene's geometry is known.
awareness. To support future smart home/shopping mall/
In an outdoor imaging and mapping scenario involving a
restaurant/hotel, and automatic factory applications,
mobile vehicle, its sensors usually have a restricted view and
objects and parts need to have dispatchable localization
limited coverage due to weather, obstacles, and the sensors'
information such as shelf level, seat number, table number,
power control. That said, nearby stationary base stations
etc. In a restaurant, robotic waiters, which have semantic
may have a greater field of view, longer sensing distance,
localization capability, can accurately deliver food to
and higher resolution because they collect and use their
guests and even go a step further and set different level of
own sensing data or sensing data of UE. Therefore, mobile
protections according to different task specifics, e.g., fragile
vehicles can achieve higher levels of autonomy by utilizing
and rigid objects can be treated with different levels of
the maps reconstructed by the base stations to determine
location and velocity accuracy during transportation.

September 2022 | Communications of HUAWEI RESEARCH | 7

Outlook
their next move. Moreover, the sensing resolution and changes beneath the skin, behind occlusion, or in darkness,
accuracy performance will significantly improve due to the poses different requirements. 6G radio wave (up to THz)
fusion of imaging results across the network. The densely based sensing can achieve the NLOS imaging ability, where
distributed base stations in an urban area and ISAC make technology for detecting hidden objects can be equipped on
environmental reconstruction and 3D localization possible, portable devices that have a powerful imaging capability. As
which in turn form the virtual city. The reconstructed map such, mobile phones can be used to detect pipelines behind
used for smart traffic control scenarios, such as traffic flow walls or perform security scans on packages by utilizing
monitoring, queue detection, and accident detection, are an the penetration characteristics of electromagnetic waves.
important use case in the dynamic virtual city. Moreover, 6G ISAC can enable atraumatic medical detection
which plays an important role in eHealth procedures such as
diagnosis, monitoring, and treatments. It provides ultra-high
2.4 Augmented Human Senses
reliability and accuracy and does not harm human bodies.

Technology progress makes augmented human sensing a Spectrogram recognition is another interesting application
reality. Augmented human sensing aims to provide a safe, that could be supported by a 6G ISAC system. It can
high-precision, low-power, sensing and imaging capability identify targets through spectrogram sensing of their
that exceeds human abilities, by means of a portable electromagnetic or photonic characteristics. This includes
terminal (e.g., 6G-enabled mobile phones, wearables, or the analysis of absorption, reflectivity, and permittivity
medical equipment implanted beneath human skin), to parameters, which helps distinguish the type and quality
sense the surrounding environments. With the help of of materials. Pollution and product quality management
scientific and technological advancement, augmented are some of the prospective applications of this technology.
human sensing can be achieved to facilitate information Spectrogram recognition can also be used in food sensing
collect ion and integrate the maximum number of applications to detect the food type and ingredients
environmental messages into the 6G network. through the transmission and reflection of THz signals. This
technology will help identify different types of food, calorie
In the 6G network, high-resolution imaging and detection
content, presence of contaminated ingredients, etc.
sensing techniques will open the door for numerous
applications, such as remote surgery, cancer diagnostics,
detection of slits on products, and sink water-leakage 2.5 Posture and Gesture Recognition
detection. A surgeon may be able to conduct surgery at
a different location through the help of an ultra-high- Device-free gesture and posture recognition using machine
resolution imaging monitor system and remote operation learning is the key to promoting human–computer
platform system. In addition, intelligent factories will interfaces that allow users to convey commands and
leverage these superior sensing solutions to implement conveniently interact with devices through body postures,
contactless ultra-high-precision detection, tracking, hand gestures, etc. In 6G system, the higher-frequency band
and qualit y control, where millimeter-level radial- will enable higher resolution and accuracy to capture finer
range resolution and ultra-high cross-range resolution postures and gestures, and the detection of motion activities
based on higher bandwidth and increased antenna array (resulting in Doppler shifts) will be more sensitive in the
aperture, respectively, are required. 6G communication higher-frequency band. Furthermore, the massive antenna
technologies, with high THz frequency and corresponding arrays allow for recognition with significantly improved
short wavelength that is less than 1 mm can increase spatial resolution and accuracy. Another important benefit
the bandwidth and decrease the array size, so that these of gesture and posture sensing by 6G is the fact that there is
augmented human sensing functions can be integrated or no risk of personal privacy information being compromised,
installed in portable devices. as is the case with cameras now, which makes it ideal for
many scenarios, especially smart home scenarios. In a future
While ultra-high-resolution scenarios require higher
gesture and posture recognition system that utilizes the
bandwidth and increased antenna aperture, another
densely distributed 6G network, devices will be collectively
application of "seeing beyond the eye" that can sense the

8 | Communications of HUAWEI RESEARCH | September 2022

used to sense the surrounding environments, and sensing indicators (KPIs) are introduced for sensing capability and
data association and fusion at an extended range will they are listed in Table 2.
significantly improve the overall recognition performance.
Table 3 presents the relevant key performance indicators
There will be advanced gesture and posture recognition along with the requirements that must be met in order
features in smart hospitals in the foreseeable future. The to realize the important use cases discussed in the earlier
medical rehabilitation system in future smart hospitals will sections.
enable the automatic supervision of patients. This ensures
that their gestures and movements during physiotherapy
3 ISAC for Centimeter-Level Positioning
conform to the standard requirements of rehabilitation
exercises. There will be prompt alerts on incorrect
movements or gestures, significantly improving patients'
3.1 Background, Motivation and
rehabilitation. In addition, an alarm alerting the hospital's
High-Level Scheme
control center will be generated if a patient falls during an
6G requires solutions for sub-centimeter level positioning
exercise, or if a suspicious person is detected intruding into
techniques for various future applications and use cases.
a restricted area.
This level of accuracy for positioning requires much more
The future smart home will be equipped with an advanced detailed knowledge of the radio signal propagation
hand gesture capturing and recognition system where environment where sensing comes into play. By learning the
it allows a hand's 3D position, rotation, and gesture to environment RF map and the way the transmitted waveform
be tracked. Thus, by simply waving our hands and other is manipulated by it, the UE position can be obtained as
gestures, many household appliances such as smart light, a function of the measurement parameters. This way, the
smart TV, etc., can be remotely controlled. Looking ahead, multipath nature of the propagation channel will be helpful
more complicated functionalities can be realized by the [4]. Moving to higher frequencies can further facilitate such
advanced hand gesture capturing and recognition function sensing-assisted positioning because the channel becomes
in the 6G network, such as playing a virtual piano in the sparser, and hence, characterizing the mapping between
air, in order to provide a completely immersive experience UE position and its propagation channel takes less effort.
anywhere, anytime. Without doubt, this futuristic concept In a reflection-dominant environment (which is the case
would open up a range of possibilities for many more in higher frequencies), one such mapping can be obtained
innovative applications related to high-accuracy finger by decomposing the multipath channel as multiple LOS
motion detection and tracking. channels coming from multiple anchors. Those anchors are
obtained by mirroring the transmission point (TP) over the
surface of the corresponding reflector for each path. Those
2.6 Key Performance Indicators
virtual anchors are referred to as virtual TPs or vTPs.

Within the ISAC context, several new key performance

Table 2 Sensing key performance indicators and their descriptions

Key Performance Indicator Description

Range and field of view limits within which objects can be
Coverage
detected by the system.
Difference between the sensed and real values in range,
Accuracy
angle, velocity, etc.
Resolution Separation between multiple objects in range, angle, velocity, etc.
Detection/False alarm probabilities Probabilities that an object will be detected when one is present/not present.

Availability Percentage of time for which a system is able to provide the sensing
service according to requirements.
Refresh rate Rate at which positioning/localization data is refreshed.

September 2022 | Communications of HUAWEI RESEARCH | 9

Outlook

· The channels between vTPs and the UE are LOS, which

means that there is no NLOS bias.

Hence, it can solve the two limiting problems of NR

positioning (i.e., synchronization error and NLOS error).

However, implementing such technology in a real cellular

(a) the real reflectors map (b) the extracted geometrical representation system is fraught with various challenges and the goal of
to the visible reflectors this section is to provide solutions for these challenges and
Figure 1 Mapping the objects/reflectors of the environment to pave the way for utilizing sensing-assisted positioning in
virtual anchors, i.e., mapping multipath components to vTPs
future 6G networks.

The advantage of such characterization is two-fold:

· Potentially large number of vTPs: In the initial stage of
environment sensing, the reflections of the TP location
· The vTPs are totally synchronous with the actual TP.
with respect to all objects in the map are obtained. The
This solves one of the prominent problems of current
issue is that number of vTPs grows linearly with the
positioning technologies, which rely on multiple TPs that
number of reflection planes and grows exponentially
are not synchronous.

Table 3 ISAC use cases along with key performance indicators and requirements

Use Case Category Coverage Resolution Accuracy Probability Availability Refresh Rate
High-accuracy localization and tracking
Module installation
10 m - 1 mm - 99.99% < 100 ms
and placement
Docking drone on a
50 m - 1 cm - 99.99% < 10 ms
moving platform
Robot/Drone as
50 m - 1 cm - 99.9% < 100 ms
waiter
Simultaneous imaging, mapping, and localization
SLAM 50 m 5 cm 1 cm - 99.9% < 10 ms
Indoor NLOS
100 m 5 cm 1 cm - 99.9% < 10 ms
localization
Urban environment
reconstruction 100-200 m 0.5 m 0.1 m - 99% < 1s
(virtual city)
Augmented human sensing

Remote surgery and -

2m 1 mm < 0.5 mm 99.9999% < 1 ms
medical diagnostics

Security scans on
packages via mobile 0.5 m 1-2 mm 0.5 mm - 99% < 100 ms
devices
Spectrogram
recognition for 0.5 m 1 mm 0.5 mm - 99% < 100 ms
calories
Posture and gesture recognition
Medical
rehabilitation activity 10 m 1 cm 0.5 cm - 99.9% < 1s
recognition
Virtual piano
anywhere, 10 m 0.5 mm 0.1 cm - 99% < 1 ms
anytime

10 | Communications of HUAWEI RESEARCH | September 2022

 with the number of allowed bounces. This is, in particular, (which is also referred to as pose estimation ) can be
problematic in outdoor scenarios. obtained.

· Association of the multipath measurements to vTPs: The proposed SAPE scheme is in contrast to most SLAM
Another major challenge in implementing the multipath techniques where all the localization burden/processing is at
assisted positioning techniques is that a UE has no idea the UE side.
how to match each measurement parameter vector
(consisting of angles, delay and Doppler) to a vTP and
3.2 Detailed Proposed SAPE Scheme
this can potentially produce a large positioning error. In
general, the matching between the observations and the
3.2.1 First and Second Step Sensing
visible vTPs is a combinatorial problem with exponential
complexity.
In the initial environment sensing stage (first step), the
TP senses the entire communication space by using a
To solve the above issues, we introduce our proposed
sensing-assisted position estimation (SAPE) scheme. The relatively wide beam or small bandwidth in order to

basic concept of SAPE is to utilize the high resolution generate a coarse RF map to the main reflectors/objects of

c a p a b i l i t i e s o f t h e m a s s i v e M I M O a n d m m Wa v e the communication space. The main goal of this stage is to

technologies in space, angular, and time domains in order to identify the potential reflectors and map them to vTPs. A
increase the resolvability of the multipath components and static RF map is then available at the TP through this first
exploit the environment RF map to identify the potential stage sensing, based on which the location and orientation
reflectors of such multipath components, thereby sensing of the static objects or reflectors can be pre-calculated.
the environment while localizing UEs with high resolution
and accuracy. This allows for exploiting the multipath In the second step of sensing, which is the stage of
components (including NLOS) to enhance the accuracy environment sensing update or dedicated sensing, the TP
of the position, velocity, and orientation information by starts targeted sensing based on the obtained RF map
providing the association between the observations reported and coarse UE location. Particularly, the TP senses certain
from the UEs and the prior information corresponding to subspaces, based on the coarse UE location and location
the main environment reflectors. Efficient association and
of the main reflectors, and processes the reflected signals
accurate mapping need careful design of specific sensing
to obtain finer sensing information of those reflectors.
signals, novel transmission and reception signal processing
Simultaneously, the UE also performs measurements on the
techniques, and their corresponding measurement and
sensing signal to obtain information including multipath
signaling mechanisms.
identification, range, Doppler, angular and orientation

In particular, the proposed SAPE scheme comprises two measurements in order to obtain the UE position. Therefore,

main steps: the second step sensing refines the pre- calculated
information obtained in the first phase and thus supports
1. First step sensing or environment sensing, in which quasi-static environment. In addition, this step can correct
the network (TP) tries to find/update the location of the potentially large location errors of the vTP locations
the main reflectors of the environment and obtains the obtained from the first step sensing. The impact of vTP
subspace for the next step sensing; location errors will be studied in Section 4.

2. Second step sensing, in which the TP sends tailored,

specific sensing signals in the subspaces obtained in
the first step sensing in order to enhance the multipath
re s o l va b i l i t y a n d a s s o c i a t i o n . T h e U E p e r fo r m s
measurements over the received sensing signals, and by
proper mapping of the measurements to the vTPs, the
UE position, as well as velocity vector and orientation

September 2022 | Communications of HUAWEI RESEARCH | 11

Outlook

3.2.2 Multipath Parameter Estimation be classified into four categories, namely, spectra-based [5–
6], subspace-based [7–8], compressive sensing-based (sparse
The problem involves estimating the parameters of the signal recovery/reconstruction) and maximum likelihood-
dominant J multipath components of the received signal based (ML) approaches [9–11]. A high-level comparison
at the UE per transmitted beam. The parameters to between the four categories is provided in Table 4.
be estimated are the delay τ j , Doppler νj , channel path
coefficients βj and angle of arrivals ϑjr , φjr , i.e., elevation and Among these algorithms, space alternating generalized
azimuth angles of the j - th path. All these parameters are expectation (SAGE) maximization is known to be a
collected into one vector denoted by θj , for all j . Given the reasonable approach for reducing the computational
transmitted signal s m (t ) over the m th beam, the received complexit y, and the slow convergence rate of the
signal is given by: maximization step in the EM algorithm is improved by
employing the alternating optimization concept over the
(1)
estimated parameters for each path. Similar to EM, the
where X j (t ; θj ) is the received signal of the j - th path. SAGE consists of two consecutive and iterative steps, i.e.,
We note here that X j (t ; θj ) subsumes the effect of the expectation and maximization. In the expectation step,
beamformer at the transmitter and the additive white the unobservable data (in our case they are the multipath
Gaussian noise at the receiver. We note also that the components θj ) is estimated based on the observation
Λ
TX beamforming, during the second step sensing stage, of the incomplete data and a previous estimate θ(i) of
makes Y (m )(t ) sparse, i.e., J (m ) is small. The joint estimation the parameters vector θ. In the maximization step, the
of these space-time-frequency parameters results in parameters vector of j- th path θj is re-estimated iteratively
complex noncovex optimization problems. Moreover, the by alternatingly optimizing the components of θj , i.e., delay,
entanglement of the paths' parameters limits the accuracy Doppler, channel coefficients and angle of arrivals. In this
and reduces the resolution of the estimated parameters, way, the multi-parameter optimization problem is reduced
thereby impeding their resolvability. In addition, the high to multiple single-parameter optimization problems.
dimensionality in space, time, and frequency, and real-
time processing requirements necessitate taking the
3.2.3 Multipath Parameter Association
computational complexity of the parameter estimation
algorithm into consideration. Thus, we are looking for a low-
The multipath parameter association problem requires
complexity super-resolution channel parameters estimator.
finding a way to associate the estimated parameters,
The literature on the multipath parameters estimation can
i.e., delays and angles of arrival, of the different channel

Table 4 Comprehensive comparison between channel parameters estimation

Category Description Pros Cons

ML-based Maximum likelihood estimator The optimal solution Prohibitive complexity

Importance sampling ML High complexity

ML-based Superior performance
Slow convergence

High complexity
ML-based Expectation maximization Superior performance
Slow convergence

Medium complexity
ML-based SAGE Super resolution
Fast convergence
Sparse signal Competitive
OMP and its variant Medium complexity
reconstruction performance

Sparse signal Based on convex relaxations such as l 1

Super resolution High complexity
reconstruction norm, nuclear norm, and atomic norm

Subspace-based MUSIC, ESPRIT, and their variants Medium resolution Medium complexity

Subspace-based FFT Low resolution Low complexity

12 | Communications of HUAWEI RESEARCH | September 2022

 Λ Λ
multipath components, Z = { θ1 ,..,θNobs }, where Nobs denotes Prior art adopts measuring the distance between the
Λ Λ
the number of observation and Z is obtained by the UE for set of the measurements Z = {θ1 ,.., θNobs } and the set of
each measurement-beam pair, to their relevant visible vTPs the e x p e c t e d g ro u n d t r u t h va l u e s G = { g 1 , . . , g N v t p } ,
Λ
represented by the set of the ground truth values where θ1 is the i th vector that contains range and angles
G = {g 1,..,g Nvtp }, where N vtp denotes the number of visible of arrival of the measurement and g k is the distance
Λ
vTPs and G is obtained by the TP through the first and vector between the expected UE position p and the k- th
second step sensing, as shown in Figure 2. vTP obtained from previous estimations. However, this
technique has two main shortcomings. First, it requires
prior information about the UE's position which might not
be available in many scenarios. Second, it requires using
long training periods and measurements to iteratively
update the prior knowledge of the UE's position to
make the association algorithm converge. To cover these
shortcomings, we propose a new technique that mainly
exploits the differential/mutual distances between the
members of the two measurement sets. The key idea of
the proposed algorithm is to match the relative/differential
distances between the members of the measurements set to
the relative distances between a subset of visible/expected
vTPs as shown in Figure 3.
Figure 2 Illustration of measurements — vTP association problem
Based on Z Based on G

Extensive research has been conducted in order to alleviate

the association error and reduce the computational
complexity of the association algorithms. These works
can be categorized into two main lines of thoughts,
namely, soft-decision/probabilistic data association and
hard decision data association [12]. In the probabilistic
Figure 3 Illustration of relative/tdifferential distance (a) based on Z (b) based on G
approach [13–16], all the vTPs are assigned to a certain
measurement/observation with different probabilities, with Λ
This is mainly aimed at avoiding the dependency of G on p .
the probabilities indicating how likely a given measurement
This requires two different modifications on the two sets, Z and
is due to a particular vTP. This requires a proper selection
G . First, instead of directly using the N obs measurements, we
of the statistical model for assigning these probabilities. In
convert them into N obs hypothetical vTP locations relative
the hard decision data association approach [12, 17], each
to the origin point. We denote this set of hypothetical vTP
measurement/observation is associated only to one vTP.
locations by H z = {h 1,..,h Nobs }, where h i is the relative
The techniques within this approach can be divided into
location vector of the i- th hypothetical vTP. Based
two categories, namely, probabilistic-based hard decision
on these relative location vectors, we calculate the
algorithms and distance metric-based selection algorithms.
differential/mutual distances between these relative
In the former, the measurement is associated to the most
locations, i.e., dij = h i - hj∀i , j , i ≠ j . The set of the differential/
likely association event according to a certain probabilistic
mutual Euclidean distances between the hypothetical vTPs'
measure such as maximum likelihood or a posteriori; in N
locations is defined as D and has a size of ( obs) elements
2
the latter, the measurement is associated to the nearest
where its n- th element is denoted by d i j (n ). The second
association event according to a certain distance metric such
modification is to build the set of the differential/mutual
as the Mahalanobis distance [12, 17]. The main drawback ~
distances between the real vTPs' locations, i.e., D . Because
of this approach is that it depends heavily on the accurate
the set of hypothetical vTPs' locations, i.e., H z and the set
knowledge of the UE position as a prior.
of the actual vTPs locations, i.e., R usually have a different
N
cardinality, we divide the later set into ( N v tp ) subsets of size
obs

September 2022 | Communications of HUAWEI RESEARCH | 13

Outlook
Nobs , with each containing a different combination of vTPs' that with 3 observations, one gets an average association
locations. We denote these subsets by r i = {r i 1,...,r iNobs }, i ∈ error of 0.07 at SNR of 5 dB, i.e., just 7 out of 100 vTPs on
N
{1,...,( N v tp )}, where ri 1 is the location vector of first vTP in average will be associated wrongly. It is also noteworthy
obs

the i th subset. For each ri , we define the differential/mutual to mention that the average association error is different
~
distances between its members as d imn = ri m - r i n ,∀m ,n ,m ≠n . from the average position error. However, the former
The set of the mutual/differential between the members of affects the later. In other words, wrongly associating one
~i
ri is denoted by D . We measure the distance between the out of 10 vTPs might not produce significant position error
~i
sets D and D by: if the measurements for this vTP have a lower weight in

(2) calculating the position error.

100
~ Number of oberservations-3
where, with a slight abuse of notation, d (n) and d iD~ iπ (n ) Number of oberservations-5
~ j
are n- th elements of D and D iπ j , respectively, and πj is the j- th
~ -1
10
permutation of the elements of D i. . Using this new metric,
the association is given by:

Association Error
-2
(3) 10

πopt contains the indices of those entries of R that are

optimally assigned to D . 10
-3

3.3 Performance Evaluation of the 10

-4

Proposed SAPE Scheme

-10 -5 0 5 10 15 20 25
SNR (dB)

Figure 4 Average association error of the proposed association algorithm

when the number of visible vTPs is 3 and 5
3.3.1 Link-Level Evaluation
We further evaluate the performance of the proposed
In this subsection, we mainly evaluate the average association algorithm and the impact on the positioning
behaviour of the proposed association algorithm using error in a real multipath environment. Without generality,
statistically generated measurements. In the simulation, we we assume that the estimation error due to the channel
generate 8 uniformly distributed vTP positions. We further parameters estimation stage follows the CRLB. Figure 5
assume each received observation θ (can be range or angle) presents the CDF of the UE positioning error bound (PEB)
has Gaussian noise with standard deviations of due to the association scheme and compares it with the
√- -
c *CRLB(θ ), where CRLB(θ ) denotes the Cramer-Rao Lower case of ideal association, which shows the promising
Bound for mean square error estimate of θ and the constant performance of the proposed SAPE technology and its
factor c accounts for the non-ideal factors in the detection potential for 6G positioning.
which results in the gap between the real estimation and
1
the lower-bound (in the evaluation results, c = 6). We Realistic association
Idealistic association
calculate the association error by counting the number of
0.8
different indices of the associated vTPs from the observed
ones.
0.6
P (PEB < x)

As shown in Figure 4, the proposed association algorithm

provides a very good performance in the synthetic 0.4

scenario. In addition, better performance is observed with

more measurements (more vTPs) because more mutual/ 0.2

differential distances are used for association. We note that

the association error in Figure 4 represents the ratio of the 0
10-5 10-4 10-3 10-2 10-1 100 101
number of vTPs wrongly associated to the total number PE (m)
Figure 5 Overall positioning performance of the
of vTPs on average. For instance, in Figure 4, it is shown
proposed SAPE technology at the link level

14 | Communications of HUAWEI RESEARCH | September 2022

3.3.2 System-Level Evaluation UMI (outdoor)

In this section, we provide the performance of the proposed 1

Sensing-assisted positioning
SAPE scheme at the system level. Similar to the system-level 0.9 Baseline NR positioning
Baseline NR with perfect LOS identification and synch
communication performance evaluation, the key step in 0.8

such an evaluation is abstracting the PHY-level performance 0.7

at the network level, which is the so-called PHY abstraction. 0.6

P (PEB < x)
The rationale behind sensing PHY abstraction is to map 0.5

0.4
the system parameters in terms of SINR, bandwidth,
0.3
time duration, and antenna configuration to a sensing
0.2
performance (i.e., range, Doppler or angle mean square
0.1
errors). The proposed SAPE scheme is evaluated and
0 -4 -2 0 2
compared with baseline NR in terms of PEB, based on the 10 10 10 10 104
PEB (m)
proposed PHY abstraction methodology in two scenarios:
InH
Idealistic scenario : where the sensing is assumed to be
perfect. In this case, the evaluation is based on applying 1

the proposed PHY abstraction methodology in SLS and 0.9

evaluating the candidate schemes in two scenarios: indoor 0.8 Sensing-assisted positioning
Baseline NR positioning with 1 ns synch. error

hotspot (InH) and outdoor urban micro (UMI). Both 0.7 Baseline NR positioning with perfect TRP synch.

0.6
P (PEB < x)

scenarios are evaluated over mmWave bands and the

simulation parameters are given in Table 5. 0.5

0.4

0.3
Table 5 Parameters for SLS evaluation
0.2
Parameter Value
0.1
Bandwidth 80 MHz
Sensing time 14 symbols 0
-5 -4 -3 -2
10 10 10 10 10-1 100
Sub-carrier spacing 60 kHz PEB (m)
Number of subcarriers 1024 Figure 6 SLS results of the proposed SAPE vs. baseline NR in idealistic scenario

Indoor hotspot, 256 × 32 Based on the results, we can observe that under ideal
Deployment UMI 32 × 16 (outdoor only),
conditions (no RF impairments, no sensing error, no
20 RRUs and 200 UEs
diffraction), SAPE can achieve an order of magnitude
Channel model SCM (stochastic) better accuracy compared to NR. In addition, the NR
Carrier frequency 60 GHz
baseline cannot achieve good performance in any scenario,
Based on SLS using the
even under ideal conditions, due to NLOS bias and
Simulation methodology proposed sensing PHY
abstraction synchronization error between the TPs.
Non-idealities modeled Sensing error
Realistic scenario : assuming sensing error, the candidate
Synch. error between TRPs 0 (perfect synch.) or 1 ns
schemes are evaluated in outdoor UMI. The simulation
parameters are the ones given in Table 5. For modeling the
Based on these parameters, the simulation results are given sensing error, we assume the vTPs corresponding to each
in Figure 6. path/cluster are Gaussian-distributed with some variances
which are also modeled as random variables. In addition,
the vTP location variance for the LOS link is set to 0 as it
corresponds to the actual TP. Based on these parameters, the
simulation results are given in Figure 7.

September 2022 | Communications of HUAWEI RESEARCH | 15

Outlook
can provide high sensing resolution in addition to high
UMI (outdoor)
communication throughput, the integration of THz sensing
Sensing-assisted positioning
Baseline NR with perfect LOS identification and synch
and communication has become an attractive and active
Baseline NR positioning
Sensing-assisted positioning, realistic sensing research area.

(4)

(5)

The application of ISAC-THz design is expected to provide

many opportunities for brand new services especially on
future mobile devices or even wearables as illustrated
Figure 7 SLS results of the proposed SAPE vs. baseline NR in the realistic scenario in Figure 8. In addition to the localization and imaging
applications, molecular spectrogram analysis is another
Based on the results, we can observe that SAPE can achieve interesting application area that could be enabled by ISAC-
an order of magnitude better accuracy (cm-level accuracy) THz, as discussed in Section 2.
when compared with NR, even with the sensing error.
Table 6 Maximum contiguous bandwidth in the range of 100-450 GHz and the
corresponding range resolution

4 ISAC for Millimeter-Level Imaging Freq. (GHz) Contiguous Range Resolution

at the THz Band Bandwidth (GHz) (mm)
102–109.5 7.5 20
THz lies between the mmWave and infrared frequencies, 141–148.5 7.5 20
and thus has millimeter-level and even sub-millimeter- 151.5–164 12.5 12
level wavelength, making the ISAC system at the THz band
167–174.8 7.8 19
(ISAC-THz) particularly suitable for high resolution sensing
191.8–200 8.2 18
applications such as millimeter-level resolution 3D imaging.
209–226 17 8.8
Like the other lower frequency radio waves, THz can
252–275 23 6.5
penetrate some obstacles, achieving high-precision sensing
in all weather and lighting conditions. 275–296 21 7.1
306–313 7 21
Recent developments in semiconductor technology have 318–333 15 10
bridged the "THz band gap" and made the hardware
356–450 94 1.6
feasible at the terminal side. ISAC-THz based portable
devices will thus open the door for numerous new sensing Table 7 Aperture size and corresponding cross-range resolution at 140 GHz
applications such as augmented human sensing with
λ = 2.1 mm, r = 30 cm
very high resolution. Table 6 shows the allocated mobile
Aperture Size (cm) Cross-Range Resolution
frequency bands with a contiguous bandwidth greater
(mm)
than 5 GHz. The ultra-wide bandwidth in THz will also
1 32
enable Terabits/second data rate transmission, especially
5 6.4
in short-range communications. The corresponding range
10 3.2
resolution (from equation 4) based on Heisenberg's
20 1.6
Uncertainty Principle is also provided in this table. Under
the assumption of synthesized aperture, the cross-range
resolution is provided in Table 7 based on equation 5
where λ is the wavelength, D is the aperture size, and r is
the distance between transceiver and target. Because THz

16 | Communications of HUAWEI RESEARCH | September 2022

 Time division & Code division
TX = 4 TX = 4
THz THz THz
Sensing Communication Spectrogram
λ

λ
λ
-
2
Localization and Access Pollution
tracking communication detection

RX = 16

RX = 64
RX = 64
RX = 16
Simultaneous High speed
Quality Element spacing = λ
λ
imaging, mapping, backhaul Displaced phase centerspacing = -
assurance 2
and localization communication

High speed
Augmented Basic scientific
chip/board
human sense research
communication Figure 9 MIMO virtual aperture

Gesture and Space Space

activity recognition communication observation To implement the overall solution of a virtual aperture, the
Figure 8 THz application in sensing and communication following three requirements need to be satisfied in the
hardware design:
In this section, we elaborate on our ISAC prototype of THz
imaging on portable devices that achieves millimeter-level
· Multiple transceiver (TRX) chains to support the MIMO
resolution. A robot arm equipped with ISAC-THz module antenna array structure as the first step for the overall
is used to represent a human arm holding a THz imaging virtual aperture.
camera. The prototype is built to operate at 140 GHz carrier
· Wide antenna pattern to cover the target scanning area
frequency with a bandwidth of 8 GHz.
in order to maintain the correlation among the reflected
samplings.
4.1 Hardware Architecture of the · Real-time position information of the device to perform
ISAC-THz Module coherent processing of the received signals.

From the THz imaging aspect, thousands of antenna The schematic of the prototype architecture is shown in
elements are required to create a large aperture for high Figure 10. The transmitter antenna array has 4 RF ports and
cross-range resolution. However, it is clear that physically the receiver antenna array has 16 RF ports, forming a 4T16R
packing thousands of antenna elements into the portable MIMO antenna array structure [3]. The per unit antenna
device is infeasible due to the size and power constraint radiation pattern is a wide beam design with a 3 dB beam
requirements of the device [18–19]. To solve this problem, width of 50° and gain of 7 dBi.
virtual aperture techniques are applied in the prototype
system [3]. In particular, the virtual MIMO antenna array
design in the hardware transceiver architecture using the
sparse sampling design in the scanning process is proposed
[3, 20–21].

First, a virtual MIMO antenna array structure is constructed

to form a virtual aperture that can achieve the same
performance with respect to its equivalent physical aperture
array as illustrated in Figure 9. Next, a sparse scanning
approach is applied, transforming the degree-of-freedom
in time and space into a larger virtual aperture, as shown
in Figure 9. The scanning performed by the robot arm thus
mimics a user holding a smartphone and imaging an object
with a zigzag scanning trajectory.

September 2022 | Communications of HUAWEI RESEARCH | 17

Outlook
To solve this challenge, we consider decomposing the
scanning trajectory on a two-dimensional (2D) plane into

DAC

DAC

DAC

DAC
Baseband several sets of linear scanning tracks along the horizontal
PLL
TX direction, where the sparseness of the sampling signals in
the vertical domain is then equivalent to the sparseness
between horizontal tracks, as illustrated in Figure 11. In
Transmitter antenna array this case, the reflected/echo information from the object
ADC
MEMS can be retrieved from these vertically sparse samplings via
ADC
RX
ADC
compressed sensing techniques [3].
ADC
As depicted in Figure 12a, the robotic arm scans at a speed

Receiver antenna array of 1 m/s with the scanning area set as 10 cm by 12

cm in the prototype. The longitudinal spacings of the scan
trajectories are controlled to simulate the sparsity in the
ADC

ADC
trajectories of the user's hand-held scanning behavior. The
RX
ADC target object to be imaged, as shown in Figure 12b, is put in
ADC a box with a cap on top of it. As we can see from Figure
12b, the smallest distance in the hallowed pattern is 3.5
Figure 10 Illustration of the architecture of ISAC prototype
mm, so the highest resolution of the imaging results can be
3.5 mm.
4 . 2 C o m p re s s e d S e n s i n g - b a s e d
Tomography Imaging Robot arm for scanning

A major challenge for the virtual aperture imaging

technique is the irregular scanning trajectory caused by ISAC-THz
the user moving the ISAC imaging module to perform THz prototype

scanning on an object. Assume a zigzag scanning routine is

used to image an object, as shown in Figure 11. The echo
samplings in the horizontal direction are continuous, i.e.,
the spatial spacing between sampling points is comparable
to the wavelength of the echo signal. However, continuous
sampling cannot be maintained in the vertical direction.
As a result, the echo samplings in the vertical direction are
sparse, which will cause high and non-uniform sidelobe Box with
target in
effects, giving rise to false artifacts, which may lead to
(a) Prototype setup for THz sensing where the ISAC-THz module is held by
imaging failure. a robot arm representing a human arm
Scanning trajectory
Vertical

= tomographic sections
Aperture plane 2D image slides 9m
24 m

Horizontal m
m

3. m
5
m
82 mm

Range

7 mm

Echo reflection vector

(= a sparse vector ) Target
(b) Target object in the box
Figure 11 Illustration of the sparse scanning approach and the tomographic
imaging techniques Figure 12 Setup of the ISAC-THz prototype

18 | Communications of HUAWEI RESEARCH | September 2022

The proof-of-concept THz imaging performances with
different sparsity configurations in the scanning patterns are
presented and compared in Figure 13. In each of the figures,
the 3D imaging results are shown on the left and the cross-
range profile perceived from top down is shown on the
right.

The non-sparse full aperture scanning in Figure 13a is an (c) Sparse scanning with 25% sparsity (most sparsity) and using the traditional
tomography approach [22]
ideal case, in which the vertical sampling is half wavelength
adjacent. This achieves the best PSLR and ISLR performance,
which is set as an upper bound performance reference.
Then, in order to simulate the sparsity in real free hand
scanning, we assume different sparsity configurations
in tests, from 50% (medium sparsity) to 25% (most
sparsity), where X % sparsity means that there are X %
of the full samplings remaining in the vertical direction.
(d) Sparse scanning with 25% sparsity (most sparsity) and using the compressed
With the collection of fewer samplings, stronger side- sensing based tomography approach

lobe interference occurs at the resulted aperture, resulting

Figure 13 Imaging results at different sparsity configurations
in worse imaging performance. From the comparison of
Figure 13c and Figure 13d, we see that when the sparsity
4.3 Multi-Channel Imaging
is too high, the traditional tomography algorithm is not
enough to recover the images. In this case, the compressed
The multi-channel imaging process can be treated as a
sensing based tomography approach showed its superior
time-domain coherent combination of electromagnetic
performance. signals from multiple receiving channels. Theoretically, n
receivers can reduce the sampling time to 1/n compared
with one receiver with the same imaging quality. Less
sampling time will reduce the difficulty of motion error
compensation, which in turn will improve the imaging
quality.

However, in multi-channel imaging, one major challenge

arises from the imbalance in gain and time delay of
(a) Non-sparse full aperture scanning (ideal case)
different receiver channels due to hardware imperfection.
The antenna mounting positional imperfection will introduce
the displaced phase center error, as shown in Figure 14.
Multi-channel amplitude and phase imbalance will lead
to azimuth ghosting, which will significantly degrade the
imaging quality. The amplitude imbalance can be easily
compensated by multichannel amplitude equalization
methods [23], while phase imbalance compensation needs
(b) Sparse scanning with 50% sparsity (medium sparsity)
auto-focusing algorithm such as gradient descent.

September 2022 | Communications of HUAWEI RESEARCH | 19

Outlook

TX
Displaced phase
center 1-16 RX 1-16

Figure 14 Illustration of the displaced phase center caused by multi-channel imaging

(a)

To validate the performance of multi-channel imaging,

we use the same testbed described in the last subsection
but a different target (resolution is similar, i.e., 3 mm) as
shown in Figure 15. In this case, we tried both the 2D target
shown in Figure 15a and the 3D target shown in Figure
15b, where the 3D imaging target was formed by placing
the two characters at different heights inside the box. With
the aforementioned benefit of multi-channel imaging, very
sparse sampling is needed for a good imaging quality. In the
prototype, only 12% sparsity is configured in the scanning
trajectory.
(b)
3 mm
Figure 16 Multi-channel imaging result of the 2D target
47 mm

10 mm Subsequently, the imaging results of the 3D target are

shown in Figure 17. The imaging result clearly depicts the
3 mm shape of the two characters and their relative distance in
the 3D space.
(a) 2D imaging target (b) 3D imaging target

Figure 15 Imaging targets used in multi-channel imaging

Figure 16 shows the imaging results of the 2D target.

The imaging results without multi-channel phase error
compensation are illustrated in Figure 16a where severe
ghosting on the final image that significantly degrades
the imaging quality can be prominently seen. Using the
geometric interpretation algorithm, the sidelobe due to the
multi-channel imbalance has been duly suppressed as can
be clearly seen in Figure 16b.

Figure 17 3D imaging result

20 | Communications of HUAWEI RESEARCH | September 2022

5 Major Challenges for Making · Typical scenarios and evaluation methodology
ISAC a Reality
The traditional indoor hotspot, urban micro, urban macro
are defined in the 3GPP 38.901 communication channel. The
5.1 Channel Modeling and Evaluation environment and the purpose of the application will deeply
Methodology affect the channel model parameters and even the channel
model generation approach. Therefore, the ISAC typical
In 6G, the channel model needs to be considered for both
scenario should be categorized and the typical use cases of
communication and sensing services. This brings significant
each category should be highlighted for further evaluation.
challenges to the channel modeling methodology. Until 5G,
because it has low computational complexity and is easily For a given evaluation use case, metrics to characterize the
standardizable, stochastic channel modeling methodology joint performance between communication and sensing
dominated the evaluation of wireless communications, are needed in order to optimize the performance trade-
and is used in many projects and standards such as 3GPP- off for both services simultaneously. To characterize
SCM, WINNER-I/II, COST2100, and MESTIS. It is adequate in the performance of both functions as well as mutual
evaluating the communication performance. However, there enhancement, scenarios and metrics need to be
is a doubt as to whether it still can meet the more diverse implemented into the system level simulations and the
requirements from different sensing applications. ISAC performance must be evaluated in a fully integrated
network.
One typical sensing channel is the echo channel, which
consists of the backscattering RCS characteristics from · Channel measurement and modeling methodology
the object and its surroundings. This type of propagation
channel brings new requirements for the phy sical Regarding ISAC channel modeling, a single channel
electromagnetic (EM) characteristic which are not supported modeling scheme may not meet the need to evaluate
in the current communication channel models. One typical all ISAC applications. Instead, stochastic, deterministic,
use case is the high resolution imaging application. This and even hybrid channel models must be considered. For
type of application requires the deterministic channel
instance, in the sensing-assisted beamforming use case, the
coherence of the antenna array aperture with the geometry
stochastic channel modeling could be adopted, whereas
information. This requirement is contradictory to the typical
for localization and tracking application, ray tracing could
stochastic channel modeling approach. Therefore, the
be considered as a strong candidate for channel modeling
traditional channel modeling methodologies deserve some
because the detailed contours for the object reflection/
rethinking and innovation.
scattering are not strictly required. On the other hand, for

Another major challenge is the evaluation performance imaging and recognition applications, there is a need to
m e t r i c s b a s e d o n t h e n e w s e n s i n g re q u i re m e n t s . consider EM algorithm when the size of the scatterers is
Conventionally, throughput, latency, and reliability are the close to the signal wavelength and therefore the interaction
main evaluation performance metrics for communication of the signal to the scatterers are strongly correlated with
systems. However, due to the different sensing applications, the EM characteristics.
there are new dimensions of evaluation metrics that need
to be considered, such as sensing resolution, accuracy,
detection probability, and update rate. So far, no KPIs have
5.2 Joint Waveform and Signal Processing
been proposed for the joint performance characterization
Design
and evaluation of both the communication and sensing
Most of the works on the joint design of sensing and
services. This implies that a new scenario-dependent
communications mainly focus on the joint waveform
evaluation methodology may need to be investigated.
design. The main challenge for the joint waveform design
is the contradicting KPIs for communications and sensing.
To address the challenges mentioned above, the following
In particular, the main target for communications is
research directions are proposed:
maximizing the spectral efficiency, whereas the optimum

September 2022 | Communications of HUAWEI RESEARCH | 21

Outlook
waveform design for sensing is focused on estimation signal. Another line of research is devoted to using single-
resolution and accuracy. Because CP-OFDM has been carrier waveform based on the code domain spreading of
proven to be a favorable option for communication, many joint radar and communication signals. For this class of
researchers have considered this waveform for sensing as waveforms, the radar performance has been shown to be
well. Although the introduction of cyclic prefix (CP) has affected by the auto-correlation of the sequences and the
been shown to degrade auto-correlation in the time domain long spreading codes result in good auto-correlation at
[24], a novel approach of frequency domain processing the expense of communication spectral efficiency [27]. In
[25] allows for efficient parameter estimation of CP-OFDM, addition, Doppler estimation requires more complicated
achieving the maximum processing gain. Furthermore, CP- algorithms [27]. The current state of the art suggests that
OFDM has been shown to be free of the range-Doppler there is still room for waveform design to strike a balance
coupling problem, which means that the range and Doppler between good communication and sensing performance to
estimation can be performed independently [25]. However, meet 6G ISAC requirements.
these favorable properties of CP-OFDM depend on perfect
synchronization (in both time and frequency domains)
5.3 Hardware Co-design
between the transmitter and the receiver, which may not be
present, especially for bistatic sensing. In addition, the large
In the design of the ISAC system, the solution that integrates
peak to average power ratio (PAPR) of CP-OFDM is another
the baseband and RF hardware reduces the overall power
major issue for radar applications where power efficiency is
consumption, system size, and information exchange
very important.
latency between the two systems. The hardware converging
strategy facilitates the mutually beneficial functions of
Alternatively, frequency modulated continuous wave
sensing and communication in distortion calibration and
(FMCW) waveform, which has traditionally been used for
compensation. The common impairments in the ISAC
radar, is not capable of carrying data at transmission rates
system due to hardware imperfections are demonstrated in
desirable for communication services. Some researchers
Figure 18.
proposed to modify the FMCW waveform to make it more
communication-friendly. Among the many contributions
It should be noted that in light of the differences in
in this line of research, we can mention [26], in which the
evaluation metrics and algorithms between communication
authors propose to use up-chirp for communication and
and sensing, hardware requirements are quite
down-chirp for radar, and [27], introducing trapezoidal
different. Considering the cost and size of the historical
f re q u e n c y m o d u l a t i o n co n t i n u o u s -wa ve ( T F M C W )
communication and radar systems, the ISAC system
modulation in which the radar cycle and communication
hardware design will closely resemble the traditional
cycles are multiplexed in the time domain. Although these
communication architecture. As a tradeoff, we need to
techniques enable efficient multiplexing of communication
consider the impact of distortion parameters on sensing
data in the sensing signal, they still suffer from low spectral
performance. For instance, a communications system
efficiency due to the existence of the chirp-like sensing
depends on full duplex isolation to achieve high capacity.

Antenna Full duplex

distortion interference
Sampling jitter
AWGN noise I/Q mismatch

Multipath
channel
WF mod DAC PA LIN ADC WF demod

PLL Non-
linearities PLL Resolution/Clipping
IIP2/IIP3 SFO
CFO Sampling jitter
Phase noise CFO
Flicker noise Phase noise
Flicker noise

Figure 18 Impairments of ISAC transmission system

22 | Communications of HUAWEI RESEARCH | September 2022

In contrast, from the OFDM ISAC perspective, limited benefit of sensing for communication would be improving
transmitter-receiver isolation is a primary concern in the the users' positioning accuracy by combining the advantages
detection of static targets [28]. Proper design of integrated of active localization and passive localization and thus
RF architecture and self-interference cancellation in the ISAC overcoming their shortcomings to satisfy 6G localization
system are key technical problems that need to be solved. requirements.
Another issue is that sensing requires the accumulation
of coherent signals to ensure performance, which makes
5.5 Communication-assisted Collaborative
the system more sensitive to sampling jitter, frequency
Sensing
offset, and phase noise [29]. This in turn leads to higher
requirements on synchronization and stability of the system.
6G ISAC takes advantage of the mobile communication
In short, we need to consider these hardware challenges in
network to support synchronized, collaborative multi-
the selection of ISAC waveforms, sensing algorithms, and
node sensing. Sensing through cooperation refers to the
non-ideal distortion compensation schemes.
sensing nodes that share their observations with each
other and attempt to reach a common consensus on the
5.4 Sensing-assisted Communication surrounding environment. This will significantly improve
localization performance. Integration of sensing capabilities
Although sensing will be introduced as a separate service into the existing communication network will be the
in the future, it might still be beneficial to look at how most viable and cost effective option where the multiple
the information obtained through sensing can be used in network nodes (base stations, UEs, etc.) can function as
communication. The most trivial benefit of sensing will a complete sensing system to enable network sensing
be environment characterization, which enables sensing- operations for the use cases highlighted in Section 2. The
assisted communication due to more deterministic and process involves the collaborating nodes forming a dynamic
predictable propagation channels. It has been shown that reference grid through distributed sensing and processing.
the environment knowledge provided by sensing not only The collaboration reduces measurement uncertainty
improves the accuracy of channel estimation in mmWave, and provides greater coverage as well as higher sensing
but also significantly reduces the overhead because the accuracy and resolution through sensing data fusion.
environment is shared by potentially many UEs and sensing- In addition, this offers interesting possibilities for being
based channel acquisition does not repeat the channel able to carry out sensing under non-line-of-sight (NLOS)
estimation process for each individual link [30]. Another conditions. The major research challenges here would lie in
example would be sensing-assisted beam alignment, the synchronization, joint processing, and network resource
especially in mmWave vehicular communication where the allocation in order to achieve the optimum sensing fusion
main challenge is the frequent link reconfiguration resulting results.
in significant overhead. In [31], it has been proposed to
use the information obtained from a radar mounted on
6 Conclusion
an infrastructure operating in a given mmWave band
to configure the beams of the vehicular communication
With the concept of ISAC being commonly accepted as
system operating in another mmWave band. Moreover,
one of the key technology trends for 6G, this paper takes
the users' location information and the environment map
a step forward and elaborates two case studies on how 6G
obtained by sensing helps identify the link blockage caused
ISAC technologies can be applied to improve localization
by large objects, especially in dense urban networks, so
and to perform high resolution imaging. In particular, the
that the power and beams can be adjusted accordingly
proposed SAPE scheme utilizes the joint benefit of device-
to improve the communication throughput [32]. Other
free and device-based sensing and greatly improves the
examples of sensing-assisted communication can also be
positioning accuracy compared with the current NR scheme.
considered and studied to reduce the latency and overhead
The prototype of the THz camera justifies the feasibility
of communication systems in future 6G networks with the
of mm-level imaging resolution on portable devices for
help of information provided by the sensing system. Another

September 2022 | Communications of HUAWEI RESEARCH | 23

Outlook
both 2D and 3D objects placed in a box. Joint efforts from
both academia and industry are needed to address further 86th Vehicular Technology Conference (VTC-Fall) ,

challenges in the system level evaluation of ISAC, new 2017, pp. 1-5, doi: 10.1109/VTCFall.2017.8287882.

channel modeling methodology, new waveform design, low

[8] F. Wen, N. Garcia, J. Kulmer, K . Witrisal, and
complexity algorithm design, and low cost hardware design.
H. Wymeersch, "Tensor decomposition based
beamspace ESPRIT for millimeter wave
MIMO channel estimation," 2018 IEEE Global

References Communications Conference (GLOBECOM) , 2018,

pp. 1-7, doi: 10.1109/GLOCOM.2018.8647176.

[1] W. Tong, P. Zhu, et al ., "6G: the next horizon:

[9] Pei Chen and H. Kobayashi, "Maximum likelihood
from connected people and things to connected
channel estimation and signal detection for OFDM
intelligence," Cambridge: Cambridge University Press,
systems," 2002 IEEE International Conference on
2021.
Communications. Conference Proceedings . ICC 2002

[2] D. K. P. Tan, J. He, Y. Li, A . Bayesteh, Y. Chen, (Cat. No.02CH37333), 2002, pp. 1640-1645 vol.3, doi:

P. Zhu, and W. Tong, "Integrated sensing and 10.1109/ICC.2002.997127.

communication in 6g: motivations, use cases,

[10] R. Carvajal, J. C. Aguero, B. I. Godoy, and G. C.
requirements, challenges and future directions," in
Goodwin, "EM-based maximum-likelihood channel
1st IEEE International Online Symposium on JC&S ,
estimation in multicarrier systems with phase
23-24 February, 2021.
distortion," in IEEE Transactions on Vehicular

[3] O. Li et al., "Integrated sensing and communication Technology , vol. 62, no. 1, pp. 152-160, Jan. 2013,

in 6G: a prototype of high resolution THz sensing doi: 10.1109/TVT.2012.2217361.

on portable device," in European Conference on

[11] M. Feder and E. Weinstein, "Parameter estimation of
Networks and Communications (EuCNC) , 8-11 June,
superimposed signals using the EM algorithm," in
2021.
IEEE Transactions on Acoustics, Speech, and Signal

[4] K. Witrisal et al., "High-accuracy localization for Processing , vol. 36, no. 4, pp. 477-489, April 1988,

assisted living: 5G systems will turn multipath doi: 10.1109/29.1552.

channels from foe to friend," in IEEE Signal

[12] Y. Bar-Shalom, F. Daum, and J. Huang, "The
Processing Magazine , vol. 33, no. 2, pp. 59-70, March
probabilistic data association filter," in IEEE Control
2016, doi: 10.1109/MSP.2015.2504328.
Systems Magazine , vol. 29, no. 6, pp. 82-100, Dec.

[5] F. Talebi and T. Pratt, "Channel sounding and 2009, doi: 10.1109/MCS.2009.934469.

parameter estimation for a wideband correlation-

[13] Paul Meissner, Christoph Steiner, and Klaus Witrisal,
based MIMO model," in IEEE Transactions on
"UWB positioning with virtual anchors and floor plan
Vehicular Technology , vol. 65, no. 2, pp. 499-508,
information," 2010 7th Workshop on Positioning,
Feb. 2016, doi: 10.1109/TVT.2015.2404571.
Navigation and Communication , IEEE, 2010.

[6] N. Ben Rejeb, I. Bousnina, M. B. Ben Salah, and

[14] Paul Meissner, Thomas Gigl, and Klaus Witrisal,
A. Samet, "Channel parameters estimation using
"UWB sequential Monte Carlo positioning using
cross-correlation matrix of a wireless SIMO system,"
virtual anchors," 2010 International Conference on
Fourth International Conference on Communications
Indoor Positioning and Indoor Navigation , IEEE, 2010.
and Networking, ComNet-2014 , 2014, pp. 1-5, doi:
10.1109/ComNet.2014.6840915. [15] Paul Meissner et al ., "Analysis of an indoor UWB
channel for multipath-aided localization," 2011
[7] L. Wei, Q. Li, and G. Wu, "Direction of arrival
IEEE International Conference on Ultra-Wideband
estimation with uniform planar array," 2017 IEEE
(ICUWB) , IEEE, 2011.

24 | Communications of HUAWEI RESEARCH | September 2022

[16] Deissler, Tobias, and Jörn Thielecke., "UWB SLAM [25] M. Braun, C. Strum, and F. K. Jondral, "Maximum
with rao-blackwellized Monte Carlo data association," likelihood speed and distance estimation for OFDM
2010 International Conference on Indoor Positioning radar," in Proc. 2010 IEEE Radar Conf ., Washington,
and Indoor Navigation , IEEE, 2010. DC, May 2010.

[17] Paul Meissner and Klaus Witrisal, "Multipath- [26] G. N. Saddik, R. S. Singh, and E. R. Brown, "Ultra-
assisted single-anchor indoor localization in an office wideband multifunctional communications/radar
environment," 2012 19th International Conference system," IEEE Trans. Microw. Theory Tech ., pp. 1431-
on Systems, Signals and Image Processing (IWSSIP) , 1437, July 2007.
IEEE, 2012.
[27] K. Wu and L. Han, "Joint wireless communication and
[18] C. B. Barneto, T. Riihonen, M. Turunen, L. Anttila, radar sensing systems - state of the art and future
M. Fleischer, K . Stadius, J. Ryynänen, and M. prospect," IET Microwaves Antennas & Propagation,
Valkama, "Full-duplex OFDM radar with LTE and vol. 7, no. 11, pp. 876-885, 2013.
5G NR waveforms: challenges, solutions, and
[28] C. B. Barneto, T. Riihonen, M. Turunen, L. Anttila,
measurements," in IEEE Transactions on Microwave
M. Fleischer, K . Stadius, J. Ryynänen, and M.
Theory and Techniques , vol. 67, no. 10, pp. 4042-
Valkama, "Full-duplex OFDM radar with LTE and
4054, Oct. 2019.
5G NR waveforms: challenges, solutions, and
[19] K. Siddiq, R. J. Watson, S. R. Pennock, P. Avery, R. measurements," in IEEE Transactions on Microwave
Poulton, and B. Dakin-Norris, "Phase noise analysis Theory and Techniques , vol. 67, no. 10, pp. 4042-
in FMCW radar systems," 2015 European Radar 4054, Oct. 2019.
Conference (EuRAD) , Paris, pp. 501-504, 2015.
[29] K. Siddiq, R. J. Watson, S. R. Pennock, P. Avery, R.
[20] C. Wang, Y. Li, Z. Li, K. Zeng, J. He, and G. Wang, Poulton, and B. Dakin-Norris, "Phase noise analysis
"A 3D imaging method for future communication in FMCW radar systems," 2015 European Radar
and imaging integrated terminal," The 2021 CIE Conference (EuRAD) , Paris, pp. 501-504, 2015.
International Conference on Radar .
[30] C. Jiao, Z. Zhang, C. Zhong, and Z. Feng, "An indoor
[21] X. Li, J. He, Z. Yu, G. Wang, and P. Zhu, "Integrated mmWave joint radar and communication system
sensing and communication in 6G: the deterministic with active channel perception," in 2018 IEEE
channel models for THz imaging," 2021 IEEE 32nd International Conference on Communications (ICC) ,
annual international symposium on personal, indoor Kansas City, MO, 2018.
and mobile radio communications (PIMRC) , 2021,
[31] N. González-Prelcic, R. Méndez-Rial, and R. W. Heath,
pp. 1-6, doi: 10.1109/PIMRC50174.2021.9569384.
"Radar aided beam alignment in mmWave V2I
[22] T. Jin, X. Qiu, D. Hu, and C. Ding, "Unambiguous communications supporting antenna diversity," in
imaging of static scenes and moving targets with the Information Theory and Applications Workshop (ITA) ,
first Chinese dual-channel spaceborne sar sensor," La Jolla, CA, 2016.
Sensors 17(8), 1709 (2017).
[32] Z. Li, S. Yang, and T. Clessienne, "Exploiting location
[23] A . Vertiy and S. Gavrilov, "Near-field millimeter information to enhance throughput in downlink
wave and microwave tomography imaging," Proc. V2I systems," in 2018 IEEE Global Communications
Int. Kharkov Symp. Phys. Engrg. Millim. Sub-Millim. Conference (GLOBECOM) , Abu Dhabi, United Arab
Waves (MSMW) , Jun. 2007, pp. 104-108. Emirates, 2018.

[24] B. Paul, A. R. Chiriyath, and D. W. Bliss, "Survey of RF

communications and sensing convergence research,"
IEEE Access, pp. 252 - 270, 2016.

September 2022 | Communications of HUAWEI RESEARCH | 25