![The comparison of size scale between a one-cent coin and two QTM052 mmWave antenna modules announced in July and October 2018, respectively. (Image source: Qualcomm Technologies, Inc. [86]).](https://www.researchgate.net/profile/Long-Zhang-55/publication/334498109/figure/fig3/AS:781454032248832@1563324568029/The-comparison-of-size-scale-between-a-one-cent-coin-and-two-QTM052-mmWave-antenna_Q320.jpg)
Content uploaded by Long Zhang
Author content
All content in this area was uploaded by Long Zhang on Mar 03, 2020
Content may be subject to copyright.
Content uploaded by Long Zhang
Author content
All content in this area was uploaded by Long Zhang on Jul 19, 2019
Content may be subject to copyright.
Available via license: CC BY 4.0
Content may be subject to copyright.
IEEE ACCESS 1
A Survey on 5G Millimeter Wave Communications
for UAV-Assisted Wireless Networks
Long Zhang, Member, IEEE, Hui Zhao, Shuai Hou, Zhen Zhao, Haitao Xu, Member, IEEE,
Xiaobo Wu, Qiwu Wu, and Ronghui Zhang
Abstract—In recent years, unmanned aerial vehicles (UAVs)
have received considerable attention from regulators, industry
and research community, due to rapid growth in a broad
range of applications. Particularly, UAVs are being used to
provide a promising solution to reliable and cost-effective wireless
communications from the sky. The deployment of UAVs has
been regarded as an alternative complement of existing cellular
systems, to achieve higher transmission efficiency with enhanced
coverage and capacity. However, heavily utilized microwave
spectrum bands below 6 GHz utilized by legacy wireless systems
are insufficient to attain remarkable data rate enhancement
for numerous emerging applications. To resolve the spectrum
crunch crisis and satisfy the requirements of 5G and beyond
mobile communications, one potential solution is to use the
abundance of unoccupied bandwidth available at millimeter wave
(mmWave) frequencies. Inspired by the technique potentials,
mmWave communications have also paved the way into the
widespread use of UAVs to assist wireless networks for future
5G and beyond wireless applications. In this paper, we pro-
vide a comprehensive survey on current achievements in the
integration of 5G mmWave communications into UAV-assisted
wireless networks. More precisely, a taxonomy to classify the
existing research issues is presented, by considering seven cutting-
edge solutions. Subsequently, we provide a brief overview of
5G mmWave communications for UAV-assisted wireless networks
from two aspects, i.e., key technical advantages and challenges as
well as potential applications. Based on the proposed taxonomy,
we further discuss in detail the state-of-the-art issues, solutions,
and open challenges for this newly emerging area. Lastly, we
complete this survey by pointing out open issues and shedding
new light on future directions for further research on this area.
Index Terms—millimeter wave (mmWave) communications,
unmanned aerial vehicle (UAV), mmWave UAV communications,
UAV-assisted wireless networks, 5G and beyond.
This work was supported in part by the National Natural Science Foun-
dation of China under Grants 61501406, 51775565, and 61802107, the
Research Program for Top-notch Young Talents of Hebei Province, China
under Grant BJ2017037, the Natural Science Foundation of Hebei Province
under Grants F2019402206, E2017402115, F2017402068, and F2018402198,
and the Natural Science Foundation of Guangdong Province under Grant
2018A030313492. Corresponding authors: Long Zhang (e-mail: zhang-
long@hebeu.edu.cn) and Hui Zhao (e-mail: Zhaohui_0507@hotmail.com).
L. Zhang, H. Zhao, S. Hou, and Z. Zhao are with the School of In-
formation and Electrical Engineering, Hebei University of Engineer-
ing, Handan 056038, China (e-mail: zhanglong@hebeu.edu.cn, Zhao-
hui_0507@hotmail.com, houshuai@hebeu.edu.cn, zhen1121@hotmail.com).
H. Xu is with the School of Computer and Communication Engineering,
University of Science and Technology Beijing, Beijing 100083, China (e-mail:
xuhaitao@ustb.edu.cn).
X. Wu is with the China Academy of Transportation Sciences, Beijing
100029, China (e-mail: wuxiaobo@motcats.com.cn).
Q. Wu is with the Department of Armaments Management and Support,
Engineering University of PAP, Xi’an 710086, Shaanxi, China (e-mail: wuqi-
wu700@163.com).
R. Zhang is with the Guangdong Key Laboratory of Intelligent Trans-
portation System, School of Intelligent Systems Engineering, Sun Yat-sen
University, Guangzhou 510275, China (e-mail: zrh1981819@126.com).
I. INTRODUCTION
UNMANNED aerial vehicles (UAVs), popularly known
as drones, refer to a class of aircrafts that operates
without a human pilot aboard. The flight of UAVs can be
realized with various degrees of autonomy: either under the
remote control by a human operator or autonomously by the
onboard sensors and computers. Due to their capability to
hover, sufficient flexibility, ease of deployment, higher maneu-
verability, lower operating and maintenance costs, UAVs have
attracted significant attention recently for both civilian and
commercial applications. They are being used in a broad range
of ways ranging from precision agriculture, smart logistics, law
enforcement, disaster response, prehospital emergency care,
and mineral exploration, to personal business drone-based
photography and drone racing [1]–[6]. For instance, Amazon
is projected to own nearly 450,000 commercial drones in its
delivery fleet by 2020 under the proposed “Amazon Prime Air”
service, operating with delivery worldwide [7].
Meanwhile, the advent of a new generation of highly
capable UAVs has also led to plenty of interest in the use
of UAVs to provide reliable and cost-effective solutions for
wireless communications in many real-world scenarios [8]–
[14]. In particular, by the aid of wireless communication
modules, UAVs can be rapidly deployed as flying base stations
(BSs) in three-dimensional (3D) space. The aerial deployment
has been regarded as a promising way to provide ubiquitous
access from the sky towards ground user equipments (UEs)
in specified areas during temporary events (e.g., hotspot areas
and large public venues) [8], [12], [13], [15]–[18]. Compared
with terrestrial static BSs, one unmatched advantage of using
UAV-based flying BSs is that their positions and altitudes can
be dynamically adjusted to provide on-the-fly air-to-ground
(A2G) communication links. The placement of flying BSs has
been viewed as an effective complement of existing cellular
systems to enhance wireless capacity and coverage footprint
on the ground with ultra dense traffic demands, to meet the
requirements of the fifth generation (5G) and beyond 5G
(B5G) cellular mobile communications. As a consequence of
practical demands, several drone programs about aerial BSs1
from industrial communities have already been initiated, e.g.,
AT&T’s all-weather flying COW (Cell on Wings) to provide
the Long Term Evolution (LTE) coverage of the fourth genera-
tion (4G) cellular systems, Verizon’s Airborne LTE Operations
(ALO) initiative to enable the access to wireless connectivity,
1Throughout this paper, the terms "aerial BS" and "flying BS" can be used
interchangeably.
IEEE ACCESS 2
Nokia Bell Labs’ F-Cell (flying-cell) to enable the “drop and
forget” small cell deployment anywhere, Facebook’s Aquila
using the solar-powered drone to blast Internet to underserved
areas via laser communications, etc.
On the other hand, UAVs can be utilized as mobile relays
to provide wireless connectivity for partitioned ground UEs
without any direct line-of-sight (LOS) transmission links be-
cause of the blocking of physical obstacles like mountains or
buildings [9], [19]. Given this scenario, the load-carry-and-
deliver forwarding strategy is well tailored to the use of a sin-
gle UAV to relay data packets from one source UE to another
destination UE, aiming to achieve higher system throughput
[20]. From the perspective of aerodynamics characteristics, the
distinct features of various mainstream drones, such as flying
altitude, energy resource, maximum payload, and endurance
time, will necessitate the configuration of multi-tier archi-
tecture for future drone-cell networks [12], [21]–[23]. It has
been further highlighted that the coordination, collaboration,
and self-organizing flocking behavior of multiple UAVs under
this architecture can create flying ad hoc networks (FANETs),
Internet of Drones (IoD), and even drone swarm like a flock
of birds which is beyond the capability of one single UAV
[24]–[28]. Technically, the integration of collective intelligence
into the drone swarm will ultimately bring lots of benefits
for 5G cellular mobile systems, e.g., higher downlink spectral
efficiency and better quality of experience (QoE) for ground
UEs.
From all the above realistic applications of interest, we
can observe that the widespread use of UAVs has exerted an
important influence on wireless networking. The networking
strategy by employing one single UAV or multiple UAVs to
assist wireless networks has attracted even more attention from
both academia and industry. Beyond that, several other typical
applications about UAV-assisted wireless networks include
wireless sensor networks (WSNs) [29], vehicle-to-everything
(V2X) communications [30], Internet of Things (IoT) [31],
flyMesh [32], wireless powered networks [33], caching aided
wireless networks [34], mobile edge computing (MEC) [35],
[36], cloud radio access networks (CRANs) [37], emergency
networks [38], device-to-device (D2D) communications [19],
cognitive radio (CR) networks [39], etc. In addition to the
role of classic aerial BS, the part played by UAVs in these
application scenarios has covered various networking entities
such as aerial relay, aerial data collector, aerial caching,
aerial MEC server, aerial power source, and so on. These
representative applications by leveraging one single UAV or
multiple UAVs to assist wireless networks are summarized in
Table I for the ease of reference.
In the past few years, the rapid proliferation of mas-
sive number of internet-connected things (e.g., smart de-
vices, industrial sensors, driverless cars, connected factories,
etc.), and their various emerging applications (e.g., ultra-
high definition video, artificial intelligence, face recognition,
Tactile Internet, augmented reality (AR), virtual reality (VR),
broadband kiosks, cloud gaming, autonomous driving, etc.)
have propelled the explosive growth in wireless data traffic.
However, radio spectrum resources in legacy cellular systems
and conventional wireless local area networks (WLANs) based
on IEEE 802.11 standards2that are all operating in the mi-
crowave frequency bands below 6 GHz frequencies have been
already heavily utilized. There is more and more concern about
limited spectrum available for use relative to the exponentially
growing data capacity demand. Several enabling techniques
have been proposed to achieve enhanced network capacity
and higher spectral efficiency for future cellular systems, e.g.,
massive multiple-input and multiple-output (MIMO) [40], cog-
nitive radio [41], network densification [42], non-orthogonal
multiple access (NOMA) [43], advanced channel coding and
modulation [44], cooperative relaying [45], etc. Nevertheless,
those advances will not be an essential step towards tackling
the issue of limited available spectrum.
To overcome the spectrum crunch crisis, one potential
solution has been to expand the use of higher frequencies in
radio spectrum, since large amount of licensed and unlicensed
spectrum bands are still available at these frequencies. As such,
the expanded use of millimeter wave (mmWave) frequencies,
ranging from around 30 GHz to 300 GHz, has been recognized
as a promising approach for providing wide available spectrum
resources and multiple Gigabit data transmission speeds [46]–
[52]. With this incomparable advantage, mmWave commu-
nications are indeed more suited to 5G and B5G wireless
systems with the requirements of massive data throughput,
high wireless bandwidth, super-fast speeds, ultra-low latency,
and increased connectivity for a networked society. They will
also be dedicated to the support of a wide variety of broadband
applications envisioned for Multiple Gigabit Wireless Systems
(MGWS) in the frequencies around 60 GHz unlicensed indus-
trial, scientific and medical (ISM) frequency band, and WiGig
(wireless Gigabit) network as the 60 GHz Wireless Fidelity
(Wi-Fi) solution [53]. In addition, the use of mmWave com-
munications has been highlighted as a key enabler for 5G New
Radio (NR) developed by the Third Generation Partnership
Project (3GPP) as a unified air interface to support many
user cases expected in 5G, e.g., enhanced Mobile Broadband
(eMBB) and Ultra-Reliable Low Latency Communications
(URLLC) [54]–[56].
Owing to the promising applications of UAVs to assist wire-
less networks along with the possibility of multiple Gigabit
data transmissions by using 5G mmWave communications, a
natural idea is to link UAV-assisted wireless networking and
mmWave communications together. Fig. 1 depicts the appli-
cation scenario of 5G mmWave communications for UAV-
assisted wireless networks. In the typical scenario, multiple
rotary-wing drones act as flying BSs, aerial radio access points,
and aerial relays via A2G mmWave beams to expand wireless
coverage footprint and provide multi-Gigabit transmission
towards physical objects including ground UEs in 5G systems,
and ground vehicles in vehicular networks via vehicle-to-
vehicle (V2V) and vehicle-to-infrastructure (V2I) communi-
cations. The information exchange and transfer among several
2Here the IEEE 802.11 standards refer to the set of WLAN protocols that
specifies the set of media access control (MAC) and physical layer (PHY)
specifications. In this work, this set of standards does not contain the recent
most expected WLAN technologies for ultra-high-speed communications, e.g.,
IEEE 802.11ad and IEEE 802.11ay operating at unlicensed 60 GHz mmWave
frequency, which are still requesting proposals from regulators, industry and
research community.
IEEE ACCESS 3
TABLE I
SUMMARY OF TYPICAL APPLICATIONS OF UAV-ASS IS TED W IR ELE SS N ETW ORK S.
Typical Application Direction Role of UAV Achieved Function for UAV Reference
Wireless cellular networks Aerial BS Ubiquitous coverage and access [12], [15]–[17]
Wireless relay networks Aerial relay Wireless connectivity [9], [19]
Multi-tier air-ground networks Aerial BS Higher downlink spectral efficiency [12], [21], [22]
Flying ad hoc networks Aerial drone swarm Enhanced usage efficiency of single UAV [24], [26]
Wireless sensor networks Aerial data collector Data collection and dissemination [29]
V2X communications Aerial RSU/relay/radio access V2X applications [30]
Internet of Things Aerial IoT device Intelligent applications [31]
FlyMesh Aerial relay Coverage and fully connected UAV networks [32]
Wireless powered networks Aerial power source Wireless power transfer [33]
Caching aided wireless networks Aerial caching Reduced wireless backhaul load [34]
Mobile edge computing Aerial MEC server Dynamic deployment of edges [35], [36]
Cloud radio access networks Aerial radio access/caching Wireless connectivity and caching distribution [37]
Emergency networks Aerial BS/relay Coverage and information relaying [38]
D2D communications Aerial power source Wireless power transfer [19]
Cognitive radio networks Aerial cognitive device Dynamic spectrum access [39]
D2DPair
D2DPair
D2DPair
CellularLink
CellularLink
CellularLink
Cellular Link
CellularLink
Cellular UE
Cellular
UE BS
BS
RSU
V2ILink
V2V Link
BS
D2DPair
CellularLink
CellularLink
CellularLink
Information
Tansfer
Information
Transfer
A2G MmWave
Beam
A2G MmWave Beam
A2G MmWave
Beam
BS Coverage
A2A MmWave
Beam
Fixed- Wing Drone
Rotary- Wing Drone
A2A Microwave Link
A2A
Microwave
Link
Fig. 1. Illustration of 5G mmWave communications for UAV-assisted wireless networks.
drones, e.g., fixed-wing drones and rotary-wing drones, can be
achieved through multi-hop drone relaying by means of air-
to-air (A2A) mmWave beams. Moreover, drones are deployed
as access networks for ground D2D transmitters (TXs) and
receivers (RXs) (i.e., D2D pairs) at hotspot areas to provide
additional capacity without the support of terrestrial BSs.
There is no doubt that the use of mmWave frequencies
will be definitely a very promising way for UAV-assisted
wireless networks to sustain ultra-high-speed and real-time
transmissions for future 5G and B5G wireless applications.
The integration of the merits from these two techniques has
inspired their respective advantages:
•MmWave communications for UAV-assisted wireless net-
works can support higher data rate due to higher band-
width. It is easier to achieve the peak data rate of up
to 10 Gbit/s, which is unlikely to be matched by any
wireless technologies with lower microwave frequencies.
Conceptually, a large amount of licensed and unlicensed
mmWave frequency bands are potentially available for
use in UAV-assisted wireless networks, e.g., 28 GHz
licensed band and 60 GHz unlicensed ISM band [57].
•Much shorter mmWave wavelength makes it feasible to
design the physically smaller mmWave circuits and an-
tennas, which enables to build the beamforming antenna
arrays with many antennas to be packed on a chip in a tiny
component size [52], [58]. It is perfect for low-altitude
short range UAVs with limited payloads.
•Low interference and increased security benefit from the
inherently directional mmWave beams and the narrow
beamwidth with a high immunity to jamming and eaves-
dropping [48], [59].
In spite of being advantageous, atmospheric absorption at
mmWave wavelengths is a concern of 5G mmWave commu-
nications for UAV-assisted wireless networks, thus restricting
IEEE ACCESS 4
Position and Trajectory
Optimization
Fig. 2. Taxonomy of related research issues in 5G mmWave communications
for UAV-assisted wireless networks.
their transmission range of mmWave signals and resulting in
high propagation attenuation at higher mmWave frequencies
[60], [61]. Another major technical challenge is blockage ef-
fect, when travelling mmWave signals are blocked by physical
obstacles in their propagation paths. For instance, human-
body blockage [62]–[65] and self-blockage [59] caused by
human activity and components of UAV itself, respectively,
which poses extra challenges on mmWave LOS propagation.
In this paper, we focus on the scenario of the combination
of 5G mmWave communications with UAV-assisted wireless
networks and the emerging studies related to the integration
of these two important areas.
Although UAV-assisted wireless networks and mmWave
communications have been studied extensively in previous
works, existing surveys and tutorials either cover UAV-assisted
wireless networks and UAV communications for 5G and
beyond [8], [11], [12], [24], [25], [66]–[68], or concentrate
on mmWave communications for 5G and vehicular communi-
cations [48]–[50], [69]–[73]. Those existing contributions are
still separated across these two important areas. To the best
of our knowledge, only two existing survey papers in [10],
[66] are related to mmWave communications for UAV-assisted
wireless networks. Nevertheless, the work in [66] by Li et
al. just shortly review the technology challenges and existing
solutions for mmWave UAV-assisted cellular networks as one
of physical layer techniques over the entire topic focused on.
Although the work in [10] by Xiao et al. discussed the major
technical challenges and possible solutions in mmWave UAV
cellular systems for the first time, they did not fully cover the
detailed progress of the state-of-the-art comprehensively. This
motivates us to deliver a detailed survey with the objective
to give a comprehensive summary of current achievements in
the integration of 5G mmWave communications into UAV-
assisted wireless networks, and to discuss the state-of-the-art
issues, solutions, and open challenges in this newly emerging
area. In this survey, a taxonomy graph to classify the related
research issues based on the existing literatures is presented
in Fig. 2. As displayed in Fig. 2, we will discuss seven
important research issues that we would like to be targeted
at: antenna technique, radio propagation channel, multiple ac-
cess mechanism, spatial configuration, resource management,
security strategy, and performance assessment. To the authors’
best knowledge, this paper is the first comprehensive effort to
review the cutting-edge solutions related with 5G mmWave
communications for UAV-assisted wireless networks in a holis-
tic way. With this survey, our objective is to provide readers
with an overall understanding of what has been accomplished
so far, in the area of 5G mmWave communications for UAV-
assisted wireless networks, and to open up new perspectives
for subsequent studies on this area.
For convenience, a list of acronyms used in this paper is pro-
vided in Table II. The remainder of the paper is organized as
follows. Firstly, existing surveys and tutorials related to UAV-
assisted wireless networks and communications along with 5G
mmWave communications are discussed and compared with
our survey in Section II. Section III presents the brief overview
of 5G mmWave communications for UAV-assisted wireless
networks including key technical advantages and challenges
of 5G mmWave communications, and brief Introduction to
the integration of 5G mmWave communications into UAV-
assisted wireless networks. In Section IV, we comprehensively
review the state-of-the-art research issues and challenges from
the perspective of the taxonomy of the related research issues.
Then Section V summarizes the current research and proposes
the open issues together with the future research directions.
Finally, Section VI concludes the paper. For the sake of clear
presentation, the organizational structure of this paper in a
top-down manner is exhibited in Fig. 3.
Notations: Throughout this paper, we use a,a,A, and
Ato denote a scalar variable, a vector, a matrix, and a
set, respectively. The cardinality for a set Ais denoted by
|A|. Further, (·)∗and (·)Hdenote conjugate and Hermitian
(conjugate transpose), respectively. We use [a]+to represent
the maximum value between 0and variable a, i.e., [a]+=
max {a, 0}. The set of real numbers is denoted by R. The
operator E[·]is the expectation operation, and the operator
IEEE ACCESS 5
TABLE II
LIST OF ACRONYMS.
Acronym Meaning Acronym Meaning
2D Two-Dimensional Space LTE Long Term Evolution
3D Three-Dimensional Space MAC Media Access Control
3GPP Third Generation Partnership Project MBH Multi-Beam Horn
4G Fourth Generation Cellular Mobile Communications MEC Mobile Edge Computing
5G Fifth Generation Cellular Mobile Communications MGWS Multiple Gigabit Wireless Systems
6G Sixth Generation Mobile Systems MIMO Multiple-Input and Multiple-Output
A2A Air-to-Air MINLP Mixed Integer Non-Linear Programming
A2G Air-to-Ground ML Maximum Likelihood
AAN Aerial Access Network MS Mobile Station
ALO Airborne LTE Operations MSE Mean Square Error
AN Artificial Noise MmWave Millimeter Wave
AoA Angle of Arrival NFV Network Functions Virtualization
AoD Angle of Departure NLOS Non-Line-of-Sight
AP Access Point NOMA Non-Orthogonal Multiple Access
AR Augmented Reality NR New Radio
AWV Antenna Weight Vector OFDM Orthogonal Frequency Division Multiplexing
BBU Baseband Unit OMA Orthogonal Multiple Access
BDMA Beam Division Multiple Access PDF Probability Density Function
B5G Beyond 5G Cellular Mobile Communications PHY Physical Layer
BS Base Station QoE Quality of Experience
CCDF Complementary Cumulative Distribution Function QoS Quality of Service
CDF Cumulative Distribution Function RF Radio Frequency
COW Cell on Wings RRH Remote Radio Head
CPS Cyber Physical System RSS Received Signal Strength
CR Cognitive Radio RSU Road Side Unit
CRAN Cloud Radio Access Network RX Receiver
CSI Channel State Information SBS Small Base Station
D2D Device-to-Device SDMA Spatial-Division Multiple Access
DBF Digital Beamforming SINR Signal-to-Interference-Plus-Noise Ratio
DM Directional Modulation SPF Stratospheric-Platform
DoA Direction of Arrival TDMA Time-Division Multiple Access
EH Energy Harvesting ToA Time of Arrival
EHF Extremely High Frequency TR Tensor Recovery
ELSM Echo Liquid State Machine TX Transmitter
eMBB enhanced Mobile Broadband UAV Unmanned Aerial Vehicle
F-Cell Flying-Cell UCA Uniform Circular Array
ESN Echo State Network UE User Equipment
FANET Flying Ad Hoc Network ULA Uniform Linear Array
FCC Federal Communications Commission UPA Uniform Planar Array
G2A Ground-to-Air URA Uniform Rectangular Array
G2G Ground-to-Ground URLLC Ultra-Reliable Low Latency Communications
GMT Ground Mobile Terminal USRP Universal Serial Radio Peripheral
HAPS High-Altitude Platform Station UUV Unmanned Underwater Vehicle
HetNet Heterogeneous Network V2I Vehicle-to-Infrastructure
IoD Internet of Drones V2V Vehicle-to-Vehicle
IoE Internet of Everything V2X Vehicle-to-Everything
IoT Internet of Things VLEO Very Low Earth Orbit
ISM Industrial, Scientific and Medical VR Virtual Reality
ITU International Telecommunication Union WiGig Wireless Gigabit
KNN K-Nearest Neighbor Wi-Fi Wireless Fidelity
LOP Line-of-Propagation WLAN Wireless Local Area Network
LOS Line-of-Sight WSN Wireless Sensor Network
LSM Liquid State Machine XR eXtended Reality
Var [·]is the variance operation. The distribution of a circularly
symmetric complex Gaussian random variable xwith mean Σ
and covariance Λis represented by x∼ CN (Σ , Λ), where ∼
stands for “distributed as”. We employ P L (d)to denote the
free space path loss measured in dB from TX to RX over
a distance d. In addition, |·| and k·k stand for the absolute
value of a scalar variable and the Euclidean distance between
the pair of vectors, respectively. Finally, Fζ(·)is the Fejér
kernel of order ζand Γ (z) = R∞
0xz−1e−xdx is the Gamma
function.
II. REVIEW OF EXISTING SURVEYS AND TUTO RI AL S
In this section, we will present the existing surveys and
tutorials, including UAV-assisted wireless networks and com-
munications as well as 5G mmWave communications. More-
over, the comparison of this paper with existing surveys and
tutorials will be also summarized.
A. UAV-Assisted Wireless Networks and Communications
During the past few years, a number of excellent surveys
and tutorials have been reported to be published in the fields
IEEE ACCESS 6
5HTXLUHPHQWVIRU8$9$VVLVWHG:LUHOHVV1HWZRUNV
6HFWLRQ ,,QWURGXFWLRQ
5HTXLUHPHQWVIRU*0P:DYH&RPPXQLFDWLRQV
0RWLYDWLRQIRU,QWHJUDWLRQRI*0P:DYH
&RPPXQLFDWLRQVLQWR8$9$VVLVWHG:LUHOHVV1HWZRUNV
6HFWLRQ ,,5HYLHZRI([LVWLQJ6XUYH\VDQG7XWRULDOV
8$9$VVLVWHG:LUHOHVV1HWZRUNV
DQG&RPPXQLFDWLRQV *0P:DYH&RPPXQLFDWLRQV
6HFWLRQ ,,,$Q2YHUYLHZRI*0P:DYH&RPPXQLFDWLRQVIRU8$9$VVLVWHG:LUHOHVV1HWZRUNV
6HFWLRQ ,95HVHDUFK,VVXHV&KDOOHQJHVDQG6WDWHRIWKH$UW
$QWHQQD
7HFKQLTXH
5DGLR3URSDJDWLRQ
&KDQQHO
0XOWLSOH$FFHVV
0HFKDQLVP
6SDWLDO
&RQILJXUDWLRQ
5HVRXUFH
0DQDJHPHQW 6HFXULW\6WUDWHJ\ 3HUIRUPDQFH
$VVHVVPHQW
6HFWLRQ 96XPPDU\2SHQ,VVXHVDQG)XWXUH5HVHDUFK'LUHFWLRQV
6XPPDU\ &KDOOHQJHVDQG2SHQ
,VVXHV
3RWHQWLDO5HVHDUFK
'LUHFWLRQV
6HFWLRQ 9,&RQFOXVLRQ
$HULDO$FFHVV
$HULDO5HOD\DQG,QWHUFRQQHFWLRQ
$HULDO%DFNKDXO
3RWHQWLDO$SSOLFDWLRQV %ULHI,QWURGXFWLRQWR7HFKQLFDO$GYDQWDJHVDQG&KDOOHQJHV
0XOWLSOH*LJDELW'DWD7UDQVPLVVLRQ6SHHGV
0XFK6KRUWHU0P:DYH:DYHOHQJWK
/DUJHVFDOH0XOWLHOHPHQW$QWHQQD$UUD\V
+LJKO\'LUHFWLRQDO6LJQDOV*UHDWHU5HVROXWLRQV
$WPRVSKHULF$EVRUSWLRQ3DWK/RVV
%ORFNDJH(IIHFW
+XPDQERG\%ORFNDJH6HOI%ORFNDJH
'\QDPLF%HDP$OLJQPHQW
,QWHJUDWLRQRI*0P:DYH&RPPXQLFDWLRQVLQWR8$9$VVLVWHG:LUHOHVV1HWZRUNV
7HFKQLFDO$GYDQWDJHV
/DUJHU%DQGZLGWK$YDLODELOLW\
6KRUWHU:DYHOHQJWK
1DUURZ%HDP
,QFUHDVHG6HFXULW\DQG,QWHUIHUHQFHLPPXQLW\
7HFKQLFDO&KDOOHQJHV
)UHH6SDFH3DWK/RVV
$WPRVSKHULF$WWHQXDWLRQ
%ORFNDJH(IIHFW
%HDP0LVDOLJQPHQW
.H\7HFKQLFDO$GYDQWDJHVDQG&KDOOHQJHVRI*0P:DYH&RPPXQLFDWLRQV
Fig. 3. The organizational structure of this paper.
of UAV-assisted wireless networks and UAV communications
for 5G and B5G wireless systems. Mozaffari et al. [8] p-
resented a comprehensive tutorial on UAV-enabled wireless
communications and networking applications, and provided
the insightful comments to some important research oppor-
tunities and challenges in flying BSs. The analytical frame-
works and optimization tools were proposed to deal with the
challenging open problems towards UAV communications and
networking. Gupta et al. [25] briefly discussed the challenges
to effectively design stable and reliable context-specific UAV
networks according to their inherent features. The authors put
more emphasis on the problems of routing, seamless handover,
and energy efficiency in UAV networks. With relation to
routing mechanism, Jiang and Han [74] carried out a brief
survey on routing protocols for UAVs based on hop counts,
i.e., single-hop and multi-hop routing. Besides, position-based
routing protocols and cluster-based routing protocols for UAV
networks were extensively surveyed and compared by Oubbati
et al. [75] and Arafat and Moh [76], respectively.
By integrating UAVs into cellular networks, Fotouhi et
al. [11] conducted a comprehensive survey on some ma-
jor networking issues about UAV cellular communications,
IEEE ACCESS 7
TABLE III
ABRIEF COMPARISON OF OUR STUDY WITH EXISTING SURVEY PAPERS AND TUTORIALS.
Survey Paper Year Topic Focused On Main Issues Elaborated and Addressed MmWave-UAV
Bekmezci et al. [24] 2013 Flying ad hoc networks FANET application scenarios, design characteristics,
communication protocols, test beds and simulators Not covered
Rappaport et al. [50] 2013 MmWave communications for
5G and B5G mobile networks MmWave propagation measurement campaigns Not covered
Rangan et al. [69] 2014 MmWave cellular wireless
networks
Channel measurements, adaptive beamforming, multi-hop
relaying, heterogeneous network architectures, and carrier
aggregation
Not covered
Niu et al. [49] 2015 MmWave communications for
5G mobile networks
Peculiar characteristics, research challenges, and potential
applications of mmWave communications Not covered
Gupta et al. [25] 2016 Important issues in UAV
communication networks Routing, seamless handover, and energy efficiency Not covered
Xiao et al. [10] 2016 UAV cellular networks with
mmWave communications
MmWave channel propagation characteristics, fast
beamforming training and tracking, mmWave SDMA,
blockage problems, and directional user discovery
Covered
Xiao et al. [48] 2017 MmWave communications for
5G and B5G mobile networks
MmWave channel measurement campaigns and modeling,
mmWave MIMO systems, multiple-access technologies, and
mmWave bands for backhauling
Not covered
Rappaport et al. [71] 2017 MmWave communications for
5G and B5G mobile networks MmWave propagation models Not covered
Mozaffari et al. [8] 2018 UAV-enabled wireless
communications and networks
Application scenarios, key research directions, challenging
open problems, and summary of analytical frameworks Not covered
Fotouhi et al. [11] 2018 UAV cellular communications
Standardization efforts for aerial UE, deployment of aerial
BSs, drone communication prototypes, UAV regulations,
and security
Not covered
Sekander et al. [12] 2018 Multi-tier drone network
architecture
Potential challenges, performance optimization of
drone-assisted cellular networks, and feasibility analysis of
multi-tier drone-aided cellular architecture
Not covered
Wang et al. [67] 2018 A CPS perspective for UAV
networks
Three CPS components in UAV networks, UAV networks of
three CPS hierarchies, and coupling effects in UAV networks Not covered
Hemadeh et al. [70] 2018 MmWave communications for
5G and B5G mobile networks
MmWave propagation characteristics, channel modeling, and
design guidelines Not covered
Sánchez-García et al. [68] 2018
UAV and unmanned
underwater vehicle (UUV)
multi-hop networks
Wireless communications, evaluation tools and application
scenarios Not covered
Zhou et al. [72] 2018 IEEE 802.11ay based
mmWave WLANs
Channel bonding and aggregation, channel access and
allocation, beamforming training and beam tracking, and
single-user MIMO and multi-user MIMO beamforming
Not covered
Jameel et al. [73] 2019
Propagation channel modeling
for mmWave vehicular
communications
MmWave channel modeling approaches, channel attributes,
and channel models based on three measurement scenarios
including V2I, inter-vehicle, and intra-vehicle
Not covered
Li et al. [66] 2019 UAV communications for 5G
and B5G mobile systems
Physical layer, network layer, joint communication,
computing and caching, and mmWave communications for
UAV-assisted cellular networks
Covered
This survey 2019
5G mmWave communications
for UAV-assisted wireless
networks
Antenna technique, radio propagation channel, multiple
access mechanism, spatial configuration, resource
management, security strategy, and performance assessment
Covered
such as standardization efforts for aerial UE, deployment of
aerial BSs, drone communication prototypes, security and
so forth. Li et al. [66] presented a comprehensive survey
on key techniques for UAV communication towards 5G and
B5G wireless networks from three aspects involving physical
layer, network layer, and joint communication, computing
and caching. Particularly, mmWave communications as a key
technology at physical layer of UAV-assisted cellular networks
was briefly discussed. From the perspective of cyber physical
system (CPS), Wang et al. [67] carried out a systematic review
of UAV networks, with an emphasis on the impact of CPS
components on the system performance of UAV networks.
A survey on wireless communications, evaluation tools and
applications in unmanned aerial and aquatic vehicle multi-hop
networks was described in the literature by Sánchez-García
et al. [68]. Similar to UAV multi-hop networks, Bekmezci
et al. [24] presented a survey on communication protocols
from four layers and cross-layer interactions in FANET which
is typically a kind of ad hoc networks. Driven by future
drone-cell networks, Sekander et al. [12] reviewed the existing
innovations in drone networks and drone-assisted cellular net-
works, and proposed the multi-tier drone network architecture
over the conventional single-tier drone networks.
In addition to these surveys and tutorials on UAV networks
[25], [67], [68], UAV cellular networks [11], [66], UAV-
enabled wireless networks [8], FANETs [24], and multi-tier
drone networks [12], several other surveys have also been pub-
lished on different topics with regard to UAV-related networks
and communications, such as channel modeling [77], civil
applications [1], [78], UAV-assisted vehicular networks [30],
wireless charging techniques [79], IoT services [6], game-
theoretic optimization approach [80], CR-based UAV [39], etc.
B. 5G MmWave Communications
Many other existing survey papers have been dedicated
to mmWave communications for 5G wireless systems and
IEEE ACCESS 8
vehicular communications. By surveying the recent channel
measurements of mmWave signals in urban environments,
Rangan et al. [69] elaborated on several techniques includ-
ing adaptive beamforming, multi-hop relaying, heterogeneous
network (HetNet) architectures, and carrier aggregation in the
context of mmWave cellular wireless networks. Xiao et al.
[48] gave a systematic survey on main technical progress of
mmWave communications for 5G and B5G mobile networks.
The authors focused primarily on mmWave channel mea-
surement campaigns and modeling, mmWave MIMO systems,
multiple-access technologies, mmWave bands for backhauling,
etc. Furthermore, Niu et al. [49] conducted a survey on ex-
isting research progress in 5G mmWave communications. The
authors mainly discussed the peculiar characteristics, research
challenges, and potential applications of mmWave commu-
nications. Other similar surveys and tutorials on mmWave
communications for 5G and B5G were also reported in re-
cent literature [50], [70], [71]. The work of Rappaport et
al. in [50] focused on mmWave propagation measurement
campaigns conducted in New York City, NY with 28 GHz
mmWave frequencies and Austin, TX with 38 GHz mmWave
frequencies. Also, the surveys carried out by Hemadeh et al.
[70] and Rappaport et al. [71] were both targeted at mmWave
propagation characteristics and channel modeling. Faced to
the challenges of ultra-high-speed WLANs, Zhou et al. [72]
presented a comprehensive survey on MAC-related issues,
especially channel access for multiple channels and beam-
forming for both single-user MIMO and multi-user MIMO,
for mmWave WLANs based on the IEEE 802.11ay standard.
Lastly, Jameel et al. [73] briefly discussed the mmWave prop-
agation channel modeling from the perspective of vehicular
communications.
Different from all the works as mentioned previously, Xiao
et al. [10] combined mmWave communications with UAV
cellular networks together, and presented a brief overview of
key technical challenges and possible solutions in mmWave
UAV cellular, including mmWave channel propagation char-
acteristics, fast beamforming training and tracking, mmWave
spatial-division multiple access (SDMA), problem of mmWave
blockage effect, and directional user discovery. To sum up,
as Table III shows, we provide a brief comparison of our
study with existing survey papers outlined before mainly from
three aspects, including topic focused on, main issues elabo-
rated and addressed, and whether the integration of mmWave
communications into UAV-assisted wireless networks or not.
In Table III, we use “mmWave-UAV” as the abbreviation to
represent the integration of mmWave communications into
UAV-assisted wireless networks due to the limited space of this
table. In comparison with the aforementioned survey papers,
this survey provides a critical and systematic overview of the
up-to-date activities related to research issues on 5G mmWave
communications for UAV-assisted wireless networks from the
point of view of literature classification.
III. ANOVE RVIEW OF 5G MMWAVE COMMUNICATIONS
FO R UAV-AS SI ST ED WIRELESS NE TW ORKS
In this section, we will present a brief overview on 5G
mmWave communications for UAV-assisted wireless network-
s. Specifically, we will first introduce the key technique advan-
tages and challenges of 5G mmWave communications. Then
the integration of 5G mmWave communications into UAV-
assisted wireless networks will be briefly discussed from two
aspects, i.e., potential applications as well as main technical
advantages and challenges.
A. Key Technical Advantages and Challenges of 5G MmWave
Communications
1) Technical Advantages: The mmWave communications
will play an important role in 5G and B5G wireless systems,
due to the ability to support orders of magnitude increases
in network capacity. In what follows, we will briefly dis-
cuss the key technique advantages of 5G mmWave com-
munications from four perspectives, i.e., larger bandwidth
availability, shorter wavelength, narrow beam, and increased
security/interference immunity.
a) Larger Bandwidth Availability: The mmWave range in
the electromagnetic spectrum is usually considered to be the
band of spectrum from around 30 GHz to 300 GHz that lies
between the microwaves and the infrared waves, also called
the Extremely High Frequency (EHF) range. The abundance
of unoccupied bandwidth available at mmWave frequencies
is one of key advantages of 5G mmWave communications
compared to limited microwave spectrum resources below
6 GHz currently used by conventional wireless systems and
existing 4G LTE. Particularly, the available bandwidth in
mmWave frequency bands of 71~76 GHz and 81~86 GHz
(widely known as E-band) is more than the sum total of all
other licensed spectrum available for existing wireless systems
[81]. It is obvious that heavily utilized microwave spectrum
below 6 GHz for existing wireless systems is insufficient to
attain the Gigabit data transmission speeds. However, there
are still huge amount of lightly licensed and unused spectrum
resources available at mmWave frequencies reserved for future
applications around the world. The larger bandwidth translates
to extremely high data rates, easily achieving the peak data
rates of 10 Gbit/sor more with full duplex capability, which
is much greater than the limit rate of 1 Gbit/sby using lower
microwave frequencies [71]. This achievement of remarkable
data rate enhancement has been theoretically verified through
the Shannon-Hartley theorem showing that the capacity in-
creases linearly with the bandwidth.
In light of the advantage of larger bandwidth availability,
a multi-layer spectrum approach was proposed by Huawei,
based on the consideration of divergent requirements of 5G
services and different characteristics of related frequency
bands. The 24.25~29.5 GHz and 37~43.5 GHz mmWave fre-
quency bands were specifically selected for early deployment
of 5G mmWave systems in accordance with the 3GPP Re-
lease 15 [82]. To provide the full-duplex Gigabit Ethernet
connectivity at the data rates of 1 Gbit/sor higher, the E-
band (i.e., 71~76 GHz and 81~86 GHz ) was initially allocated
by the International Telecommunication Union (ITU) as early
as 1979, to support the ultra high capacity point-to-point
communications [83]. The ITU and 3GPP also launched a
plan for two phases of research for 5G standards, covering
IEEE ACCESS 9
19.05 mm
July 2018 October 2018
Fig. 4. The comparison of size scale between a one-cent coin and two
QTM052 mmWave antenna modules announced in July and October 2018,
respectively. (Image source: Qualcomm Technologies, Inc. [86]).
two mmWave frequency bands to be used with commercial
needs, i.e., 40 GHz and 100 GHz [84]. In 2015, the Federal
Communications Commission (FCC) proposed the flexible ser-
vice rules for the mmWave frequency bands including 28 GHz,
37 GHz, and 39 GHz for licensed usage, and 64~71 GHz
for unlicensed usage [85]. In addition to the ongoing efforts
from the regulators and industrial communities, some service
providers are all fighting to establish their 5G commercial
networks by using the extraordinary amount of bandwidth
available at mmWave frequencies. In many instances, AT&T
and Verizon have been focused on the 39 GHz and 28 GHz
mmWave frequency bands, respectively, for their initial 5G
deployments.
b) Shorter Wavelength: Within the range of the electro-
magnetic spectrum, higher frequency of electromagnetic wave
implies shorter wavelength of wave, which means that more
information can be transmitted per unit of time. Compared
to microwave signals below 6 GHz previously used by in-
creasingly crowded traditional cellular systems and WLANs,
mmWave signals have much shorter wavelengths that span
between 10 mm and 1 mm3. For example, mmWave signals
at 28 GHz,60 GHz, and 300 GHz have the extremely short
wavelengths of 10.7 mm,5 mm, and 1 mm, respectively. In
particular, the wavelengths of mmWave signals at 28 GHz and
73 GHz are nearly 10~30 times smaller than the wavelength
of microwave signal at 2.5 GHz [69]. Additionally, shorter
wavelengths for mmWave signals also bring about several
distinct benefits:
•Tiny component sizes: Compared with lower microwave
frequencies, the components and antennas for mmWave
frequencies can be designed in tiny physical dimension-
s owing to smaller millimeter-long wavelengths. Thus,
dozens to hundreds of antenna elements are easily packed
in an array of tiny size similar to fingernail, which makes
it feasible to miniaturize physically smaller circuits, mod-
ules, and equipments for 5G mmWave applications [50],
[52], [58], [69]. A typical half-wave dipole at 900 MHz is
152.4 mm long, but it is only 2.5 mm at 60 GHz in free
space or even less when built on a dielectric substrate
[87]. More recently, Qualcomm designed and developed
the world’s first fully-integrated 5G NR mmWave antenna
module QTM052 for next generation smartphones and
3Note that the range of the wavelengths between 10 mm and 1 mm exactly
corresponds to the mmWave frequencies between 30 GHz and 300 GHz.
Actually, the wavelengths of the mmWave signals may be slightly larger than
10 mm due to the adopted 28 GHz mmWave frequency band.
mobile devices. The antenna module packs a tiny phased
antenna array with a compact dimension, roughly the
size of a one-cent coin, as shown in Fig. 4. With these
smaller antenna modules, equipment manufacturers will
have more options for antenna placement, providing them
with more freedom and flexibility in the design of 5G NR
devices, e.g., more space for batteries and more choices
for smaller smartphones.
•Higher antenna gains: Based on the electromagnetics
and antenna theory, shorter wavelengths of mmWave
signals can obtain proportionally higher antenna gains
under the given effective antenna aperture. Because of
smaller physical dimensions at mmWave frequencies, the
application of large-scale multi-element antenna arrays
enables greater antenna gain and achieves highly di-
rectional beamforming. Through the configuration of a
pair of transmit antennas with the same aperture areas
and the equal transmit power, 30 dB more gain can be
achieved at 80 GHz mmWave beam compared to that of
2.4 GHz microwave frequency [70]. With the fully digital
processing, massive MIMO can be integrated to achieve
large beamforming gains [88]–[90].
c) Narrow Beam: By jointly combining digital precoding
and analog beamforming, shorter wavelengths of mmWave
signals enable the integration of massive MIMO with dozens
to hundreds of antenna elements to be packed into a s-
mall space at mmWave transceivers. The antennas can steer
mmWave beams rapidly in different directions. According
to the electromagnetics and antenna theory, the increased
frequencies result in the decreased beamwidth. As a result,
highly directional steerable narrow beams can be formed for
mmWave antenna arrays to direct the transmit power precisely
towards the intended users along a desired direction [91]. It
has been suggested that beam orientation of mmWave signals
can be optimized, aiming to reduce the energy loss during
transmission. Therefore, mmWave antennas with narrow di-
rectional beams can send and receive much more energy to
compensate for stronger propagation attenuation and higher
free space path loss of mmWave signals. These characteristics
make mmWave communications be more specifically suited
to ultra high capacity point-to-point transmission. Moreover,
another major advantage of the formed narrow beamwidth is
spatial reuse over the same spectrum to simultaneously serve
multiple users with different radios within a specific small
geographical area. It is further revealed that system capacity
has an apparent enhancement by employing mmWave antennas
with narrow beamwidth than the capacity by using wider
beamwidth antennas of microwave frequencies [92].
d) Increased Security and Interference immunity: In con-
trast to the congested spectrum and the easily intercepted
signals caused by wide beams at lower microwave frequen-
cies below 6 GHz, narrow beams of mmWave signals allow
the ability to obtain highly directional signals and greater
resolutions. When malicious eavesdroppers want to intercept
and decode the confidential messages successfully, they need
to physically be within the transmission path of mmWave
signals with narrow beams. With the help of complicated and
IEEE ACCESS 10
high-cost hardware equipments as well as accurate location,
eavesdroppers may have the possibility to intercept mmWave
signals. Obviously, the inherent benefits including highly di-
rectional beams and greater resolutions make the interception
of mmWave signals much more difficult and costly because
mmWave signals are only restricted to a relatively small area.
Therefore, the security of mmWave communications and the
privacy of mmWave users can be enhanced, by protecting the
sensitive and legitimate messages from passively eavesdrop-
ping and actively jamming [48], [59], [93]. Additionally, nar-
row beams also make mmWave transmission highly immune to
interference and noise at the receiving end due to the capability
to focus their transmit power levels of mmWave signals.
2) Technical Challenges: Despite many attracting tech-
nique benefits and potentials for 5G mmWave communica-
tions, there still have several challenges and limitations that
we should try to overcome for mobilizing mmWave.
a) Free Space Path Loss: The free space path loss is the
loss in signal strength of a signal in terms of radio energy when
it travels between the feedpoints of two antennas through free
space (i.e., unobstructed LOS channels in the air) [94]. The
signal strength of a signal can be measured as the transmitter
power output received by a receiving antenna at a transmission
distance from the transmitting antenna. We generally employ
the Friis transmission equation as follows to calculate the
power (denoted by PR) received from a receiving antenna with
gain GR, when transmitted from the transmitting antenna with
gain GT[95]:
PR
PT
=GTGRd−nλ
4π2
,(1)
where PTis the transmit power, λis the signal wavelength, d
is the transmission distance between the TX and RX antennas,
and nis the path loss exponent. For free space scenario, n= 2
[95]. We also note that nusually has different values that
depend on radio propagation channels for various complex
environments [96], e.g., n∈[2.7,4.0] for normal urban area
cellular radio and n∈[1.6,1.8] for indoor LOS scenarios.
According to (1), free space path loss can be expressed by
P LFS =4πd
λ2=4πdf
c2
, where fis the signal frequency
and cis the speed of light. We can actually use P LFS to predict
the signal strength of a signal at a given distance dof interest.
For the typical applications of wireless communications and
networking, P LFS can be described through a convenient way
in unit of dB by adopting fin GHz and din km:
P LFS = 20 log10 (d) + 20 log10 (f) + 92.45.(2)
Based on (2), P LFS can be obtained in direct proportion
to signal frequency and transmission distance. Compared
to microwave signals below 6 GHz, free space path loss
is much higher for mmWave signals at higher frequencies
under the condition of the same transmission distance and
antenna configurations. With different transmission distances,
free space path loss at microwave and mmWave frequencies
is depicted in Fig. 5. As the signal frequency increases, free
space path loss will raise as well. From the figure, it is also
0 50 100 150 200 250 300 350
50
60
70
80
90
100
110
120
130
140
150
Frequency (GHz)
Free Space Path Loss (dB)
d = 0.02 km
d = 0.05 km
d = 0.1 km
d = 0.5 km
d = 1.2 km
1 5 10 15 20
60
80
100
120
Fig. 5. The free space path loss at microwave and mmWave frequencies
under different transmission distances.
clearly revealed that free space path loss will increase for
both microwave and mmWave frequencies with the growth
of transmission distance between the TX and RX antennas.
This demonstrates that the use of mmWave frequencies limits
the transmission distance, e.g., 60 GHz indoor short-range
broadband communications at 3.5 Gbit/sover a range of 5 m
[97]. However, to extend the communication distance, several
advanced solutions, e.g., distance-adaptive design at physical
layer, ultra-massive MIMO, reflectarrays, and HyperSurfaces,
have been proposed with the improved transmission distance
up to 100 m in both LOS and NLOS areas at 60 GHz [98].
Recent field trial jointly carried out by Huawei and NTT
DoCoMo has further attained the long-distance transmission
with 5G high data speed at 39 GHz in both stationary and
mobility scenarios [99]. In addition, the path loss of mmWave
NLOS links is more larger than the free space path loss of
mmWave unobstructed LOS links. For instance, based on the
measurement campaigns at 60 GHz, the NLOS path loss is
147.4 dB which is obviously higher than the LOS free space
path loss with the value 119.3 dB under the transmission
distance 200 m [100].
b) Atmospheric Attenuation: The free space path loss just
reflects only a kind of signal attenuation which occurs while
traveling in the ideal vacuum environment. But beyond that,
mmWave signals traveling in free space are also effected by
frequency-related atmospheric attenuation. In principle, the
propagation of mmWave signals is typically influenced by
interaction of atmospheric molecules in the form of oxygen,
water vapor, rain, fog, cloud, etc, within the Earth’s atmo-
sphere. The atmospheric attenuation is primarily caused by
the vibrating attribute of atmospheric molecules when they
interact with mmWave propagation. More precisely, these
molecules can absorb a certain portion of signal energy of
mmWave propagation and vibrate with a strength proportional
to signal frequency [70], [101], [102]. The atmospheric effect
below 10 GHz is relatively low and can be also measured
by using the Friis transmission equation [103], however at-
IEEE ACCESS 11
0 50 100 150 200 250 300 350
Frequency (GHz)
10-3
10-2
10-1
100
101
102
Specific Attenuation (dB/km)
60GHz
120GHz
180GHz
Water Vapor Density = 7.5 g/m3
Water Vapor Density = 0.0 g/m3
Fig. 6. The specific attenuation caused by atmospheric oxygen and water
vapor at microwave and mmWave frequencies under whether there has water
vapor with the density of 7.5 g/m3or not. Here, atmospheric pressure P=
101.325 kPa, temperature T= 20°C, and distance d= 1 km.
0 50 100 150 200 250 300 350
Frequency (GHz)
10-5
10-4
10-3
10-2
10-1
100
101
102
Rain Attenuation (dB/km)
Light Rain, Rain Rate = 1.0 mm/h
Moderate Rain, Rain Rate = 4.0 mm/h
Heavy Rain, Rain Rate = 25.0 mm/h
Intense Rain, Rain Rate = 50.0 mm/h
1 5 10 15 20
0
2
4
6
Fig. 7. The rain attenuation at microwave and mmWave frequencies under
different rainfall intensities in terms of rain rates. Here, distance d= 1 km.
mospheric attenuation for mmWave frequencies and higher
increases significantly, specially at certain frequencies. Hence,
atmospheric effects on mmWave propagation not only restrict
the transmission distance of mmWave signals, but also affect
the utilization of 5G mmWave communications.
With the varying frequencies from 1 GHz to 350 GHz, at-
mospheric attenuation per kilometer generated by atmospheric
oxygen and water vapor at sea level is displayed in Fig. 6. We
can observe from this figure that atmospheric gases make the
signal attenuation very high at mmWave frequencies, especial-
ly for the water vapor density of 7.5 g/m3. Moreover, there are
three attenuation peaks of atmospheric absorption in the fre-
quency bands approximately 60 GHz,120 GHz, and 180 GHz,
wherein the attenuation reaches the maximum values, namely,
14.65 dB/km,2.0 dB/km, and 27.77 dB/km, respectively.
Specifically, atmospheric oxygen absorption is particularly
high at 60 GHz and 120 GHz, and water vapor absorption
is especially high at 180 GHz. These attenuation peak points
are caused by absorption resonances of atmospheric molecules
such as oxygen and water vapor at those frequencies, which is
similar to resonant oscillations in nonlinear isolation systems
[104].
In addition to atmospheric absorption by oxygen and water
vapor, rain attenuation can be regarded as the most signifi-
cant propagation impairment for mmWave frequencies. The
raindrops are approximately the same order of the size as the
wavelengths of mmWave signals. Therefore, the rainfall causes
additional attenuation due to the scattering and absorption
of electromagnetic waves by rain particles. Theoretically,
rain attenuation has a bearing on raindrop shape, raindrop
size distribution, rain rate, signal polarization and frequency,
etc [102], [105], [106]. The impact of rainfall intensity on
rain attenuation per kilometer at sea level under the varying
frequencies between 1 GHz and 350 GHz is further shown
in Fig. 7. As can be observed, rain attenuation at mmWave
frequencies is much higher than that of microwave frequencies.
On the other hand, as the rain rate increases, the attenu-
ation will also become greater. Despite the inevitable rain
attenuation, there are also a number of frequency windows
wherein atmospheric attenuation is significantly smaller. In
consequence, these windows at mmWave frequencies provide
the opportunity for mobilizing mmWave.
c) Blockage Effect: Although mmWave communications
own the capability to provide ultra-high-speed indoor wireless
transmission, wireless backhaul, small cell access, cellular
access, etc, mmWave bands are not robust enough and the
system performance degrades significantly due to the sensitiv-
ity of mmWave links to blockages, e.g., static, dynamic, and
self-blockage [107]. Typically, the propagation of mmWave
signals is rather susceptible to blockage effect due to severe
penetration loss and higher reflection/diffraction loss [107]–
[109]. These losses of mmWave signals are caused by different
materials or surfaces of the physical obstacles and even foliage
penetration while traveling in their propagation paths blocked
by the obstacles. It should be admitted that several factors
including shape, dimension, and material type of the obstacles
have a dramatic impact on blockage effect at mmWave bands.
In many instances, the attenuation of mmWave signals at
60 GHz for drywall with the thickness of 2.5 cm and mesh
glass with the thickness of 0.3 cm reaches 6.0 dB and 10.2 dB,
respectively [108]. Even more serious, mmWave signals can-
not penetrate most solid materials compared to microwave
signals. With respect to the duration of blockage with 0.2 s,
the blockage of human body can result in the reduction of
signal strength of mmWave signals at 60 GHz by 20 dB,
much greater than at microwave frequencies [110]. As a
consequence, frequent blockages of mmWave LOS links and
large blockage duration lead to performance degradation of 5G
mmWave communications. Fortunately, traveling signals still
have the alternative ways to reach the RX side by the aid of
diffraction, reflection and scattering, in spite of the blockages
of mmWave LOS links [111].
d) Beam Misalignment: The narrow beams of mmWave
IEEE ACCESS 12
signals allow the ability to achieve highly directional beams
and greater resolutions along the desired directions. We can
utilize the highly directional antennas to focus their power of
steerable narrow beams on the intended users. With the large-
scale antenna arrays, highly directional beamforming can be
leveraged to compensate for stronger atmospheric attenuation
and higher free space path loss at mmWave frequencies. In
order to fully take advantage of highly directional mmWave
transmission or reception, the beams between TX and RX
sides need to be well steered and aligned [112]–[115]. Efficient
beam alignment policies, e.g., beam tracking, beam training,
hierarchical beam codebook design, accurate estimation of the
channel, etc, are required to reduce the delay overhead with
or without a priori knowledge of the TX and RX locations
[116]–[119].
Nevertheless, due to the decrease of beamwidth of mmWave
signals and the mobility of UEs, it will become more and more
difficult to effectively align the beams between TX and RX
sides [112], [120]. To this end, beam misalignment between
TX and RX sides is unavoidable from practical view points.
The misalignment of beams not only reduces the probability of
successful transmission and reception for TX and RX sides,
but also degrades the network performance, such as severe
delay spread, reduced system throughput, and significant loss
of beamforming gain. Based on the measurements, beam mis-
alignment can cause additional propagation loss of mmWave
links that is almost symmetric with respect to the misaligned
angle [121]. For instance, in the mmWave system with the
beamwidth of 7°, the beam misalignment of 18° results in the
link budget drop by about 17 dB, which further generates the
maximum throughput reduction by up to 6 Gbit/sor leads to
the entire link interruption [122].
B. Integration of 5G MmWave Communications into UAV-
Assisted Wireless Networks
1) Potential Applications: As has been mentioned before,
the use of 5G mmWave communications can bring multiple
benefits including larger bandwidth availability, shorter wave-
length, narrow beam, increased security/interference immunity,
etc. Meanwhile, UAVs are being used to provide a promising
solution to reliable and cost-effective wireless communications
from the sky. On the one hand, they can be deployed as
flying BSs to enable aerial access to ground UEs in target
areas. On the other hand, UAVs can serve as aerial relays
for partitioned ground UEs, and can even create multi-tier
architecture for FANETs or connected drones. As such, the
application of UAVs has been considered as an alternative
complement of legacy cellular systems, with the potentials
to achieve higher transmission efficiency, enhanced wireless
coverage, and improved system capacity. To meet the exponen-
tially growing data capacity demand of future 5G and beyond
wireless applications, an intuitive way is to integrate UAV-
assisted wireless networking with mmWave communications.
With this integration of two promising techniques, three kinds
of typical applications can be summarized as follows:
a) Aerial Access via A2G/G2A MmWave Links: As previ-
ously introduced, UAVs can be used to provide aerial access
from the sky to ground UEs in a certain area of interest. As
summarized in Table I, we can deploy the UAVs as many
forms of aerial access equipments, e.g., aerial BS, aerial RSU,
aerial data collector, and even aerial MEC server and caching,
to achieve on-the-fly wireless access. To keep up with the
exponentially growing data capacity demand, a promising idea
is to link aerial access in UAV-assisted wireless networks with
mmWave communications together. Particularly, bidirectional
mmWave links are allowed to be properly designed for aerial
access applications in our considered scenario. To be precise,
depending on the transmission direction between UAVs and
ground UEs, bidirectional mmWave links consist of the A2G
links (i.e., downlink) and the ground-to-air (G2A) links (i.e.,
uplink). In many instances, typical applications for aerial
access using A2G and G2A mmWave links include downlink
NOMA transmission to serve ground UEs simultaneously
[123], flying cache-enabled remote radio heads (RRHs) by
deploying cache storage units at UAVs [37], flying BSs
with multiple transmit antennas to provide wireless access to
ground legitimate receivers [124], etc.
b) Aerial Relay and Interconnection via A2A MmWave
Links: UAVs can further serve as aerial relays to provide
wireless connectivity towards ground UEs without direct trans-
mission links. Given this scenario, data packets from source
UE are first sent to an initial UAV via aerial access, and
then transferred to other UAVs in a multi-hop fashion by
means of UAV interconnection in FANETs or directly to
destination UE through UAV relay using the load-carry-and-
deliver forwarding mechanism. To meet the requirement of
the sustainable increase of transmission capacity and data
traffic, it will be far more realistic to apply higher mmWave
frequencies instead of lower microwave frequencies to design
the A2A communication links for both aerial relay and multi-
hop interconnection. Additionally, as exhibited in Fig. 1, in-
formation exchange and transfer among the UAVs can be also
achieved through multi-hop drone interconnection based on
the A2A mmWave links. By using multiple antenna arrays to
obtain multiple beams, each UAV can establish the directional
mmWave links with its neighboring UAVs in the mmWave
flyMesh architecture [32].
c) Aerial Backhaul via A2G/G2A MmWave Links: With
the ultra dense small cell deployment in 5G wireless systems,
the backhaul between BS and core network is traditionally
connected by optical fiber or coax in terrestrial networks,
to support multiple Gigabit data transmissions. However, it
will not be flexible, easier to deploy, and cost-effective for
massive deployment of small cells [49], [125]. In comparison
with wired backhaul, UAVs can serve as aerial BSs and
RRHs to provide wireless backhaul due to flexibility, ease of
deployment, and lower operating and maintenance costs. With
the heavy data traffic burdens for small cells, UAV based aerial
backhaul via A2G/G2A links by using mmWave frequencies
has been proposed as a potential solution to attain Gigabit
data transmission speeds. Moreover, ultra-high-speed wireless
backhaul by integrating UAVs and mmWave communications
will further improve the flexibility and apparently reduce
the costs of wired backhaul in terrestrial networks [8]. For
instance, the mmWave flyMesh has been proposed to provide
IEEE ACCESS 13
ultra-high-speed wireless backhaul between mmWave flyMesh
and relay BS [32].
2) Brief Introduction to Technical Advantages and Chal-
lenges: Inspired by the technique benefits and potentials
for 5G mmWave communications, UAV-assisted wireless net-
works with mmWave communications still have some inherit-
ed merits:
•Multiple Gigabit data transmission speeds can be guaran-
teed for UAV-assisted wireless networking with mmWave
communications, due to the abundance of unoccupied
bandwidth available at mmWave frequencies. To be pre-
cise, the peak data rate of 10 Gbit/sor more can be
easily achieved with the help of full-duplex Gigabit
A2G mmWave connectivity. This attracting peak data
rate is much greater than the limit rate of 1 Gbit/s
by using traditional wireless technologies operating at
lower microwave frequencies. A large amount of licensed
and unlicensed mmWave frequency bands are potentially
available for usage in UAV-assisted wireless networks
including 28 GHz licensed band and 60 GHz unlicensed
ISM band [57]. From the perspective of policymakers,
FCC has adopted rules for wireless broadband operations
at frequencies above 24 GHz [126], and approved the
unlicensed V-Band use of 57~64 GHz mmWave bands to
enhance wireless backhaul [127]. These mmWave bands
will be the candidate for A2G and G2A links in UAV-
assisted wireless networking scenarios.
•Much shorter mmWave wavelength (ranging form 1 mm
to 10 mm) makes it feasible to miniaturize physically
smaller circuits, modules, and equipments for mmWave
applications in UAV-assisted wireless networks. In par-
ticular, hundreds of mmWave antennas can be easily
integrated as a beamforming array on a chip in a tiny
component size [50], [52], [58], [69]. This advantage
makes mmWave antenna arrays and other related modules
and equipments very suitable for low-altitude short range
UAVs with limited payloads. Thereby, we can offer more
choices for placing antennas, and more freedom and flex-
ibility in system design and optimization for computing,
memory, and communications of UAV payloads.
•Large-scale multi-element antenna arrays can be poten-
tially applied into UAV-assisted wireless networking with
mmWave communications. It has been demonstrated that
greater antenna gain and highly directional 3D beamform-
ing will be easily designed and achieved under this sce-
nario [10], [128], [129]. By adopting the 3D beamform-
ing, the beamforming gain can be markedly concentrated
in the target area due to ubiquitous access from the sky
[128]. The spatial beamforming gain enhancement can
effectively combat stronger atmospheric attenuation and
higher free space path loss at mmWave frequencies. This
will bring better system performance in UAV-assisted
wireless networks with mmWave communications.
•Highly directional signals and greater resolutions can be
achieved owing to narrow beams of mmWave signals by
integrating mmWave communications into UAV-assisted
wireless networks. On the one hand, highly directional
transmission that is similar to optical-like propagation
can enable low interference and increased security [59].
It will further help to protect the sensitive and legitimate
messages for ground UEs from passively eavesdropping
and actively jamming, especially for A2G mmWave links
under the scenario of aerial BSs enabled cellular sys-
tems. On the other hand, narrow directional beams can
concentrate much more wireless energy on target area,
to compensate for stronger propagation attenuation and
higher free space path loss of mmWave signals.
Although there exist many attracting advantages, mmWave
communications also pose several technique challenges and
limitations towards the integration of mmWave communica-
tions with UAV-assisted wireless networking:
•Propagation behavior and characteristics at mmWave fre-
quencies, i.e., severe atmospheric absorption and higher
path loss, have restricted the transmission distance for all
the potential application scenarios as mentioned above.
This requires us to consider short distance of mmWave
links for UAV deployment no matter in aerial access or
aerial backhaul. In addition, for actual system design, we
should avoid using three attenuation peaks of atmospheric
absorption in the frequency bands approximately 60 GHz,
120 GHz, and 180 GHz. Within these mmWave bands,
atmospheric attenuation can reach the maximum values,
which may further affect the system performance. This
in turn tells us that we can take full advantage of several
frequency windows with relatively smaller atmospher-
ic attenuation, to achieve mmWave communications in
UAV-assisted wireless networks.
•It has been highlighted that the propagation of mmWave
signals is more susceptible to blockage effect along the
propagation channel because of severe penetration loss
and higher reflection/diffraction loss as stated previously.
Compared with terrestrial static BSs, UAV-based flying
BSs are characterized by dynamic time-varying positions
and altitudes. This kind of high flexibility for UAV
deployment together with the use of mmWave frequency
bands make mmWave signals more vulnerable to the
blockage problem. The typical blockage primarily in-
cludes human-body blockage incurred by human activity
[62]–[65] and self-blockage caused by components of
UAV itself [59]. The self-blockage of mmWave signals
along the propagation channel is especially related with
the scenario of UAV-assisted wireless networks, due to
the components of UAV itself, e.g., the rotating propeller.
Despite the impact of self-blockage, UAV-assisted wire-
less networks are more practical and applicable towards
mmWave transmission owing to the easily established
A2G LOS links.
•UAV-based aerial BSs can provide ubiquitous access from
the sky owing to the typical characteristic of sufficient
flexibility and high mobility with dynamic time-varying
positions and altitudes. In addition, narrow mmWave
beams and possible mobility of ground UEs make it very
difficult to properly steer and align the highly directional
beams for mmWave antenna arrays at the side of UAVs
or ground UEs [10], [32]. Apparently, it will be far
IEEE ACCESS 14
Fig. 8. The classification of antenna technique solutions for UAV-assisted
wireless networks with mmWave communications.
more realistic to dynamically adjust directional beams
according to current instant time in practical dynamic en-
vironment. As such, achieving dynamic beam alignment
between UAVs and ground UEs with lower overhead and
computational complexity will be incredibly complicated
and challenging.
IV. RESEARCH ISS UE S, C HALLENGES,AND
STATE-OF-THE-ART
From the perspective of the taxonomy of the state-of-the-
art research advances, we will give readers a comprehensive
summary of key topics on 5G mmWave communications
for UAV-assisted wireless networks in this section, including
antenna technique, radio propagation channel, multiple access
mechanism, spatial configuration, resource management, secu-
rity strategy, and performance assessment.
A. Antenna Technique
As an important component of UAV-assisted wireless net-
works with 5G mmWave communications, the antenna acts
as wireless interface to transmit and receive mmWave signals
for UAVs and ground UEs. In this regard, the antennas can
be figuratively described as the “eyes and ears” of advanced
wireless communication systems. Due to the extension from
conventional microwave frequency bands to mmWave fre-
quency bands, there will be much higher propagation at-
tenuation and severe free space path loss under the same
communication distance and antenna gains. Compared with
traditional user devices, high mobility and physical structures
of UAVs also pose extra challenges on accurate tracking of
dynamic mmWave beam directions. Thus, the design of high
gain and high-efficiency antennas is of great importance for
realization of UAV-assisted wireless networks with mmWave
communications. As shown in Fig. 8, recent efforts and
solutions on the antenna-related research including antenna
design [130]–[133], beam tracking [10], [32], [134], [135], and
beam optimization [136] will be summarized in the following
subsection.
1) Antenna Design: The success deployment of UAV-
assisted wireless networks with 5G mmWave communications
has given fresh impetus to the development and design of
wireless antennas, complying to the requirements of mmWave
frequency bands, propagation characteristics, circuit compo-
nents, gain rating, geometries of mounting devices, transceiver
architecture, communication system performance, and so on.
Based on the state-of-the-art solutions about antenna design in
our considered scenario, we will present the existing research
efforts on the design issues and solutions of two representative
antennas, i.e., array antennas and beam antennas.
Fig. 9. Illustration of the top view of the devised 1×4star shaped antenna
array.
a) Array Antenna: The use of UAVs flying at a high
altitude of about 20 km to build up the high-altitude platform
station (HAPS) system and the stratospheric-platform (SPF)
system has been recognized as a low-cost and flexible method
for future wireless infrastructure. Compared to the outermost
layer of the atmosphere in which the satellites orbit the earth,
the deployment of UAVs at the layer of the atmosphere
has many advantages, such as free-space like channel, lower
transmit power similar to terrestrial systems, reduced prop-
agation delay, etc. Based on this insight, Tsuji et al. [130]
conducted the experiments to test the performance of digital
beamforming (DBF) antenna and multi-beam horn (MBH)
antenna for mmWave frequency band under the stratospheric
conditions. To be specific, as for array antenna design, the
authors proposed an electronically controlled mmWave DBF
antenna for HAPS. In particular, this kind of array antenna
was mounted on the helicopter, and was operating under
the stratospheric conditions where the temperature was below
−60 °C and the atmospheric pressure is 1/20 of that on the
earth. Based on this setting, the performance of the devised
mmWave DBF antenna under the stratospheric condition-
s was tested to examine the optimal forming of antenna
beams for mmWave SDMA. More precisely, the contents of
the experiments included beam-tracking performance, beam
compensation, measurement of antenna beam patterns, and
transmission of data packets. According to the experimental
results, despite the harsh conditions of low temperature and
low pressure, mmWave DBF antenna system can still work
properly.
Bandwidth enhancement and miniaturization for microstrip
antenna are one of the major challenges towards the design of
conventional compact antenna. Especially, for the application
scenario with mmWave communications, tiny component size
of mmWave circuits and increased bandwidth result in lower
antenna gain, which can be improved by building large antenna
arrays. Under this background, the dual-band mmWave printed
microstrip antenna patch was designed by Siddiq et al. [131]
for UAV applications. The devised antenna array with coaxial
feed resonated at 29~30 GHz and 57~66 GHz mmWave bands.
Beyond this, the shape of the structure of UAV’s wing was
based on the size of 1250 mm ×350 mm ×50 mm by using
the star shaped printed microstrip antenna. Specifically, the
authors put forward two kinds of star shaped antenna arrays,
i.e., 1×2array and 1×4array. Fig. 9 illustrates the top view
of the 1×4star shaped antenna array with core parameters.
It should be admitted that antenna gain can increase for both
antenna arrays with the growth of frequency, and 1×4array
IEEE ACCESS 15
can achieve more gain than that of 1×2array.
In addition to the advantage that the positions and altitudes
are dynamically adjusted to match the requirement of ground
users, UAV-based flying BSs can be also deployed with
the characteristic of directional mobility and quasi-stationary
hovering. How to improve the directivity of flying BS and
facilitate better beam-steering resolution and lower cross-beam
interference is an important issue we need to resolve. Fortu-
nately, phased array antenna is a good choice to tackle this
problem. Huo and Dong [132] proposed a uniform rectangular
based phased array beamforming model for flying BSs with
mmWave communications. The proposed phased array antenna
employed a dimension of 2×8antenna elements to realize
the 3D beamforming. In order to provide a broad coverage
towards ground users, the elevation radiating direction of
main lobe (i.e., main beam) was required to host sufficient
wide beamwidth, and the azimuth radiation pattern of main
lobe should require narrow beamwidth. Through the field trial
tests, it is observed that the measurement results for both
single user and multiple users can obtain multiple Gigabit data
transmissions over flying BSs with mmWave communications.
b) Beam Antenna: In the context of UAV-enabled HAPS
system and SPF system with mmWave communications, Tsuji
et al. [130] also developed an mmWave MBH antenna that can
achieve high-speed transmission at 47 GHz and 48 GHz. To
be specific, the antenna direction of mmWave MBH antenna
was devised to own three control modes: i) fixed mode: MBH
antenna cannot adjust the direction; ii) inclination adjustment
mode: the direction of beam was adjusted as the helicopter
changes its inclination; iii) inclination/position adjustment
mode: the footprint can be controlled in the same spot.
Furthermore, mmWave MBH antenna was demonstrated to
possess several advantages in controlling the antenna beams,
such as broad bandwidth, low development cost and low power
consumption. Similar to the experiments for mmWave DBF
antenna in [130], MBH antenna was also installed on the
helicopter to test its performance over the HAPS system. The
measurement mission included beam tracking, beam pattern,
and quality of communication signals. Finally, experiment
results showed that the system can work properly and further
demonstrated that highly directional beams can solve the dis-
advantage of UAV-assisted wireless networks, which becomes
a major component of mmWave UAV communications for 5G
and beyond. The main features of the devised antennas in [130]
are summarized in Table IV.
With the help of 3D beamforming and tracking antennas,
the advantage of using highly directional mmWave antennas
at TX and RX sides is that a continuously maintained antenna
gain can be obtained, although there exist higher free space
path loss and severe atmospheric attenuation caused by higher
mmWave frequencies. Motivated by that, Heimann et al. [133]
performed an experimental verification by implementing the
mmWave link at 28.5 GHz with the bandwidth of 800 MHz
between a fixed ground station equipped with an active antenna
(i.e., pencil beam antenna) and a lightweight UAV equipped
with a passive antenna (i.e., lightweight horn antenna). As
shown in Fig. 10, a mmWave TX-RX scheme was designed
UAV
MmWave TX MmWave RX
Pencil Beam
Antenna
Optical Reference
System
Pencil Beam
Horn Antenna
Ground Station
Fig. 10. The experiment configuration for verifying the mmWave link at
28.5 GHz with the bandwidth of 800 MHz. Here, the mmWave TX and RX
connect the UAV and the ground station, respectively.
by sending from the UAV based platform to the static in-
frastructure side including ground station and mmWave RX.
From Fig. 10, the optical reference system was used to obtain
the accurate position and orientation information of UAV for
tracking performance evaluation. Based on this setup, the
author carried out three tests: i) pencil beam alignment in static
UAV environment, ii) precise tracking of UAV in dynamic
environment, and iii) effect of the tracking precision. Based
on the insightful discussions and tests, the authors analyzed
the performance of beam tracking in UAV communications
at mmWave band according to signal strength, quality and
throughput for different antenna tapers, tracking algorithms
and mobility patterns.
2) Beam Tracking: Beamforming, also known as spatial
filtering, is a signal processing technique that has been already
used in Wi-Fi routers with multiple antennas, to dramatically
improve Wi-Fi performance, smoothing the path to the next
generation IEEE 802.11ac standard and beyond. In practice,
beamforming technology concentrates the signal energy over
a narrow beam by means of highly directional signal trans-
mission or reception to improve the spectral efficiency. By
applying the beamforming technology into mmWave commu-
nications, it is efficient to compensate for higher free space
path loss and stronger atmospheric attenuation at mmWave
frequencies [137]. Evidently, signal interference will be severe
when many people gather in small space to access wireless
networks. Fortunately, beamforming technology achieves the
equivalent “bunching” wireless signal into a rope, which can
effectively reduce the cumulative signal interference to obtain
better signal-to-interference-plus-noise ratio (SINR) at RX.
However, due to the fast-moving objects (e.g., vehicles,
drones, and trains) under the context of 5G mmWave com-
munications for UAV-assisted wireless networks, significant
challenge of beamforming application is how to achieve fast
and accurate tracking of dynamic mmWave beam directions
with low computational complexity and overhead. To deal
with this problem, Li et al. [134], [135] developed a recursive
analog beam tracking algorithm for tracking dynamic beams
in mmWave communications using analog beamforming. The
beam tracking error minimization problem was formulated as
a constrained sequential control and estimation problem by
IEEE ACCESS 16
TABLE IV
THE C OMPA RIS ON O F MAI N FEAT URE S BE TWE EN T HE DBF AN TEN NA AN D TH E MBH ANT EN NA.
Main Feature DBF antenna MBH antenna
Array configuration 16 elements (4×4square array) 3elements (TX , RX, respectively)
Bandwidth About 4 MHz >300 MHz
Type of antenna Microstrip antenna Corrugate horn
Antenna gain 15.7 dBi 28.4 dBi
MmWave frequency 28.2 GHz (TX), 31.1 GHz (RX) 47.2∼47.5 GHz (TX), 47.9∼48.2 GHz (RX)
Total weight 74.2 kg (TX+RX) 62.2 kg (TX+RX)
Beamwidth About 13 °C —
Number of beams Optional (3) (TX), Optional (3) (RX) —
Stratospheric conditions Low temperature (−60 °C), low pressure (1/20 on the earth) Low temperature (−60 °C), low pressure (1/20 on the earth)
jointly optimizing both the analog beamforming vectors and
the beam direction estimators. To resolve this problem, the au-
thors relaxed the dynamic scenario into a static beam tracking
scenario. In this scenario, the devised recursive mmWave beam
tracking algorithm was focused on two stages: i) coarse beam
sweeping, and ii) recursive beam tracking. Moreover, the lower
bound of the variance of beam direction estimator using the
mean square error (MSE) converges quickly to the minimum
Cramér-Rao lower bound by optimizing among all the analog
beamforming vectors. However, in the dynamic scenario, beam
direction maybe changes over time. In order to keep track of
the changing mmWave beam direction, the step-size parameter
αnwithin the n-th time slot was proved to be subject to:
αn=λ
√M(M−1) πd ,(3)
where λis the mmWave wavelength, Mis the number of
antennas at the linear antenna array receiver, and dis the
distance between neighboring antennas. Compared with three
reference algorithms (i.e., IEEE 802.11ad, least square, and
compressed sensing) through simulations, the proposed algo-
rithm simultaneously can achieve faster tracking speed, higher
tracking accuracy, low complexity, and low pilot overhead.
Due to high mobility of UAVs, the time to complete the
beamforming training is more stringent. In the meantime, the
overall mmWave beam search time is excessively costly for
the exhaustive search algorithm because of a large number
of candidate beam directions. To reduce the beamforming
training time and mmWave beam search overhead, Xiao et
al. [10] proposed to adopt a hierarchical beam search scheme
based on the tree-structured beamforming codebook which
covers the whole search space in angle domain. As shown
in Fig. 11, there are klayers in the typical tree-structured
codebook, and there are also Mkcodewords with equal
beamwidth and different steering angle in the k-th layer, where
Mis a positive integer which denotes the degree of tree-
structured codebook, for M≥2. It suffices to mention that the
mmWave beam search overhead of hierarchical beam search
scheme is greatly lower than the exhaustive search scheme.
On the basis of this advantage, the authors further designed
the hierarchical coarse codebook based on a binary tree like
structure wherein there are log2N+1 layers and the k-th layer
contains 2k−1best beams or known as antenna weight vectors
(AWVs), for k= 1,2,· ·· ,log2N[138]. We wish to remark
that the AWVs of the last layer in the coarse codebook can
hold the narrowest mmWave beam under the given number of
antennas N. It is also revealed that the beam coverage of the
AWVs of the last layer can be characterized as the sum of the
AWVs of the next layer.
In order to improve the efficiency of mmWave beam track-
ing and training, Xiao et al. [10] further pointed out two
important strategies: i) priori information with respect to the
distribution range of beamforming angles, and ii) hierarchical
tree structure of the beamforming codebooks. According to the
potential position relations between UAVs and ground UEs,
the distribution range of beamforming angles can be actually
predicted as a priori knowledge. Moreover, the candidate beam
directions can be also obtained in advance through mmWave
beam training process. Apparently, these strategies can also
significantly reduce the overhead for mmWave beam tracking
and training.
Owing to flexible configuration and mobility nature, UAVs
can be not only deployed as aerial BSs, but also serve as
aerial relays to form the flyMesh aiming at achieving a fully
connected UAV networks. However, as for the flyMesh with
mmWave communications, frequent movement of UAVs at
different altitudes causes the mmWave beam misalignment be-
tween UAVs or between UAV group leader and relay BS. Even
worse, one of the UAVs will not work properly, and the nearby
UAVs have to detect another one to establish the mmWave
links. To address this challenge, Zhou et al. [32] proposed
a fast mmWave beam tracking mechanism by employing the
beam tracking policy of IEEE 802.11ad and IEEE 802.11ay
standards, in order to find the suboptimal solution of the beam
alignment maximization problem in mmWave flyMesh. This
adopted mechanism can infer the directions of the relative
movement relations between UAVs or between UAV group
leader and relay BS according to the variations of SINR values
obtained by beam tracking. To be specific, it is noticeable
that the mmWave link for beam pair was not good enough
if the SINR-based link quality was below a given threshold.
Then the beam tracking initiator thus triggered the transmit
beam tracking in order to select the best beam as the new
transmit beam for subsequent data transmission. Through the
analytical derivation, the overhead of the proposed fast beam
tracking mechanism was further formulated as a function of
the beam offset angle and the beamwidth. Simulation results
indicated that the performance in terms of tracking overhead
of the proposed fast beam tracking approach was superior to
that of the method by using the IEEE 802.11ay standard.
3) Beam Optimization: In practice, intra-group interference
involving data transmission from other UAVs and mmWave
IEEE ACCESS 17
w
w w
w w w w
w
w w w w
w w
w
w
0
S
k
wwwwwwwwwwwwwww
nĂk
w k n
ĂĂ ĂĂ ĂĂĂĂ
Fig. 11. The hierarchical beam codebook for beamforming training under a tree structure, where w(k, n)stands for the n-th codeword in the k-th layer,
for n= 1,2,···,2kand k= 0,1,2,···.
T
d
1M
X
2
X
1
X
0
X
Wavefrontsignal
Fig. 12. Illustration of antenna structure with uniform linear arrays wherein
θis the AoA of the incoming wavefront signal, dis the distance between
two adjacent array elements, and Xmis the m-th antenna array element for
m= 0,1,2,···, M −1.
signal reflections and scattering has a bearing on the system
performance of mmWave UAV group communications. As a
consequence, it is of paramount importance to investigate the
performance of beam anti-interference through a kind of beam
optimization. In this context, Zhong et al. [136] elaborated
the beam interference through the beam optimization based
on window function and codebook design for the antenna
with uniform linear arrays (ULA), and also established the
interference model of mmWave UAV group communications.
Fig. 12 shows the specific structure of the adopted ULA
antenna under this scenario. On the basis of this structure of
the ULA antenna, the mmWave beam response towards the
incoming wavefront signal with the angle of arrival (AoA)
denoted by θwas calculated as follows:
p(θ) = w(θ)α(θ) =
M−1
X
m=0
wmej2π
λmd sin θ,(4)
where w(θ)is the mmWave beam weight vector, α(θ)is the
array response for the AoA, Mis the number of the antenna
array elements, wmis the beam weight of the m-th antenna
array element, and λis the mmWave wavelength.
As for the beam optimization based on window function, the
target is to suppress the side lobes by designing the mmWave
beam weight vector w(θ)of weighting functions. Through the
window function optimization, the sidelobes of the incoming
wavefront signal can be effectively inhibited and the main lobe
gain can be also been greatly improved. As for the beam
optimization based on codebook design, Zhong et al. [136]
further presented an improved N-phased codebook design
scheme to optimize the weight vector w(θ). Lastly, simulation
results suggested that the phase adjustment of the beam can
be realized by using this codebook design and the sidelobe
interference by using N-phased codebook was greatly smaller
than that of IEEE 802.15.3c codebook.
4) Summary and Lessons Learned: Due to the important
role of antenna in 5G mmWave communications for UAV-
assisted wireless networks, we have reviewed the recent efforts
and solutions about the antenna-related research from three
perspectives, i.e., antenna design, beam tracking, and beam
optimization. With relation to antenna design, array antenna
and beam antenna have been discussed, respectively. Based
on the classification of antenna design, key parameters and
main features for DBF antenna and MBH antenna have been
compared and summarized in Table IV. By considering the
narrow directional beams of mmWave signals, we have then
surveyed the existing beam-related issues in terms of beam
tracking and beam optimization. The important lessons learned
from the review of the antenna-related research issues are
summarized as follows:
•The design of array antenna typically benefits from the
feature of tiny physical dimensions, and the development
of beam antenna actually relies on the attribute of highly
directional mmWave beams. We can obtain higher anten-
na gain by increasing the number of antenna elements,
utilizing the 3D beamforming, and taking advantage of
the highly directionality of mmWave beams.
•The fast and accurate tracking of dynamic mmWave beam
directions is very important for successful transmission
and reception of data packets. Generally, beam tracking
mechanism and algorithm with low computational com-
plexity and overhead are required to maintain the beam
alignment and improve the system performance. To deal
with the interference problem of adjacent mmWave beam-
s, we can employ the window function and codebook
design to optimize the beams for mmWave UAV group
communications.
B. Radio Propagation Channel
The radio propagation channels in mmWave frequencies are
significantly different from the channels in lower microwave
frequencies due to shorter mmWave wavelengths. Besides, at-
mospheric absorption at the mmWave wavelengths is very seri-
ous, which further leads to higher propagation attenuation and
IEEE ACCESS 18
Fig. 13. The classification of channel modeling solutions for UAV-assisted
wireless networks with mmWave communications.
reduced transmission range. Also, the propagation channels of
mmWave signals are susceptible to the blockage problem that
affects the design of LOS propagation links. Correspondingly,
modeling the radio propagation channel and capturing the
propagation characterization of mmWave signals for UAV-
assisted wireless networks with 5G mmWave communications
are of paramount importance to design and optimization
of UAV-enabled A2G and A2A mmWave communications.
Moreover, mmWave channel sparsity in angular domain ne-
cessitates the efficient and accurate channel estimation and
tracking technique, aiming to obtain better and stable system
performance. Another significant technical challenge is how
to detect the blockage of mmWave propagation channel. In
the light of the above issues and challenges as already stated,
we will summarize these related studies in the subsection
in terms of channel modeling [10], [37], [57], [58], [64],
[65], [123], [124], [135], [139]–[146], channel estimation and
tracking [58], [142], [147], [148], and blockage detection and
countermeasure [59], [143].
1) Channel Modeling: To fully capture perfect knowledge
and understanding of actual propagation behavior and char-
acteristics of mmWave signals, several state-of-the-art efforts
have been made to solve major problems targeted at mmWave
channel measurement and modeling techniques under the en-
vironment of UAV-assisted wireless networks. Based on those
efforts, we will provide a detailed description and discussion
about recent advances in mmWave channel modeling solutions
for UAV-assisted wireless networks with mmWave commu-
nications from three categories, namely, propagation mea-
surement, empirical channel modeling, and analytical channel
modeling, as depicted in Fig. 13.
a) Propagation Measurement: Realistic propagation be-
havior and characteristics at mmWave frequencies can be
better captured by carrying out real-world measurement cam-
paigns in frequency domain or time domain over the target
environments. The commonly used channel measurement tools
include channel sounder and vector network analyzer [149].
As for the environment of UAV-assisted wireless networks
with mmWave communications, propagation measurements
generally focus more on communication types, operational
frequency bands, flight dynamics of UAV, propagation physical
scenarios, antenna configuration, TX/RX placement, channel
sounding process, etc.
To understand and analyze the characterization of A2G
mmWave channels for UAV communications, Khawaja et al.
[57] utilized the Remcom Wireless InSite ray tracing software
to conduct the ray tracing simulations in frequency domain for
capturing the behavior of A2G mmWave bands at 28 GHz and
60 GHz frequencies. In the A2G ray tracing simulations, four
different scenarios were adopted, i.e., urban, suburban, rural,
and over sea. Besides, three key factors about the scatterers
(i.e., number, material, and height) were also incorporated into
the involved scenarios of the simulations. Under this setting,
the authors analyzed two performance metrics, namely, the
received signal strength (RSS) and the root mean square delay
spread of multipath components, for mmWave frequencies
of 28 GHz and 60 GHz, respectively. It is observed that the
fluctuation rate of the RSS versus the distance between UAV
and ground station at 60 GHz was higher than that of 28 GHz.
Furthermore, the behavior of root mean square delay spread of
multipath components highly depended on the height of UAV
along with the density/height of the scatterers around UAV.
On the basis of the A2G ray tracing simulations, Khawaja
et al. [57] presented a preliminary framework of A2G channel
sounding for mmWave UAV channels. To be even more
concrete, a lightweight and compact A2G mmWave channel
sounder was built up based on the high-performance software
defined radio platform of the universal serial radio peripheral
(USRP) X310 and the 60 GHz TX/RX development system
PEM009-KIT from the product of Pasternack. The objective
for the channel sounder development was to more accurately
characterize the A2G mmWave propagation features via the
propagation measurements at 60 GHz frequency.
As the continuation of the work about the characterization of
A2G mmWave channels for UAV communications, Khawaja et
al. [139] also performed the ray tracing simulations by taking
advantage of the Remcom Wireless InSite ray tracing software
to study the small-scale temporal and spatial characteristics of
A2G mmWave LOS channels at 28 GHz frequency. For the
simulations, four different scenarios including dense-urban,
suburban, rural, and over sea, were further created and classi-
fied in terms of the number, distribution and dimensions of the
buildings. Especially, the vertically polarized half-wave dipole
antennas were mounted at both TX and RX with the omnidi-
rectional pattern in azimuth direction. In the context of simula-
tion setup, the ray tracing results for the UAV trajectories were
characterized over sea and dense-urban scenarios. Moreover,
the power variation of multipath components was observed to
be dependent on the scatterer properties and the RX sensitivity.
Meanwhile, the small-scale temporal and spatial characteristics
of A2G propagation channels were validated to be subject to
the constraint of the scatterer properties.
To study the impact of human-body blockage on mmWave
links of flying BSs with mmWave communications as depicted
in Fig. 14, Gapeyenko et al. [64] used the distance-based
path loss model to characterize the A2G mmWave LOS and
NLOS links. Conceptually, this kind of path loss model can be
deemed to satisfy a standard linear model with respect to the
IEEE ACCESS 19
R
h
B
h
B
g
D D D
, ,
x
y h
Fig. 14. Illustration of human-body blockage on mmWave links of a flying
BS with mmWave communications. Here, the flying BS is currently located
at a 3D position point (xD, yD, hD), and user 1 and user 2 are blocked by
two human-body blockers. The user is located at the height of hR, and the
blocker is located at the height of hBand the diameter of gB.
Euclidean distance between flying BS and RX side. In order
to determine key parameters of this estimated model (i.e., the
least square fits of floating intercept and slope over the given
distance), Akdeniz et al. [140] carried out the real-world mea-
surements of mmWave outdoor cellular propagation at 28 GHz
and 73 GHz frequencies in New York City, NY. Particularly,
extensive measurements were performed by employing the
highly directional horn antennas at both TX and RX sides
under microcellular type deployments, aiming to obtain both
the bulk path loss and the spatial structure of mmWave chan-
nels. What’s more, the power measurements were performed
at various angular offsets from the strongest angular locations.
Through the measurements, the authors in [140] provided a
scatter plot of the approximate omnidirectional path losses
which can be easily observed as a function of the distance
between TX and RX. Table V gives the list of key parameter
values in the estimated path loss models through the realistic
measurements in [140]. From the results of measurements, it
should be also admitted that the theoretical free space path
loss based on the Friis transmission equation can exhibit a
good fit for the measurement data of mmWave LOS links.
b) Empirical Channel Modeling: Actual radio propa-
gation behavior of the channel in realistic environments is
generally too complex to model accurately. Nevertheless,
empirical channel models can be developed based on a large
amount of statistical data collected from the in-situ channel
observations and measurement campaigns. In what follows,
we will give a review of empirical channel modeling for UAV-
assisted wireless networks with 5G mmWave communications
from the perspective of fading types for mmWave propagation
channel.
Small-scale fading is characteristic of radio propagation
resulting from the presence reflectors and scatterers that
give rise to multiple components of transmitted signal while
traveling [150]. At RX, the combination of these different
components through multipath propagation may cause rapid
amplitude fluctuations of the received signal over a small
travel distance or time interval. To characterize the small-
scale fading, statistical models have been used to describe
the empirical fading distribution of amplitude of the received
signal along the radio propagation channel, e.g., Nakagami-
mmodel, Rayleigh model, Rician model, Weibull model, etc
[151]. The probability density function (PDF) is generally
employed to give the quantitative analysis about this kind of
fading distribution.
Under the scenario of UAV-based flying BSs with mmWave
communications, Zhu et al. [124] modeled the amplitude of
the received A2G mmWave signal as Nakagami-mfading
distribution for both the LOS and NLOS propagation con-
ditions at mmWave frequency bands. Based on this A2G
channel modeling, their target was to explore the secrecy
rate performance at physical layer for UAV-enabled wireless
networks using mmWave communications. The Nakagami
fading parameter mis a shape factor of Nakagami distribution,
which can be specifically expressed by m=(E[R2])2
/Var[R2],
where R≥0is the amplitude of received signal. Especially,
parameter mcan be empirically estimated, for m≥1/2. In
[124], the authors used mLand mNto describe Nakagami-
mfading model for mmWave LOS link and mmWave NLOS
link, respectively.
Rayleigh fading is regarded as a reasonable statistical model
to describe the distribution of the envelope amplitude of
received signal, which is made up of the multipath reflected
and scattered waves as well as the significant LOS component.
Under the scenario of next generation drone-assisted HetNets
with mmWave communications, Meng et al. [141] character-
ized the radio propagation channel for information exchange
between the BSs and the lead drone as Rayleigh fading model
at mmWave frequency bands. As a special case, Nakagami-
mfading model includes Rayleigh fading model when we set
m= 1. Thus, we can use a PDF f(R; Ω) to quantify Rayleigh
fading model, where Ω = ER2is the average fading power.
On the basis of this fading model, the authors focused on the
problem of drone swarm formation control for enhancing the
formation strength of drones during flight control.
In comparison with small-scale fading, large-scale fading or
shadowing is usually used to describe the average signal-power
attenuation and path loss of the received signal strength at RX
after traveling over a large travel distance [152]. Large-scale
signal-power attenuation caused by shadowing of obstacles has
been shown to follow a log-normal distribution. We generally
employ the path loss model P L (d)in dB to analyze this kind
of fading distribution, where dis the distance between TX and
RX.
Under the context of the proactive deployment of cache-
enabled UAVs in cloud radio access network (CRAN), Chen et
al. [37] formulated the mmWave propagation channel between
UAVs serving as the RRHs and ground mobile users as the
standard log-normal shadowing model using path loss metric.
IEEE ACCESS 20
TABLE V
LIS T OF KE Y PARA ME TER VAL UES I N TH E EST IM ATED PATH LO SS M ODE LS OB TAIN ED F ROM T HE RE AL IST IC M EAS UR EME NTS .
Path Loss Model Key Parameter Value at 28 GHz Key Parameter Value at 73 GHz
LOS: P LLOS (d) = αLOS + 10βLOS log10 (d) + χLOS,
χLOS ∼ CN 0, σ 2
LOS
αLOS = 61.4,βLOS = 2,
σLOS = 5.8 dB
αLOS = 69.8,βLOS = 2,
σLOS = 5.8 dB
NLOS: P LNLOS (d) = αNLOS + 10βNLOS log10 (d) + χNLOS,
χNLOS ∼ CN 0, σ 2
NLOS
αNLOS = 72,βNLOS = 2.92,
σNLOS = 8.7 dB
Case 1 (Combined two antenna
heights at RX):
αNLOS = 86.6,βNLOS = 2.45,
σNLOS = 8 dB
Case 2 (Only one kind of antenna
height at RX):
αNLOS = 82.7,βNLOS = 2.69,
σNLOS = 7.7 dB
TABLE VI
THE COMPARISON OF EMPIRICAL CHANNEL MODELS FOR UAV-A SSI STE D WI REL ES S NET WO RKS W IT H MMWAVE COMMUNICATIONS.
Reference Scenario Communication
Type
Adopted Channel
Model
Fading
Type
Quantitative Description
[124] Aerial BS A2G (downlink:
UAV→ground RXs)
Nakagami-m
fading model
Multipath
fading:
small-scale
fading
PDF:f(R;m, Ω) = 2mmR2m−1
Γ(m)Ωmexp −mR2
Ω,
where R≥0is the amplitude of received signal, and
Ω = ER2is the average fading power, for Ω>0.
Key parameter: For mmWave LOS link, Nakagami fading
parameter m=mL= 3; For mmWave NLOS link,
Nakagami fading parameter m=mN= 2.
[141] Swarm
Formation
Control
A2G (downlink:
UAV→terrestrial BS)
G2A (uplink:
terrestrial BS→UAV)
Rayleigh fading
model
Multipath
fading:
small-scale
fading
PDF:f(R; Ω) = 2R
Ωexp −R2
Ω,
where R≥0is the amplitude of received signal, and
Ω = ER2is the average fading power, for Ω>0.
Key parameter: —
[37] Cache-
enabled
flying RRH
A2G (downlink:
UAV→ground
mobile users)
Log-normal
shadowing model
Multipath
fading:
large-scale
fading
Path Loss for mmWave LOS link:
P LLOS
k,i dk,i=P L (d0) + 10nLOS log dk,i+χσLOS ,
Path Loss for mmWave NLOS link:
P LNLOS
k,i dk,i=
P L (d0) + 10nNLOS log dk,i +χσNLOS ,
where
dk,i =kwk−wik=q(xk−xi)2+ (yk−yi)2+h2
k
is the distance between the k-th UAV and the i-th ground
users, and P L (d0)is the free space path loss given by
20 log 4πd0fc
cwith the free-space reference distance d0,
the carrier frequency fc, and the speed of light c.
Key parameter: For mmWave LOS link, nLOS = 2 and
χσLOS = 5.3 dB; For mmWave NLOS link, nNLOS = 2.4
and χσNLOS = 5.27 dB;d0= 5 m;fc= 38 GHz.
By choosing the specific channel parameters, the standard log-
normal shadowing model was exploited to characterize the
mmWave LOS and NLOS links under the constraint of the
given 3D position wk= (xk, yk, hk)of the k-th UAV (hkis
the altitude of the k-th UAV) and 2D position wi= (xi, yi)of
the i-th ground user. More precisely, the channel parameters
include the path loss exponents nLOS and nNLOS for LOS
link and NLOS link, respectively, and the shadowing random
variables χσLOS and χσNLOS for LOS link and NLOS link,
respectively. With the 800 MHz broadband sliding correlator
channel sounder, measurement campaigns for outdoor cellular
channels at 38 GHz were conducted to obtain the measured
data, aiming to quantify these channel parameters [153]. In
Table VI, we provide the comparison among three adopted
empirical channel models for UAV-assisted wireless networks
with 5G mmWave communications. As shown in Table VI, the
quantitative description about each empirical channel model is
given in detail in terms of PDF or path loss.
c) Analytical Channel Modeling: In contrast to empirical
channel models, analytical channel modeling is used to char-
acterize the radio propagation behavior of the channel through
the way of conceptually and mathematically convenient sim-
plifications. This kind of model is very popular for predicting
the dedicated propagation phenomena by devising simplified
models based on the analytical results. In the meantime, the
effect of radio propagation channel on the performance of
communication system can be also analyzed by leveraging the
analytical models under the given channel parameters.
In the light of UAV-based MIMO system with mmWave
communications, Zhao et al. [58] formulated the A2G channel
between flying BS and ground UE as a 3D geometry-based
model. In the considered system, UAV-based flying BS was
equipped with M×Nuniform rectangular arrays (URA),
and the variation of mmWave propagation channel mainly
generated from UAV movements. Under this setting, apart
from the state parameters of UAV movements in 3D space,
the adopted channel model was designed based on the steering
vector of URA antenna and channel parameters, e.g., small-
scale fading/large-scale fading coefficients, Doppler frequency,
etc. Similar to the main-servo loop control of ground vehicle
[154], the state parameters of UAV movements can be obtained
through sensor fusion of flight control system of UAV, and
the channel parameters can be estimated by means of pilot
transmission.
IEEE ACCESS 21
TABLE VII
THE COMPARISON OF ANALYTICAL CHANNEL MODELS FOR UAV-AS SI STE D WI REL ES S NET WO RKS W IT H MMWAVE COMMUNICATIONS.
Reference Scenario Communication
Type
Adopted
Channel Model
Quantitative Description
[58] Aerial BS
with MIMO
communi-
cations
A2G (downlink:
UAV→ground
UEs)
3D
geometry-based
channel model
Channel vector (at time slot with length t):
hk(t) = αk
(Dk)γe−j(2πfdtTscos ϕk+δk)a(φk, θk),
where αkis the small-scale fading coefficient for the k-th UE, γis the
large-scale fading coefficient, Dkis the distance between the antenna-1 of
UAV and the k-th UE, fdis the maximum Doppler frequency, Tsis the
sampling period, ϕkis the angle between uplink transmitted signal and the
motion direction of the k-th UE, δkis the initial phase, and a(φk, θk)is the
steering vector of URA.
[142] Multi-user
MIMO
communi-
cations
A2A (aerial
users→aerial BS,
aerial BS→aerial
users)
D-delay channel
model
Channel matrix (for the d-th delay tap, the n-th channel slot and the u-th
aerial user):
H(n)
u,d =qNRNT,u
Lu
Lu
P
l=1
g(n)
u,l κdTs−τ(n)
u,l fRφ(n)
u,l f∗
T,u θ(n)
u,l ,
where Luis the number of channel paths, NT,u the number of transmit
antennas of the u-th UAV, NRis the number of receive antennas of aerial BS,
g(n)
u,l is the complex gain of the l-th channel path, κis the combined effects
of pulse shaping and analog filtering, Tsis the sampling period, τ(n)
u,l is the
delay of the l-th channel path, fRφ(n)
u,l is the array response vector of
receive UPA, and fT,u θ(n)
u,l is the array response vector of transmit UPA.
[143] MmWave
hybrid UAV
communi-
cations with
blockage
A2G (downlink:
UAV→ground
users)
Saleh-Valenzuela
channel model
Channel vector (between the m-th UAV and the k-th ground user):
hm,k =β(0)
m,kaφ(0)
m,k+
L
P
l=1
β(l)
m,kaφ(l)
m,k,
where β(l)
m,k is the complex gain for both LOS and NLOS component,
β(0)
m,kaφ(0)
m,kis the LOS component of the k-th user, β(l)
m,kaφ(l)
m,kis
the l-th NLOS component of the k-th user, for l= 1,2,···, L,Lis the total
number of the NLoS path, and aφ(l)
m,kis the N×1steering vector of the
k-th user served by the m-th UAV.
[123],
[144]–
[146]
mmWave-
NOMA
transmis-
sion
A2G (downlink:
UAV→ground
users)
Simplified
channel model
Channel vector (between the UAV and the k-th ground user):
hk=√Mαka(θk)
rP Lqd2
k+h2,
where Mis the number of antennas in ULA array, αkis the small-scale gain
of the LOS path which is circularly symmetric complex Gaussian distributed
with CN (0,1),P L qd2
k+h2is the path loss between the UAV and the
k-th user, and a(θk)is the steering vector with AoD θkfor ULA.
[10] MmWave
UAV
cellular
G2A (uplink:
ground
UEs→UAV)
Wideband
time-varying
continuous
channel model
Channel matrix (at time tfor wideband time-varying continuous):
H(t) = √NUENBS
L(t)
P
l=1
λl(t)p(t−τl(t)) a(ψ(t)) aH(Ωl(t)),
where NUE is the number of UE’s antennas, NBS is the number of antennas
of flying BS, L(t)is the number of multipath components, λl(t)is the
complex coefficient of the l-th multipath component, p(t)is the raised
cosine pulse, τl(t)is the relative delay of the l-th multipath component,
a(ψ(t)) is the steering vector with AoA ψ(t)for flying BS, and a(Ωl(t))
is the steering vector with AoD Ωl(t)of the l-th multipath component from
the UE.
[10] MmWave
UAV
cellular
A2G (downlink:
UAV→ground
UEs)
Narrowband
discrete channel
model
Channel matrix (narrowband discrete):
H=√NUENBS
L
P
l=1
λla(ψ)aH(Ωl),
where Lis the number of multipath components, λlis the complex
coefficient of the l-th multipath component, a(ψ)is the steering vector with
AoA ψfor flying BS, and a(Ωl)is the steering vector with AoD Ωlof the
l-th multipath component from the UE.
Under the condition of multi-user MIMO system wherein
multiple UAVs serving as aerial users communicate with
UAV-based aerial BS via the A2A mmWave communication
links, Rodríguez-Fernández et al. [142] characterized the A2A
channel between aerial BS and aerial user by a D-delay
model. Each aerial user was equipped with a hybrid MIMO
architecture, and also employed a hybrid analog precoder to
transfer the data streams denoted by a vector to aerial BS
within each channel slot. In addition, aerial BS and aerial
users were equipped with the uniform planar arrays (UPA).
Particularly, the authors used the d-th delay tap channel matrix
for each aerial user under the given channel slot to describe
the D-delay channel model. Some representative parameters
for mmWave channel and antenna array were incorporated into
the adopted channel model, e.g., the array response vector
of the UPA, the azimuth angle of departure (AoD)/AoA, the
elevation AoD/AoA, etc.
By jointly considering the LOS and NLOS components
of the propagation in mmWave hybrid UAV communications
with blockage problem, Zhao and Jia [143] utilized the Saleh-
Valenzuela channel model to describe the propagation channel
between flying BS and ground user. Each UAV was equipped
IEEE ACCESS 22
by NULA antennas with uniform spacing d. Due to the hybrid
precoding scheme, the number of radio frequency (RF) chains
for each UAV (denoted by NRF) should be much smaller than
the number of antennas with ULA, i.e., NRF N. The
proposed channel model underlined not only the multipath
channel vector for the scenario of multiple ground users (i.e.,
the NLOS component), but also the LOS component of each
ground user with the given complex gain. Additionally, both
the LOS and NLOS components of the propagation depended
on the ULA’s steering-response vectors that were related with
the physical direction of arrival (DoA) of each path between
flying BS and ground user. It is implicitly understood that the
NLOS component strongly characterized the radio propagation
behavior in mmWave communications.
Under the downlink mmWave-NOMA transmission sce-
nario where one UAV-based flying BS equipped with an M
element of ULA array serving multiple ground users with
single antenna, Rupasinghe et al. [123], [144]–[146] used
the simplified channel vector to model the LOS propagation
channel between UAV and ground users. Different from the
Saleh-Valenzuela channel model, the authors just incorporated
the LOS component into the simplified channel model because
they believed that the effect of LOS path is dominant compared
to the NLOS path due to relatively high hovering altitudes of
UAV. In addition, Xiao et al. [10] analyzed the propagation
characteristics of mmWave channels in UAV-assisted cellular
systems with mmWave communications, and discussed some
challenges for mmWave channel modeling, e.g., multipath
components caused by first- and second-order reflections and
sparse AoDs and AoAs in angle domain. Based on these
observations and challenges, the authors proposed to employ
two channel models for mmWave UAV cellular, i.e., the
wideband time-varying continuous channel model and the
narrowband discrete channel model. Beyond that, Li et al.
[135] and Gapeyenko et al. [65] exploited the multipath chan-
nel model to characterize the propagation channel for UAV-
assisted wireless networks with mmWave communications.
Similar to [10], [142], [143], the multipath model in [65], [135]
was formulated bearing in mind the assumption that there are
multiple alternative paths (i.e., multipath components) between
TX and RX. It is noted that each path was featured by some
metrics such as delay, path loss, AoA, AoD, etc. Tables VII
summarizes the above mentioned analytical channel models
and compares the quantitative descriptions for each models in
terms of channel vector or channel matrix.
2) Channel Estimation and Tracking: The technical chal-
lenges including high propagation attenuation, atmospheric
absorption, limited shadowing and diffraction of mmWave
communications can lead to the issue of mmWave channel
sparsity in angular domain. That is, only a limited number of
mmWave channels can be used as feasible and effective prop-
agation paths between TX and RX. By taking advantage of the
channel sparsity of mmWave communications, Talvitie et al.
[147] proposed a method to accurately estimate the channel
parameters, such as AoD, AoA and time of arrival (ToA) of
UE, in mmWave MIMO system by employing a distributed
compressed sensing method known as the simultaneous or-
thogonal matching pursuit algorithm. Note that the considered
MIMO system contained a terrestrial BS with known position
and antenna orientation, and a UE with unknown position
and antenna orientation. An iterative refinement algorithm
was presented to improve the estimation accuracy with the
reasonable complexity in comparison with the large dictionary
based approach. Most important of all, this proposed method
can be extended to the tracking scenario of UAV networks by
replacing UE with UAV.
Under the scenario of multi-user A2A mmWave MIMO
system with the hybrid architecture, Rodríguez-Fernández et
al. [142] devised a channel estimation and tracking algorithm
by using the priori information of the trajectory of each
UAV (i.e., aerial user) in order to reduce both overhead
and computational complexity. In this case, during a training
stage, UAV-based aerial BS was required to estimate the
multi-user mmWave MIMO channel matrices by taking the
priori trajectory information of aerial users into consideration.
To estimate these channel matrices, the maximum likelihood
(ML) estimator assisted with prior information was utilized,
and the optimal ML estimator of the frequency-selective
mmWave MIMO channel was further derived for multi-user
scenario. Moreover, the upper and lower bounds of the azimuth
and elevation AoA and AoD for different uplink channels
were obtained. Simulation results showed that the proposed
channel estimation and tracking algorithm can achieve lower
estimation errors under low overhead and low SINR regime.
Due to the essential attribute of high mobility of UAVs in
UAV-assisted wireless networks with mmWave communica-
tions, one fundamental challenging task is how to accurately
track a large number of dynamic propagation paths while
incurring lower pilot overhead. In the meantime, the overhead
of channel estimation can be also reduced with the help of an
effective channel tracking technique, i.e., tracking the temporal
variations of channel parameters [155], [156]. Motivated by
this observation, Zhao et al. [58] designed an efficient channel
tracking method for flight control system in UAV-based MIMO
system with mmWave communications. The problem of UAV
channel tracking was transformed into the track of state
information of UAV movements, and the estimate of unrelated
parameters of UAV movements. Besides, sensor fusion was
adopted to obtain the UAV movement information. Due to the
fact that the UAV channel tracking was a typical nonlinear
procedure, unscented Kalman filter as a widely recognized
nonlinear filter was used for nonlinear estimation. Compared
with the traditional channel tracking method, the proposed
method required a much lower training overhead.
By investigating the mmWave UAV communications with
beam squint effect, Zhao and Jia [148] proposed an efficient
channel tracking strategy under the comb-type orthogonal
frequency division multiplexing (OFDM) structure (i.e., pilot-
data-multiplexed OFDM structure). In this context, the channel
tracking problem was transformed into the parameter esti-
mate of the channel vector such as DOA, Doppler shift,
and uplink/downlink channel complex gain. Simulation results
demonstrated that the complexity of the downlink channel
tracking was obviously reduced by making efficient use of
both the angle reciprocity and the Doppler reciprocity.
IEEE ACCESS 23
BS
CellularLink
CellularLink
A2G MmWave
Link
Aerial User 1
Aerial User 2
A2A MmWave
Link
Self-blockage
Self-blockage
A2A MmWave
Link
Ground User
Aerial BS
A2G MmWave
Link
Fig. 15. Illustration of self-blockage of mmWave signals incurred by the
rotating propeller of UAV. Here, UAV-based aerial BS with the lightweight
horn antenna can communicate with ground users and aerial users via A2G
mmWave links and A2A mmWave links, respectively.
3) Blockage Detection and Countermeasure: One of the
major technical challenges for UAV-assisted wireless networks
with mmWave communications is blockage effect over the
propagation channel, namely, travelling mmWave signal is
blocked by physical obstacles in their propagation paths.
Different from the other blockage patterns that most of work
discussed, Bao et al. [59] placed more emphasis on self-
blockage problem of mmWave signals incurred by the com-
ponents of UAV itself, i.e., the rotating propeller of UAV, as
illustrated in Fig. 15. In order to confirm that there indeed
exists this kind of self-blockage, the authors conducted the
experiments to identify the blockage effect of UAV propeller
based on hardware testbed. This kind of testbed consisted of
the mmWave TX/RX modules with horn antennas at 60 GHz
frequency and the fine-tuned propulsion system. In particular,
the blockage loss measurement and the blockage pattern
identification were carried out. Experimental results verified
that self-blockage effect of UAV propeller was obvious, which
can severely deteriorate the system performance.
On the basis of the experiments for blockage effect, the
authors in [59] further presented a sequential quickest change
detection strategy to identify the blockage pattern character-
ized by three parameters, i.e., beginning time, time duration,
and SINR drop. Specifically, the detection of blockage pattern
was formulated as a multi-channel quickest detection for
identifying the change of distribution in random process by
utilizing multiple series of observations. Moreover, the Holm
procedure was employed to determine whether a significant
change of channel occurred. Simulation results showed that the
proposed blockage detection approach can accurately identify
the blockage with a modest blockage loss.
To deal with the blockage problem in mmWave UAV
communications with hybrid structure, Zhao and Jia [143]
further provided the corresponding countermeasure proposal
by devising an angle domain channel transmission scheme. In
this scheme, the DoA of LOS path with the most power can
be regarded as the optimal transmission direction. Meanwhile,
the deployment of multiple UAVs was also presented by
taking deployment diversity into account, aiming to reduce
the probability of blockage. A user scheduling mechanism to
maximize the achievable sum rates was also proposed based
on the water-filling algorithm.
4) Summary and Lessons Learned: We have surveyed the
existing efforts on channel modeling, channel estimation and
tracking, and blockage detection and countermeasure for UAV-
assisted wireless networks with 5G mmWave communications.
In particular, channel measurement and modeling techniques
have been reviewed and discussed from three aspects, i.e.,
propagation measurement, empirical channel modeling, and
analytical channel modeling. The detailed comparisons of
estimated path loss models based on realistic measurements,
empirical channel models, and analytical channel models have
been summarized in Table V, Table VI, and Table VII,
respectively. The important lessons learned from the review
of research issues and solutions related to radio propagation
channel are summarized as follows:
•Through the realistic measurement campaigns within
indoor or outdoor environments, we can use channel
sounder and vector network analyzer as effective tools to
obtain the measurement results of propagation behavior
and characteristics of mmWave signals along the bidirec-
tional mmWave links. The distance based free space path
loss model can be properly employed to characterize the
mmWave LOS and NLOS channels by choosing the key
parameters (e.g., α,β, and σ) reasonably.
•For empirical channel models, we need to give more
insights into the current fading types (i.e., small-scale or
large-scale fading) of A2G and G2A mmWave propaga-
tion channels when we apply them into our scenario. We
should also pay attention to what kind of PDFs or path
loss models can be utilized to describe fading distribution.
For the application of analytical models, we ought to take
the antenna and its associated parameters into account.
Moreover, channel vector or channel matrix are required
to be properly formulated to characterize the propagation
behavior of mmWave channels between UAVs and ground
UEs.
•To overcome the challenge of mmWave channel sparsity
in angular domain, channel estimation and tracking ap-
proach should be well designed to achieve lower overhead
and higher estimation accuracy. The commonly used
methods that we learned include compressed sensing,
priori information based policy, Kalman filter, etc. The
blockage effect of mmWave propagation channel not
only covers conventional blockage caused by physical
obstacles, but also includes self-blockage due to UAV
itself. The latter one is a novel scenario for further
research on the blockage problem.
C. Multiple Access Mechanism
The UAVs acting as flying BSs can provide aerial access to
ground UEs during temporary events, e.g., hotspot areas and
IEEE ACCESS 24
large public venues, where a large number of UEs straining
the available wireless resources. Consequently, much more
spectrally efficient multiple access techniques are required for
enabling multiple UEs to share the same available wireless
resources, to support massive connectivity while maintaining
the requirements of different quality of service (QoS). Recent-
ly, NOMA has been regarded as an effective multiple access
solution for 5G to improve spectral efficiency and support
many thousands of UEs [43], [157]–[159]. In addition, with the
advanced multiple antenna technique, SDMA as the channel-
based multiple access mechanism has also attracted increasing
research interests, especially for inherently directional A2G
mmWave beams under our considered scenario. In the fol-
lowing subsection, we will summarize the related research on
NOMA [123], [144]–[146] and SDMA [10], [160] for UAV-
assisted wireless networks with mmWave communications.
1) Non-Orthogonal Multiple Access: To increase the spec-
tral efficiency and serve more ground users simultaneously,
Rupasinghe et al. [123], [144], [145] proposed a downlink
NOMA transmission mechanism at UAV-based flying BS
operating in mmWave frequency bands. By maximizing the
achievable mmWave-NOMA sum rates under the constraint of
the given hovering altitude of UAV, a beam scanning strategy
was introduced to identify the best physically radiated region
of mmWave beam due to the partial coverage of user region by
downlink beam. According to the constraint that the transmit
power allocation of mmWave-NOMA TX to ground users was
based on channel quality between ground user and flying BS,
each ground user should send its feedback information about
channel quality back to mmWave-NOMA TX. More precisely,
the transmit power allocation towards each ground user via the
NOMA transmission should be conducted in a way inversely
proportional to its channel quality.
By considering that full channel state information (CSI)
would generate more link overheads, Rupasinghe et al. also
[146] presented two limited feedback schemes in terms of user
distance and user angle instead of full CSI feedback. The
authors pointed out that both user distance and user angle
were changing much slowly compared with full CSI. The
summary of three feedback schemes to measure the channel
quality used in [123], [144]–[146] is provided in Table VIII. In
addition, an analytical framework was developed to optimize
the outage probability of each ground user and the outage
sum rates based on the user distance feedback policy. The
authors noted that mmWave-NOMA with distance feedback
can exhibit better outage sum rates compared to conventional
orthogonal multiple access (OMA).
2) Spatial-Division Multiple Access: With the aid of the
inherently directional A2G mmWave beams from flying BS,
multiple ground users using different spatial beams can access
the channel with the same set of frequencies concurrently.
Based on this insight, Xiao et al. [10] explored the use of
SDMA or known as beam division multiple access (BDMA) in
UAV-assisted cellular systems with mmWave communications,
as illustrated in Fig. 16. To reach this goal, ground users should
be dynamically divided into several different parallel spatial
user groups or clusters according to the entire range of AoDs
TABLE VIII
SUM MARY O F TH E CHA NNE L QUA LI TY FE EDB ACK S CHE ME S USE D IN T HE
PRO POS ED M MWAVE-NOMA TRANSMISSION MECHANISM.
Feedback Scheme Measurement Metric
Full CSI |αk|2
M×P Lqd2
k+h2
sinπM(¯
θ−θk)
2
sinπ(¯
θ−θk)
2
2
Ground user distance dk
Ground user angle FMπ¯
θ−θk
Parameter description:αk: small-scale gain of single LOS path
between the k-th user and flying BS, which is given by circularly
symmetric complex Gaussian distributed with CN (0,1);M: number
of antennas in uniform linear antenna array; dk: horizontal distance
between the k-th user and flying BS; h: hovering altitude of flying BS;
P L qd2
k+h2: path loss between the k-th user and flying BS; θk:
AoD of LOS path between the k-th user and flying BS; ¯
θ:
beamforming azimuth AoD of flying BS beam; FM(·): Fejér kernel
with Mantennas in uniform linear antenna array.
UserGroup1
User Group2
User Group 3
UserGroup4
UserGroup5
Multiple
Beam
Directional
Antenna
A2G MmWave
Beam
Flying BS
Same Set of MmWave
Frequencies
Channel
Fig. 16. Illustration of the SDMA technique in UAV-assisted cellular systems
with mmWave communications. Here, flying BS is equipped with a smart (or
adaptive) antenna (e.g., multiple beam directional antenna), and each ground
user is equipped with a single antenna. Each A2G mmWave beam from flying
BS should correspond to only one user group, and the number of beams is
equal to the number of transceivers of flying BS using the smart antenna
system.
of ground users, and each user group must be identified by
a beamforming codeword pair. In particular, the beamforming
codeword pair of the i-th ground user was characterized by
{wi, fi}, where widenotes the beam combining codeword for
the i-th ground user at the side of BS, and firefers to the
beamforming codeword for the i-th ground user at the side
of mobile station (MS) on the ground. It is worth noting that
codewords wiand fiwere both obtained from the predefined
codebook. Compared to conventional UAV cellular operating
at lower microwave frequencies, UAV-assisted cellular sys-
tems with mmWave-SDMA strategy demonstrated a superior
performance in terms of the total achievable rate of uplink
transmission and multi-user capacity.
Considering that the ULA antenna would lead to huge phys-
ical size of the transceiver of UAV, Tan et al. [160] adopted
the uniform circular arrays (UCAs) for phased array antennas
to be deployed at flying BS in UAV-based MIMO systems
under Ricean fading channel. With the only knowledge of
the CSI at the transceiver of UAV, a statistical-eigenmode
mmWave-SDMA approach was used in the downlink transmis-
IEEE ACCESS 25
Spatial Configuration
Position and Trajectory
Optimization UAV Deployment Control
Position
Optimization
Trajectory
Optimization
Small Cell
Densification
Deployment
UAV Clustering and
Swarm Formation
Control
Fig. 17. The classification of spatial configuration solutions for UAV-assisted
wireless networks with mmWave communications.
sion for two ground users with single-antenna. Furthermore,
a suboptimal beamforming precoder was exploited to maxi-
mize the SINR of each ground user. Based on this setting,
the authors constructed a general theoretical framework to
analyze the achievable rate of UAV-based MIMO systems.
Based on the suboptimal beamforming precoder, a closed-form
expression of achievable rate was rigorously derived, which
was a function of the number of antennas, the radius of UCA
configuration, and the azimuth and elevation AoDs. Moreover,
it has been shown that the achievable sum-rate of the systems
was quite close to a fixed saturation value when SINR was
larger than 20 dB.
3) Summary and Lessons Learned: We have provided a
review on existing works about two kinds of multiple access
techniques for mmWave downlink transmission, i.e., NOMA
and SDMA, for UAV-assisted wireless networks with 5G
mmWave communications. Three feedback schemes to mea-
sure the channel quality used in downlink NOMA transmission
mechanisms have been summarized in Table VIII. The impor-
tant lessons learned from the review of related research on
NOMA and SDMA are summarized as follows:
•The proposed NOMA and SDMA mechanisms all fall
into the category of mmWave downlink transmission,
wherein UAV-based flying BS serves as the TX side of
multiple access. Due to the effect of channel quality on
transmit power allocation, feedback information about
channel quality should be sent back to flying BS in
mmWave NOMA transmission. We need to turn our
attention to the use of feedback information, because
different feedback information determines different link
overheads.
•The SDMA approach, also known as BDMA, is well
suited to our considered scenario, owing to the spatial
attribute for inherently directional A2G mmWave beams.
The difficult is how to adaptively and effectively assign
ground users into different parallel spatial clusters.
D. Spatial Configuration
Due to the capability to hover, sufficient flexibility, ease of
deployment, higher maneuverability, rapid reconfiguration, etc,
UAVs can be effectively deployed at any positions of interest
in 3D space, to serve as the aerial relay/BS/access point. For
the rotary wing UAVs, their real-time positions are relatively
fixed and static over a given geographic area. However, the
spatial positions will be in the dynamic time-varying states
for the fixed-wing UAVs according to current instant time. For
such insight about UAV spatial configuration, one important
problem for the design of UAV-assisted wireless networks with
mmWave communications is to determine and optimize the
spatial positions and trajectories of UAVs, to comply with the
requirement of system performance. Additionally, except for
spatial deployment in regard to the optimization of positions
and trajectories, another challenge is how to effectively control
the multi-UAV deployment, e.g., UAV clustering and UAV
swarm formation. In the following, as shown in Fig. 17, we
will review the existing studies about spatial configuration of
UAVs in terms of position and trajectory optimization [37],
[64], [161]–[164] as well as UAV deployment control [17],
[141], [165].
1) Position and Trajectory Optimization: For the position
and trajectory optimization, the objective is to reconfigure
UAV’s position and trajectory in order to fulfill the require-
ments of system performance and resource optimization. As
described in Fig. 17, we will provide readers a review of
related work about position optimization and trajectory op-
timization of UAVs in the context of UAV-assisted wireless
networks with mmWave communications.
a) Position Optimization: To find and determine the
optimal position of UAV as an aerial relay accurately and
quickly is an extremely challenging task in mmWave UAV
relay systems. This is because UAV generally has no priori
knowledge about its optimal position. In this context, Kong
et al. [161] proposed an Autonomous Relay solution to tackle
this problem. The main idea of the Autonomous Relay is to
apply compressive sensing technique into online measurement
and estimate of link quality of mmWave beam in 3D space. To
describe the link quality, the authors constructed a 3D matrix
Q∈Rn1×n2×n3, aiming to take advantage of compressive
sensing for 3D Matrix, where n1,n2, and n3are the coordi-
nate scales of the 3D space position point. Specifically, link
quality was defined as the product of the receiving quality
Ro(i, j, k)at the 3D point with coordinate (i, j, k)from
the mmWave TX o(origin point) and the receiving quality
Rc(i, j, k)at the 3D point (i, j, k)from the mmWave RX c,
i.e., Q(i, j, k) = Ro(i, j, k)×Rc(i, j, k ). In order to find
the optimal position point p∗
x, p∗
y, p∗
zfor UAV relay so that
link quality Qp∗
x, p∗
y, p∗
zwas maximized, link quality matrix
update procedure was formulated as a matrix recovery problem
by using MatrixUpdate method.
Based on the work by Kong et al. [161], Thomas and
Chandran [162] reviewed the Autonomous Relay approach
specialized for quick and accurate determination of mobile
relay’s position in mmWave communications. With the help
of this approach, link qualities of mmWave beams can be
sampled while UAV is moving. According to the real-time
sampling results, UAV adjusted its trajectory to further reach
its optimal position. It has been shown that the Autonomous
Relay approach can obtain more accurate relay position and
can also generate more stable results than existing classic
methods including the K-nearest neighbor (KNN) and the
tensor recovery (TR).
IEEE ACCESS 26
By taking the use of the caching in UAVs and the user-
centric information on the ground into account, Chen et al.
[37] investigated the proactive deployment issue of cache-
enabled UAVs to optimize the QoE of ground mobile users
in the CRAN. In this work, UAVs acted as the flying cache-
enabled RRHs and the A2G communication links between
UAVs and ground users employed mmWave frequencies in
the CRAN system. However, G2A communication links (i.e.,
wireless fronthaul links) between terrestrial baseband units
(BBUs) and UAVs employed traditional licensed cellular bands
at microwave frequencies. In particular, a realistic mobility
model for users was leveraged by considering the periodic,
daily, and pedestrian mobility patterns, and the QoE metric
was defined as the concrete human-in-the-loop metric that
captures data rate, delay, and device type of each user. Under
this setup, their goal was to find an effective deployment (i.e.,
optimal positions) of cache-enabled UAVs to improve each
user’s QoE while minimizing the transmit power of UAVs.
Specifically, this optimization problem can be expressed by:
minimize
Ck,Uτ,k ,wτ,t,k
T
X
τ=1 X
k∈K X
i∈Uτ,k
F
X
t=1 2δR
i,n|Uτ,k |/BV−1σ210
¯
lt,ki
10
(5a)
subject to hmin ≤hτ,k, k ∈ K,(5b)
i6=j, i, j ∈ Ck,Ck⊆ N , k ∈ K,(5c)
0< P min
τ,t,ki ≤Pmax , i ∈ U, k ∈ K,(5d)
where Uis the set of ground users, Nis the set of popular
contents, Kis the set of UAVs, Ckis the set of Ccached
contents in the storage units of the k-th UAV, Uτ,k is the set
of ground users served by the k-th UAV at the τ-th time slot,
wτ,t,k is the coordinate of the k-th UAV with altitude of the
k-th UAV denoted by hτ,k at the τ-th time slot, hmin is the
minimum altitude of UAV, Tis the number of time slots, Fis
the number of intervals in each time slot, δR
i,n is the minimum
rate that is required to maximize the QoE of user, BVis the
total bandwidth available for each UAV, σ2is the noise power
spectral density, ¯
lt,ki is the average path loss from the k-th
UAV to i-th user at the t-th interval, Pmin
τ,t,ki is the minimum
transmit power required to ensure the QoE requirement of the
i-th user receiving the n-th content at the t-th interval within
time slot τ, and Pmax is the maximum transmit power of UAV.
To solve this problem, Chen et al. [37] proposed a prediction
algorithm by taking advantage of machine learning framework
of the conceptor-based echo state networks (ESNs), aiming to
find the optimal positions of UAVs as well as the user-UAV
association and the content caching at UAVs. In this case, the
optimal positions of UAVs were the 3D points where UAVs
were serving their associated ground users with minimum
transmit power. The algorithm to find the optimal position of
each UAV just had the runtime complexity O(|K|)in linear
time. It is also observed via simulations that the proposed
algorithm can achieve better performance gains in respect
of the minimum transmit power of UAVs compared to the
traditional ESN methods.
Considering the impact of human-body blockage on the
LOS links of flying BSs by using mmWave communications
as shown in Fig. 14, Gapeyenko et al. [64] explored the
Fig. 18. Illustration of dynamic aerial relay used for data collection in WSNs.
optimized deployment problem of mmWave-based flying BS,
which captures the optimal altitude, coordinate, and coverage
radius of flying BS. An approximate A2G mmWave path
loss model was exploited for quasi-stationary flying BS with
rotary wing, which combines the LOS and non-line-of-sight
(NLOS) transmission links at mmWave frequencies together.
Thus, the 3D position point (xD, yD, hD)of mmWave-based
flying BS was associated with the adopted path loss model,
where (xD, yD)is the two-dimensional (2D) space position
of flying BS over the horizontal plane coordinate and hDis
the altitude of flying BS. To obtain the optimal 3D position
point (x∗
D, y∗
D, h∗)of flying BS, the authors formulated a 3D
placement problem of flying BS as follows:
maximize
xD,yD,h,{mi}X
i∈M
mi(6a)
subject to miσi≥miQ, i ∈ M,(6b)
X
i∈M
mi≤N, i ∈ M,(6c)
xl≤xD≤xu, yl≤yD≤yu, hl≤h≤hu,
(6d)
mi∈ {0,1}, i ∈ M,(6e)
where Mis the set of ground users, miis a binary variable
indicating whether the i-th user is covered or not, h=hD−hR
(hRrefers to the height of the users), σiis the SINR of the
i-th user, Qis the target SINR level, Nis the maximum
number of users that flying BS can concurrently serve, and
the subscripts uand lstand for the upper and the lower limits
of available positions in 3D space. Through the analytical
derivations by solving the 3D placement problem, the authors
demonstrated their theoretical results can provide a tight match
with those obtained by using the interior-point optimization
method through the MOSEK optimization software.
Conventionally, highly capable sensor node with large trans-
mit power is generally chosen as the relay in WSNs. In
contrast to this static placement of relay, Fu et al. [163] used
a single UAV operating at mmWave frequency in 3D space
as the dynamic aerial relay between ground sink node (i.e.,
concentrator) and terrestrial BS (i.e., border gateway) over
WSNs as shown in Fig. 18. The authors attempted to find
IEEE ACCESS 27
the optimal position of UAV by minimizing the system power
consumption. Under such circumstance, the authors defined
the system power consumption denoted by PTotal as the sum
of the transmit power PSof sink node and the transmit power
PUof the UAV, i.e., PTotal =PS+PU. Thus, their objective
was to minimize the total of system power consumption, which
can be further formulated as:
minimize σ2
S
|hS|22C
BS−1+σ2
U
|hU|22C
BU−1,(7)
where σ2
Sand σ2
Udenote the noise power spectral density at
UAV and terrestrial BS, respectively, |hS|2and |hU|2refer to
the wireless channel gain from sink node to UAV and from
UAV to terrestrial BS, respectively, BSand BUstand for the
wireless bandwidth of sink node and UAV, respectively, and C
represents the traffic capacity for transmission. Based on this
optimization problem, the algorithm to determine the optimal
position of UAV was proposed, and the performance of UAV
based relay in WSNs was also verified by comparing with
traditional wireless transmission without relay.
b) Trajectory Optimization: As stated previously, the ob-
stacles to the use of mmWave communications pose additional
challenges on optimizing the operations of UAVs operating at
mmWave frequencies for various communication tasks while
minimizing energy consumption and service time. Faced to
these challenges, Ghazzai et al. [164] devised a generic
optimization framework to smartly assign UAVs as aerial
relays to serve some pairs of transceivers. More precisely,
each UAV can support the dual-band communication module
operating at both mmWave and microwave bands to complete
data transmission for pairs of transceivers. For each pair of
transceiver, TX or RX may be located on the ground or in the
air. Thus, three path loss models were adopted for microwave
band in terms of A2G, A2A, and ground-to-ground (G2G)
communication links, and two path loss models were also
applied for mmWave band according to A2G and A2A links.
To quantify the total energy consumption of each UAV, the
authors characterized the hover and transition energy required
for its movement, and the communication energy needed to
perform the relaying of the transceiver’s data. Given this
scenario, the authors formulated the service time Sn,d that
is required to serve the n-th pair of transceivers by the d-th
UAV as follows:
Sn,d =X
m∈N \{n}
pm,n,dSm,d +X
m∈N \{n}
pm,n,dTf
m,n,d
+Tc
n,d
X
m∈N \{n}
pm,n,d
,
(8)
where Nis the set of the pairs of transceivers, pm,n,d is a
binary variable indicating whether the d-th UAV is directly
serving the n-th pair of transceivers after the m-th pair, and
Tf
m,n,d is the flying time of the d-th UAV to move from the
position where it serves the m-th pair of transceivers to the
position where it serves the n-th pair. It should be noted that
the first term of the right hand side of (8) is the service time
of the previous m-th pair of transceivers served by the d-th
UAV, and the second term of the right hand side of (8) is the
time where the d-th UAV flies from the 3D position xmto the
F
o
max
H
min
H
F
,B o
F
o
x
y
z
Horizontal Plane
Flying Space
Maximum
Altitude
Minimum
Altitude
Candidate UAV
Candidate UAV
Fig. 19. The cooperative UAV clustering based on the cylinder B(o, χ)+.
3D position xn, and the last term of the right hand side of (8)
is the communication time of the d-th UAV to complete the
relaying of the transceiver’s data. For DUAVs denoted by a
set D, the service time of the n-th pair of transceivers can be
further calculated as Sn=Pd∈D Pm∈N pm,n,dSn,d .
The proposed framework by Ghazzai et al. [164] was
aimed at optimizing the trajectories of UAVs such that a
weighted sum of service times of all pairs of transceivers was
minimized. By considering both the communication time of
UAVs to relay the data and the flying time of UAVs, this
objective problem was formulated as a mixed non-linear pro-
gramming problem, which was optimally solved to determine
the trajectory of each UAV based on a hierarchical iterative
approach. It is important to emphasize that the iterative
approach comprised four steps with the aim to determine:
i) the potential 3D relaying positions of UAVs by solving
the unconstrained non-convex problem, ii) the trajectories of
UAVs by converting the proposed objective problem into a
mixed integer non-linear programming problem (MINLP) and
solving the MINLP problem, iii) the adjustment of UAV stops
by using the 3D hierarchical search, and iv) the algorithm
convergence. Simulation results were presented to demonstrate
the performance of the proposed approach and to evaluate the
trajectories of UAVs by adjusting the system parameters.
2) UAV Deployment Control: According to Fig. 17, related
studies for UAV deployment control generally focus on three
directions, i.e., construction of UAV cluster within a predefined
cylinder space, formation control of drone swarm to enhance
the network capacity, and small cell densification via UAV
deployment.
a) UAV Clustering and Swarm Formation Control: The
use of the energy harvesting (EH) enabled caching UAVs in
terrestrial cellular networks brings about two major advan-
tages: i) to ease the fronthaul congestion by directly providing
the cached popular contents at UAVs towards the ground
mobile terminals (GMTs), and ii) to prolong the operational
duration of UAVs by harvesting the energy from ambient
environment such as wind and solar. However, the intermittent
energy arrival through EH and the uncertainty of caching
IEEE ACCESS 28
poses additional challenges on the robust connectivity and
the ubiquitous coverage in UAV-assisted wireless cellular net-
works. In response to this challenge, Wu et al. [17] discussed
the problem of coordination and cooperation between UAVs
and ground BSs under this scenario. Moreover, a user-centric
cooperative UAV clustering scheme was proposed aiming
to offload GMTs from BSs to UAVs within the predefined
cylinder space. The core of the cooperative UAV clustering
was to construct the cylinder B(o, χ)+, where ois the 2D
space projection position centered on a target GMT with the
content requirement over the horizontal plane coordinate and
χis the radius of the cylinder, as depicted in Fig. 19. Once the
cylinder was determined, a candidate UAV can be selected to
send its cached contents that a target GMT requested, which
must be subject to three necessary conditions simultaneously
as follows:
•The candidate UAV was under a flight condition with a
minimum altitude Hmin and a maximum altitude Hmax,
and also owned enough energy including the harvested
energy and the onboard energy to support the wireless
communication modules.
•The candidate UAV had the cached contents that should
match the requirement of target GMT.
•The connection capacity (i.e., cell loads) of the UAV was
more than the number of other candidate UAV serving
GMTs or the number of other candidate UAV serving
GMTs was much greater than the connection capacity of
the UAV.
Based on those necessary conditions, several candidate
UAVs constituted the cooperative UAV cluster. On the basis
of the cooperative UAV clustering, the explicit expressions of
successful transmission probabilities for a typical GMT were
further obtained with the help of the Gamma approximation
for the distributions of aggregated signal strength.
To effectively control and manage the drone swarm by
terrestrial BSs in the deployment of low power ground nodes
for next generation HetNets, Meng et al. [141] proposed a
robust drone swarm formation control approach to raise the
formation strength of drones during flight control. Their target
was to enhance the network capacity in traffic hotspots by tak-
ing into account the use of mmWave communications during
information exchange between terrestrial BSs and lead drone.
Due to the constraints of traditional shape of antenna such
as limited angles and inconvenience for deployment on flying
objects, the authors presented a novel barrel antenna shape of
antenna to be deployed for drone swarm formation. Especially,
ray tracing model along with the receiver spatial distribution
theory were employed in the design of the barrel antenna. The
advantage of this kind of antenna was to improve the quality
and the received power of mmWave signals between terrestrial
BSs and lead drone. In the proposed approach, the drones
equipped with the adopted antenna structure can fly around the
given regions in which the lead drone transmitted its position
signals to terrestrial BS as the backhauling and received the
fronthaul signal from terrestrial BS simultaneously.
b) Small Cell Densification Deployment: In order to
overcome the limitation of blockage effect of mmWave com-
munications, the technique of small cell densification (i.e.,
network densification) is a good choice for design of 5G
system architecture. The reason is that the application of small
cell densification can raise the probability of LOS transmission
conditions due to the increased number of access points (APs).
To enable the small cell densification, Khosravi et al. [165]
applied the UAVs as the mmWave APs which were deployed
in a Manhattan grid manner as the typical urban scenarios,
aiming to reduce the probability of blockage. The effect of
the UAV-mounted mmWave APs on the throughput of ground
UEs and the coverage probability in the case of a non-uniform
load was investigated. Through the simulations, the authors
showed that the presence of flying UAV APs with mmWave
communications under the given urban scenarios can lead
to ping-pong effects, where the throughput performance was
found to fluctuate because of frequent handovers of UE access
between UAVs and nearest APs. This would cause the overall
per UE throughput performance deterioration.
3) Summary and Lessons Learned: We have presented a
detailed review of the state-of-the-art related to UAV spatial
configuration in UAV-assisted wireless networks with 5G
mmWave communications. Especially, optimization of posi-
tions and trajectories as well as how to effectively control
the deployment of multiple UAVs have been reviewed. The
important lessons learned from the review of existing studies
about UAV spatial configuration are summarized as follows:
•No matter position optimization or trajectory optimiza-
tion, the target is to achieve system performance improve-
ment and optimal allocation of wireless resource, while
satisfying a certain metric. In order to obtain the optimal
position of UAV, we can take the prediction algorithm into
account, such as prediction of link quality of mmWave
beam via the Autonomous Relay, and machine learn-
ing framework of ESNs. The trajectory optimization of
UAV is always associated with mathematical optimization
problem under the given starting and ending positions of
UAV. It’s a general idea for designing effective algorithm
to obtain the optimal or suboptimal trajectories.
•For the UAV deployment control, the essential issue
is to deploy and control multiple UAVs according to
specific design requirements. We can construct the UAV
cluster consisting of multiple candidate UAVs within a
cylinder space. The aim of this clustering is to carry out
the theoretical analysis of system performance, e.g., the
successful transmission probabilities.
E. Resource Management
The integration of numerous emerging services and applica-
tions into UAV-assisted wireless networks with mmWave com-
munications requires a fundamental understanding on design
principles and control mechanisms, to fulfill the requirements
of various optimization criteria, e.g., QoS, energy efficiency,
throughput maximization, etc. Under these requirements, one
of the main challenges for design principles and control mech-
anisms is how to efficiently manage and schedule network re-
sources, such as transmit power, radio spectrum, transmission
rate, computing capability, subcarrier, backhaul capacity, stor-
age availability, content caching, etc. The problem of resource
IEEE ACCESS 29
management has been broadly involved in several network
scenarios [36], [166]–[172], but not well explored in UAV-
assisted wireless networks with mmWave communications.
In what follows, the related research progress on resource
management in our considered scenario will be summarized
from two aspects according to what kind of network resources
to be managed, i.e., power and subcarrier allocation [173],
[174] and caching optimization [37], [175].
1) Power and Subcarrier Allocation: In order to efficiently
exploit the advantages of 5G HetNets assisted by UAVs,
Chakareski et al. [173] and Naqvi et al. [174] envisaged
a novel HetNet architecture that employed the dual opera-
tional frequency bands in comparison with traditional HetNets.
Specifically, the HetNet architecture consisted of a macro
ground BS operating at microwave band, multiple dual-mode
small ground BSs operating at two kinds of mmWave bands,
and multiple UAV-based flying small BSs operating at mi-
crowave band. Additionally, ground users can be served by
three kinds of cases: i) microwave ground macro BS, ii)
microwave flying small BSs (SBSs), and iii) ground SBSs with
lower mmWave bands represented by Land higher mmWave
bands denoted by H.
Under this scenario of interest, Chakareski et al. [173]
and Naqvi et al. [174] designed an efficient radio resource
management optimization framework for mmWave 5G Het-
Nets empowered by flying SBSs. For the case of microwave
ground macro BS, joint optimization of achievable rate and
energy efficiency of all ground users was formulated as an
optimization problem to maximize the sum rate and minimize
the total power consumption for ground users simultaneously,
while satisfying the minimum QoS and the maximum transmit
power constraints. Mathematically, the presented optimization
problem was expressed as follows:
maximize
pBS
m,n
φX
m∈MBS X
n∈N BS
κm,nrBS
m,n
Rnorm −(1 −φ)X
m,n
pBS
m,n
Pnorm
(9a)
subject to X
m∈MBS X
n∈N BS
pBS
m,n ≤PBS
max,∀m, ∀n, (9b)
X
m∈MBS X
n∈N BS
κm,nrBS
m,n ≥RBS
min,∀m, ∀n, (9c)
pBS
m,n ≥0,∀m, ∀n, (9d)
κm,n ∈ {0,1} ∀m, ∀n, (9e)
where pBS
m,n is the power allocated to the m-th ground user on
the n-th subcarrier by ground macro BS, rBS
m,n is the achievable
rate of the m-th ground user on the n-th subcarrier associated
with ground macro BS, MBS is the set of ground users
associated with ground macro BS, NBS is the set of subcarriers
available for ground macro BS, κm,n is the binary variable
indicating whether the n-th subcarrier is assigned to the m-th
ground user, Pnorm is the maximum transmit power of ground
macro BS, and Rnorm is the maximum achievable rate under
the constraint of Pnorm. By using the Lagrangian function,
the optimal power allocated to the m-th ground user on the
n-th subcarrier can be derived via the Lagrangian multipliers
which was further obtained through the sub-gradient method.
Based on the optimal power allocation solution, the optimal
subcarrier allocation can be also computed by using the
Hungarian method.
For the case of microwave flying SBSs, the maximum
interference threshold constraint was introduced to guarantee
QoS to ground users associated with macro ground BS.
Thus, the transmit power on a reused subcarrier by flying
SBSs should be subject to the minimum rate requirement.
Finally, for the case of dual-mode ground SBSs, the authors
in [173], [174] proposed a step-wise algorithm for subcarrier
allocation and user association to a specific mmWave band
by using the Time-division multiple access (TDMA) scheme
for subcarrier access. Numerical results demonstrated that the
proposed optimization framework outperformed the conven-
tional approaches in terms of maximizing the system sum rate
or minimizing the system power consumption.
2) Caching Optimization: Caching the popular contents at
the edge of networks closer to end users during off peak hours,
has emerged as a promising alternative to mitigate the burden
of backhaul links in core networks, enhance the capacity of the
bandwidth constrained fronthaul links, and improve the QoE
of end users. Motivated by these advantages, Chen et al. [37]
explored the proactive deployment problem of cache-enabled
UAVs to optimize the QoE of ground mobile users in the
CRAN by proactively downloading and caching the popular
contents at UAVs during off peak hours. The UAVs as the
flying cache-enabled RRHs can directly deliver the caching
contents to the requested ground users via A2G mmWave
frequencies. Additionally, the QoE metric was characterized
by the concrete human-in-the-loop metric that captures data
rate, delay, and device type of each user. Based on this
setup, an optimization problem was formulated as a UAV’s
transmit power minimization problem in (5) while satisfying
the QoE metric constraint for each ground user. To solve this
problem, a prediction algorithm by using the conceptor ESN
was proposed to generate the content request distribution and
the mobility pattern of each ground user. In this way, the user-
UAV associations were determined via the K-mean clustering
approach, and the optimal contents to be cached by each UAV
was further obtained. Particularly, within time duration T, the
optimal set of contents Ckto be cached by the k-th UAV can
be calculated by:
Ck= arg max
Ck
T/H
X
j=1
H
X
τ=1 X
i∈Uτ,k X
n∈Ck
(pj,in∆Pj,τ,ki,n),(10)
where His the number of time slots to collect user mobility,
pj,in is the content request distribution of the i-th user during
the j-th time period, and ∆Pj,τ,ki,n is the reduction of UAV’s
transmit power by caching the contents during the j-th time
period. Simulation results showed that the deployment of the
cache-enable UAVs can greatly reduce the rate required for
reaching the QoE metric threshold of each ground user with
IEEE ACCESS 30
VR content
collection
UAV
VR users
SBS
SBS
SBS
UAV
UAV
Content
caching
VR users
VR users
A2G mmWave
beam
Fig. 20. Illustration of the UAV-assisted wireless VR networks.
the low wireless fronthaul rate for each ground user. Further-
more, compared with two cases without cache and without
optimizing the locations of UAVs, the proposed prediction
algorithm via the conceptor ESN can achieve 40% and 25% of
the reduction of the average transmit power of ground users,
respectively.
Recently, the emerging VR technology has enjoyed both
technological advances and widespread adoption to merge
physical and virtual environment in real time. However, ex-
isting cellular networks cannot match the 360° VR content
transmission due to limited backhaul and fronthaul capacity.
Faced to this problem, Chen et al. [175] presented a framework
of content caching and visible content transmission for UAV-
assisted wireless VR networks, aiming to reduce the data
traffic burden of the backhaul and enable the ground VR users
to achieve the application requirements of low latency. In this
scenario, as shown in Fig. 20, UAVs firstly collect VR contents
with visible format and 360° format that the VR users request,
and then send the collected contents to the cache-enabled
SBSs through wireless downlink backhaul links at mmWave
frequencies. In the meantime, the SBSs can directly transmit
the caching contents to VR users via traditional licensed
cellular bands. With this in mind, joint content caching and
visible content transmission was formulated as an optimization
problem with the goal to maximize the reliability of VR user.
The reliability of VR user was defined as the probability of
the VR user’s successful transmission, which indicated that
the content transmission delay of VR user should satisfy the
target of the VR delay requirement.
To maximize the reliability of VR user so as to find the
optimal contents to cache and content transmission format,
Chen et al. [175] further proposed a deep learning framework
known as echo liquid state machine (ELSM) by combining the
neural network concepts of the liquid state machine (LSM) and
the ESN together. Specifically, the LSM was used to adjust
the association policy of VR user, the cached contents and
the cache content format according to the VR user’s content
request. Due to the higher complexity of training the traditional
LSM, the ESN serving as the output function of the LSM was
presented. The proposed ELSM algorithm can record large
number of historical information and use it to predict the
future output. Furthermore, compared with the ESN algorithm,
the total reliability gains of the proposed ELSM algorithm
can reach 14.5% and 25.4%, respectively. Simulation results
showed that the ELSM algorithm can achieve 14.7% and
20.2% reliability gains, respectively, in comparison with the
ELSM algorithm with random caching and the ESLM with
random transmission format.
3) Summary and Lessons Learned: We have pointed out
the importance of efficiently managing and scheduling net-
work resources. The related research progress on resource
management in our considered scenario has been summarized
based on what kind of network resources to be managed. The
important lessons learned from the review of research issues
and solutions about resource management are summarized as
follows:
•Generally, the target of resource management is to ef-
ficiently and optimally allocate network resources, e.g.,
transmit power, subcarrier, caching, computing capability,
etc. As such, resource allocation problem can be formu-
lated as a mathematical optimization problem, while com-
plying with one or more optimization criteria, e.g., QoE,
energy efficiency, total power consumption, throughput
maximization, and so on.
•Compared with the power and subcarrier allocation, the
optimization of the cached popular contents at the net-
work edge is an emerging hot topic in UAV-assisted
wireless networks with mmWave communications. In
essence, caching optimization together with other re-
source allocation can be initially formulated as a joint
optimization problem. We can take advantage of machine
learning framework, e.g., conceptor ESN, ELSM, multi-
agent deep reinforcement learning, to design efficient
algorithm for allocating the cached contents.
F. Security Strategy
The broadcast nature of radio propagation of mmWave
signals makes wireless transmissions be open and accessible
to both legitimate users and unauthorized users (i.e., mali-
cious eavesdroppers and attackers) in UAV-assisted wireless
networks with 5G mmWave communications. The open trans-
mission environment and the lack of physical protection lead
to serious security risk of malicious attacks and wiretapping
under our considered scenario, e.g., passive interception of
sensitive information and disruption of legitimate transmission
via active jamming [176]. To combat the attacks, efficient
security mechanism needs to be designed to enable the legiti-
mate receivers to successfully obtain the confidential messages
from the transmitter, while preventing the eavesdroppers from
interpreting the legitimate transmission. The traditional key-
based cryptographic techniques generally focus on applying
the encryption and decryption at the higher layers such as
network layer and application layer to secure wireless trans-
missions [11]. Alternatively, by taking advantage of physical
characteristics of radio propagation, information theoretic se-
curity can be also achieved at the physical layer to provide di-
rect secure communications in UAV-assisted wireless networks
[177]–[180]. By integrating mmWave communications into
UAV-assisted wireless networks, there also have been some
IEEE ACCESS 31
Eavesdropper
Eavesdropper
Legitimate
Receiver
T
QPSK Constellation
Diagram
DM Transmitter
A2G MmWave
Beam
DOA
Fig. 21. Illustration of DM system with QPSK modulation in UAV-assisted
wireless networks with mmWave communications.
recent studies on physical layer security that are focused on
directional modulation [181]–[183] and jamming transmission
[124]. In the following subsection, we will give a summary
about these related studies.
1) Directional Modulation: Owing to the inherent direc-
tional mmWave beams, it goes without saying that there exists
the line-of-propagation (LOP) channels (i.e., LOS propagation
channels) in UAV-assisted wireless networks with mmWave
communications. As an important physical layer security tech-
nique, directional modulation (DM) [181], [183]–[187] has at-
tracted considerable attention from both academia and industry
to provide an efficient solution for secure wireless communi-
cations, especially for the LOP channel scenarios. Compared
with conventional cryptographic methods at the upper layers,
DM has the potential to enhance the security at the phys-
ical layer in UAV-assisted wireless networks with mmWave
communications, based on the feature of spatial direction-
dependent signal-modulation-formatted transmission.
As we all know, in conventional modulation, malicious
eavesdroppers have the chance to eavesdrop on the captured
messages from the receiving modulated signals, because the
only difference among the received messages from different
directions lies in the signal power for traditional wireless trans-
mitter. However, DM is typically a transmitter side security
technique due to its ability to project the modulated signals of
confidential messages into one or more desired spatial direc-
tions in the definite constellation patterns. At the same time,
in all the other spatial directions, the constellation patterns
of the same modulated signals will be purposely distorted in
the phase and amplitude such that the eavesdroppers cannot
intercept and decode the confidential messages successfully.
As a result, the directional angles of the desired receiver and
eavesdroppers (i.e., the DoAs of the modulated signals) are
greatly important for the DM transmitter, in that DM can
achieve high performance gain by beamforming technique in
the desired spatial directions. From a practical view point, the
desired spatial directions can be defined in advance as a priori
information for the DM transmitter and the legitimate receiver.
The typical DM system with QPSK modulation under the
scenario of UAV-assisted wireless networks with mmWave
communications is illustrated in Fig. 21, where there are one
UAV with the multi-beam antenna array in the air as a DM
transmitter, one ground legitimate receiver, and two ground
eavesdroppers. At the side of the DM transmitter, the QPSK
modulated signals with four phase-dependent unique symbols
of confidential messages are generated in mmWave frequency
by using the 3D beamforming technology. For the legitimate
receiver, the modulated signal of the confidential messages
is projected into the a priori spatial direction denoted by the
DOA θwhich owns a definite constellation pattern as shown
on the top left corner of Fig. 21. For the eavesdroppers,
the constellation patterns of the same modulated signal are
specially scrambled along all the directions other than θ, which
also means the artificial noise (AN) projection. The joint
operation of beamforming and AN projection at the physical
layer makes the eavesdroppers very difficult to intercept the
confidential messages, which can further enhance the security
of the system.
However, when the eavesdropper moves inside the direction
of main beam of the desired receiver, and the eavesdropper can
also intercept the confidential messages. To solve this problem,
a new concept of secure and accurate wireless transmission
was proposed by Shu et al. [181]–[183], in which DM, random
frequency diversity, and phase alignment were incorporated.
Particularly, a DOA measurement method based on Bayesian
learning was presented, and the precoding vector and the arti-
ficial noise projection matrix at the DM transmitter were fur-
ther designed. Compared to the single-snapshot measurement
without machine learning, the proposed DOA measurement
approach can generate a substantial secrecy rate performance
gain. Additionally, the Bayesian learning based method can
improve the DOA measurement precision with fully-digital
structure. To be specific, when the size of the training data set
was increasing, the root mean squared error decreased and the
DOA measurement precision was also improved. Moreover,
the secrecy rate can increase with the growth of the SINR
of system using the DM technique, but the secrecy rate gain
would gradually decrease when the SINR increased.
2) Transmit Jamming: In consequence of the broadcast
nature of the A2G wireless links, UAV-assisted wireless net-
works are particularly susceptible to the eavesdropping threats.
Faced to the challenges of the security vulnerabilities, Zhu et
al. [124] employed the physical layer security technique to
analyze the secrecy performance of UAV-assisted wireless net-
works with mmWave communications. In this scenario, UAVs
serving as the flying BSs with multiple transmit antennas were
used to provide wireless access from the sky to ground legiti-
mate receivers, in coexistence with multiple ground eavesdrop-
pers. To guarantee the minimum safety distance between the
randomly deployed UAVs, the Matérn hardcore point process
was leveraged to implement the UAV deployment. The average
secrecy rate between the associated UAV and the typical
ground legitimate receiver was also evaluated, in terms of the
complementary cumulative distribution function (CCDF) of
the average rate between the associated UAV and the typical
legitimate receiver, and the cumulative distribution function
IEEE ACCESS 32
Fig. 22. Illustration of the coexistence scenario of the associated UAV and
the UAV jammer to to transmit the jamming mmWave signals to the ground
eavesdroppers.
(CDF) between the associated UAV and the most detrimental
eavesdropper. Even more important, as a core idea of applying
the physical layer security strategy, a part of the UAVs of all
the deployed UAVs was utilized as the jammers in the air to
transmit the jamming mmWave signals to the eavesdroppers,
as shown in Fig. 22. Based on the incorporation of the UAV
jammers into the scenario, the improved average achievable
secrecy rate for the jamming-aided UAV transmission was
further characterized by:
R(J)
Sec ="1
ln 2 Z∞
0
P(J)
cov,R(γ) + F(J)
E∗(γ)−1
1 + γdγ#+
,(11)
where γ > 0is the threshold, P(J)
cov,R(γ)is the CCDF of the
SINR at the typical receiver, and F(J)
E∗(γ)is the CDF of the
SINR at the most detrimental eavesdropper. Simulation results
showed that optimizing the jamming factor (i.e., the percentage
of UAVs that transmit the jamming mmWave signals) can quite
improve the secrecy rate. Furthermore, the use of the proper
number of UAVs to transmit the jamming mmWave signals can
reduce the eavesdropper’s rate more than the typical receiver’s
rate.
3) Summary and Lessons Learned: We have presented a
summary of recent studies on physical layer security that are
focused on directional modulation and jamming transmission.
The important lessons learned from the review of research
issues and solutions about physical layer security are summa-
rized as follows:
•The directional modulation depends on the feature of
spatial direction-dependent signal-modulation-formatted
transmission to enhance the security at the physical
layer. Typically, directional angles of the desired receiver
and eavesdroppers should be different to distinguish the
constellation patterns of modulated signals. In particular,
directional angles can be defined previously as a priori
spatial direction that owns a definite constellation pattern.
Therefore, directional angles are quite similar to the
frequency hopping sequence in frequency-hopping spread
spectrum system.
•For the jamming transmission, the primary idea is to
apply the partial portions of UAVs as the jammers to
proactively transmit the jamming mmWave signals to the
eavesdroppers. Through this kind of method to transmit
jamming signals, the average achievable secrecy rate
between the associated UAV and the ground legitimate
receiver can be improved.
G. Performance Assessment
The assessment of the performance of UAV-assisted wireless
networks with 5G mmWave communications plays a signifi-
cant role in guiding us to design network protocols, optimize
networking structures, analyze system parameters, diagnose
performance issues, and so forth. The commonly used meth-
ods to evaluate the performance of UAV-assisted wireless
networks with mmWave communications include theoretical
performance analysis [65], [188]–[190] and simulation tool
based measurement campaign [191]. Both of the methods
can be exploited to quantitatively and qualitatively assess the
network performance and provide some valuable data, metrics,
models, approaches, frameworks, and mechanisms to improve
the network performance. In the following, we will review
the recent studies that apply performance analysis and mea-
surement campaign into the performance assessment of UAV-
assisted wireless networks with mmWave communications.
1) Performance Analysis: Considering the importance of
evaluating the performance of UAV-assisted wireless networks,
especially for mmWave scenarios, Yi et al. [188] proposed
a unified 3D spatial framework to theoretically assess the
network performance by employing the tractable blockage
model and the sectorized antenna pattern. In this scenario,
UAV acting as the flying BS was deployed to provide the
wireless link for ground users during the entire transmission
process consisting of uplink phase and downlink phase. For
the downlink phase, the authors analyzed the coverage perfor-
mance of downlink transmission by considering the distance
distributions of UAVs and the Laplace transform of interfer-
ence in Poisson point processes. The losed-form expression for
the downlink coverage probability at the typical BS was further
derived, to enhance the evaluation efficiency. For the uplink
phase, the coverage performance of uplink transmission was
also analyzed by characterizing the distance distributions in
the corresponding or other clusters and the Laplace transform
of interference in Poisson cluster processes. Moreover, the
general expression for the uplink coverage probability at the
corresponding UAV was rigorously obtained. Based on the
coverage probabilities of uplink and downlink phases, the
system coverage probability was formulated as the function of
SINR thresholds for downlink and uplink phases. Simulation
results indicated that a large number of antenna elements had
the capability to improve the coverage performance.
Under the scenario of the low-altitude UAVs deploying
alongside the ground BS, Galkin et al. [189] proposed to
utilize the stochastic geometry method to analyze the backhaul
performance of ground stations serving the low-altitude UAV
IEEE ACCESS 33
network in the urban environment. In particular, the authors
analyzed the directional antenna alignment in 3D space be-
tween a typical UAV in the scenario and the ground stations
including the associated backhaul ground station of UAV
and other interfering ground stations. The antenna array was
equipped by each ground station, and the 3D beamforming was
used to concentrate the highly directional beam towards UAV,
which forms the backhaul link operating at both 2 GHz LTE
and mmWave frequencies. Additionally, by setting the SINR
threshold, analytical expression for the probability of success-
fully establishing the backhaul link was obtained. The expect-
ed data rate over the backhaul link was further derived based
on the backhaul probability. Compared with the interference-
limited LTE signals, the mmWave signal was the noise limited
signal which can generate a higher backhaul probability for
UAV. Numerical results showed that the backhaul probability
can increase monotonically with the growth of the height of
the ground station, which also demonstrated that the backhaul
performance was improved by deploying the ground station as
high as possible above the ground.
The inherent challenges including multipath propagation
of urban environments and dynamic blockage of mmWave
links impose additional constraints and requirements for the
mmWave backhaul connections in UAV-assisted wireless net-
works. To address these challenges, Gapeyenko et al. [65]
adopted the flexible and reconfigurable backhaul architecture
in UAV-assisted wireless networks, for the purpose of achiev-
ing the capability to dynamically reroute the backhaul links
due to the unreliable blockage-prone mmWave links. The
typical mathematical framework was presented by capturing
the 3D multipath mmWave channel model, the heterogeneous
mobility of UAVs and blockers, and the dynamic blockage
effects. Based on this framework, the authors theoretical-
ly carried out the performance assessment of the flexible
mmWave backhaul operation in the case of the crowded
urban deployments, aiming to explore the spatial and temporal
characteristics of the mmWave backhaul. The evaluation of
the mmWave backhaul performance specifically covered two
metrics of interest, i.e., time-averaged and time-dependent
metrics. Numerical results demonstrated that both the outage
probability and the spectral efficiency were improved with the
growth of the intensity of UAV-BS traversals. In terms of the
flight speed, the decrease of the UAV-BS speed presented an
obvious positive effect on the mmWave backhaul performance.
As one major challenge, blockage effect for mmWave
signal along the LOS propagation path between the drone-
enabled aerial access point and the UE poses the additional
constraints on the design of the analytical model for drone-
based mmWave communications. Based on the extension of
conventional stochastic geometry method, Kovalchukov et
al. [190] proposed a 3D analytical model for drone-based
mmWave communications by jointly capturing the random
heights of communicating entities (i.e., drone-enabled aerial
access points and UEs) and the high directionality of transmis-
sion. Their objective was to analyze the performance of drone-
based mmWave communications through joint consideration
of the above factors for the 3D analytical model. Furthermore,
both the mean interference and the SINR were derived as
explicit expressions. The system performance was assessed
with the emphasis on the impact of the vertical dimension in
aerial mmWave connectivity. Numerical results showed that
capturing the vertical exposure probability along with the
random altitude of communicating entities and UEs to access
the aggregate interference were of great importance to the
accurate performance assessment.
2) Measurement Campaign: In addition to the performance
assessment using the theoretical analysis, simulation tool based
measurement campaign can be also conducted to evaluate the
performance of UAV-assisted wireless networks with mmWave
communications. By utilizing the existing analytical models
for free space path loss, spatial consistency, and small-scale
fading, Vasiliev et al. [191] assessed the end-to-end efficiency
of UAV-assisted LTE-like mmWave cellular networks in the
crowded area with the aid of NS-3 simulation tool. Owing to
the deployment of antenna arrays and the high throughput, the
28 GHz band was used for mmWave communications. For the
node layout, ground BS and UE were located in the opposite
corners of the simulation area, and UAV as the aerial relay was
located in the middle of the area. Two simulated scenarios
including the source-destination case and the source-UAV-
destination case were selected in the measurement campaign.
On the basis of this setting, end-to-end performance in terms
of the QoS metric, i.e., the average goodput and the goodput
gain, was measured for these two kinds of scenarios with
different parameters of free space path loss and blockage
density. The results via the measurement campaign showed
that the performance of the source-UAV-destination case was
more efficient than that of the source-destination case in terms
of average goodput and goodput gain. Especially, more than
80% of goodput gain was obtained by using the UAV as aerial
relay with high blockage density.
3) Summary and Lessons Learned: We have provided read-
ers a survey of existing research efforts related with the per-
formance assessment of UAV-assisted wireless networks with
mmWave communications. To be more concrete, performance
analysis and measurement campaign have been reviewed and
discussed, respectively. The important lessons learned from
the review of recent studies on the performance assessment
are summarized as follows:
•The performance analysis of UAV-assisted wireless net-
works with mmWave communications focuses more on
analyzing and assessing one or more performance metrics
from a theoretical perspective, e.g., coverage perfor-
mance, backhaul performance, and the effect of multiple
factors on system performance. The main factors influ-
encing the system performance include blockage effect of
mmWave propagation channels, antenna patterns, vertical
heights of communicating entities, and aggregate interfer-
ence.
•Compared to the performance assessment using the the-
oretical analysis, we can also conduct the simulation
tool based measurement campaign to verify the system
performance. The major problem of using the simulation
tool (e.g., NS-3 network simulator) is how to initialize
the network parameters that are in accordance with the
actual environments.
IEEE ACCESS 34
V. SUMMARY, OPE N ISS UE S AN D FUTURE RESEARCH
DIRECTIONS
Many research efforts have been devoted to the development
of 5G mmWave communications for UAV-assisted wireless
networks. In the previous section, we have presented a com-
prehensive summary of current achievements in the integration
of 5G mmWave communications into UAV-assisted wireless
networks, and have also discussed the state-of-the-art issues,
solutions, and open challenges in the newly emerging area. The
survey has covered and highlighted seven different research
issues that we would like to focus on, i.e., antenna technique,
radio propagation channel, multiple access mechanism, spatial
configuration, resource management, security strategy, and
performance assessment. Based on the current great research
efforts in the literature, in this section, we will summarize the
existing research works in the way of comparison, and outline
the open research issues along with the potential directions
that need to be well investigated to encourage future research
studies on this area.
A. Summary
The key research information related to existing research
works in the area of 5G mmWave communications for UAV-
assisted wireless networks has been summarized in Fig. 23
and Fig. 24. From Fig. 23, the percentage of existing research
works in terms of seven important research issues is vividly
described as a pie chart. As can be observed, the majority of
existing research works, close to 40%, are aimed at modeling
the radio propagation channel and capturing the propagation
characterization of mmWave signals for UAV-assisted wireless
networks with 5G mmWave communications. This is because
that the propagation channels in mmWave frequencies steeply
differ from the channels in traditional microwave frequencies
due to the technique challenges caused by higher frequencies,
e.g., higher atmospheric attenuation and severe free space path
loss. Correspondingly, the study and analysis on the radio
propagation channel including channel modeling, channel esti-
mation and tracking, and blockage detection and countermea-
sure, have great theoretical significance and applicable value.
From this figure, we can also find that the percentage of other
research issues occupies a comparatively small proportion,
which indicates a potential research opportunity. In addition,
Fig. 24 illustrates the time distribution of existing research
works. From Fig. 24, it is clearly revealed that almost all
of existing research works are distributed in the last three
years. These research efforts appear a continually increasing
trend, although six months have passed in 2019. Particularly,
the research trend shows that the integration of 5G mmWave
communications into UAV-assisted wireless networks has been
recognized as one of the research hotspots in research com-
munity recently.
B. Challenges and Open Issues
1) MmWave Beam Alignment and Beam Switching: As a
consequence of the decrease of mmWave beamwidth and the
mobility of ground UEs, a main challenge in our considered
15%
38%
10%
15%
7%
7%
8% Atenna Technique
Radio Propagation Channel
Multiple Access Mechanism
Spatial Configuration
Resource Management
Security Strategy
Performance Assessment
Fig. 23. The comparison of the percentage of existing research works in 5G
mmWave communications for UAV-assisted wireless networks.
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
0
5
10
15
20
25
30
35
Antenna Technique
Radio Propagation Channel
Multiple Access Mechanism
Spatial Configuration
Resource Management
Security Strategy
Performance Assessment
Fig. 24. The time distribution graph of existing research works in 5G
mmWave communications for UAV-assisted wireless networks.
scenario is to effectively steer and align the mmWave beams
between TX and RX sides at the beginning of communications
[112]–[115]. Efficient beam alignment together with highly
directional beamforming can achieve a large directional gain
to combat their stronger atmospheric attenuation and higher
free space path loss. Therefore, 5G mmWave communications
for UAV-assisted wireless networks need to require advanced
beam alignment techniques, e.g., beam tracking, beam train-
ing, hierarchical beam codebook design, beam search, accu-
rate estimation of the channel, etc. Among them, there are
lots of issues need to be resolved, but one important issue
is to mitigate the beam alignment overhead. Moreover, an
alternative solution to perform a proper beam switching is
also required when conventional beam alignment schemes by
using beam sweeping become inefficient if not impossible
[192]. In practical scenarios, both TX and RX sides cannot
purely guarantee perfect beam alignment due to time-varying
channel conditions and inaccurate beam tracking. As such, by
considering the beam alignment errors, the impact of imperfect
beam alignment between TX and RX sides on the system
performance and directional gain should further be analyzed
and evaluated.
IEEE ACCESS 35
2) Dynamic Resource Allocation by Integrating Content
Caching and Computation Offloading: In practical environ-
ment, both the users and the UEs on the ground are always
moving continuously, and even quickly, e.g., vehicle nodes.
Meanwhile, the UAVs’ spatial positions and trajectories are
also changing overtime in a dynamic time-varying state es-
pecially for the fixed-wing UAVs. Apparently, it will be far
more realistic to dynamically manage and schedule network
resources according to the current instant time in the dynamic
scenario. That is, we should perform dynamic resource allo-
cation in which the network resource will be allocated in a
continuous-time manner [193]. This kind of continuous-time
resource allocation differs from traditional resource allocation
by assuming that the finite time horizon is divided into fixed
time slots in a discrete way [19], [33], [194], [195]. By
using the edge caching, the popular social contents can be
cached at the edge of network during off-hours, to reduce
backhaul loading at peak-hours. With the help of the MEC,
the resource-limited UEs can offload their computation tasks
to closer computation servers, to obtain multiple benefits, e.g.,
reduced latency, alleviated backhaul congestion, and enhanced
user experience. Current research efforts have been targeted at
content caching and computation offloading from/to the sky
[17], [34]–[36], [196]–[198], but not well investigated in UAV-
assisted wireless networks with mmWave communications.
Therefore, one potential issue that needs to be resolved is to
achieve joint dynamic resource allocation for content caching
and computation offloading.
3) Cross-Layer Optimization Framework for Routing and
Distributed Resource Allocation: Instead of using one single
UAV, multiple UAVs have potentials to bring many benefits to
assist wireless networks by creating drone swarm, such as high
scalability, enhanced fault tolerance capability, longer network
lifetime, improved multitasking ability, and so forth [74]–[76],
[199]. Meanwhile, information exchange and transfer among
multiple UAVs can be achieved by using the multi-hop drone
interconnection via A2A mmWave links. In comparison with
the lower layer solutions explored by existing research efforts,
recent work indicates that there are many new challenges
to routing problem at the network layer in our considered
scenario. Several characteristics and constraints that affect the
design of routing mechanism should be well incorporated,
e.g., large scale spatial distribution, high degree of mobility,
high dynamic network topology, possible blockage effect, etc.
In addition, with a joint cross-layer optimization, other layer
is able to share its information and configuration about its
core function with the network layer without breaking the
hierarchical structure of the layered architecture [200]. The
cross-layer design perspective can improve network perfor-
mance and efficiency through the cross-layer coupling between
network layer and other layers. As a result, joint cross-layer
optimization for routing and distributed resource allocation is
not only desirable, but also necessary.
4) Joint Optimization of 3D Beamforming and AN Projec-
tion in Directional Modulation: As a transmitter side security
technique, the DM is used to project the modulated signals
of the confidential messages into one or more desired spatial
directions with the definite constellation patterns. In our con-
sidered scenario, by adopting the 3D beamforming architecture
with multi-beam antenna array, the UAV serves as the DM
transmitter to provide the secure communications at the physi-
cal layer for ground legitimate receivers. Nevertheless, existing
works only take single desired legitimate receiver into account
[181], [185]. That is, only one mmWave beam generated by the
DM transmitter through modulating the confidential messages
can be securely conveyed along a desired spatial direction. As
such, one of the key challenges is to extend single legitimate
receiver to multiple receivers by projecting multiple modulated
signal of the confidential messages along multiple different
spatial directions simultaneously. Additionally, as a spatial
direction-dependent transmission, secure mmWave commu-
nications via the DM depends directly on the predefined
spatial direction, i.e., DOA, for ground legitimate receiver.
For the eavesdroppers, the same modulated signals will be
particularly scrambled like the AN projection due to their
different spatial directions. However, the eavesdroppers and
legitimate receiver can be located in the same region. More
seriously, one of the eavesdroppers has the chance to move
inside the direction of the main beam of the desired receiver.
Hence, efficient and adaptive spatial direction measurement
and adjustment mechanisms for the 3D beamforming are
required to provide more accurate control of DOAs for ground
legitimate receivers.
C. Potential Research Directions
1) UAV-Assisted Wireless Power Transfer with MmWave
Communications: A challenging issue in UAV-assisted wire-
less networks is that no matter ground UEs or UAVs usually
have limited energy storage for normal operations. To prolong
the operational time of energy-constrained UEs and UAVs as
well as next generation electric vehicles [201], a straightfor-
ward and effective approach is to apply wireless power transfer
to provide a controllable and sustainable energy supply [202]–
[204]. For our considered scenario, on the one hand, UAVs are
employed as mobile energy transmitters over the air to deliver
wireless energy towards ground UEs. On the other hand, UAVs
can be recharged by other power source, e.g., the solar energy
by installing solar panels on UAVs [205] and the RF-enabled
wireless power transfer performed by other UAVs or ground
fixed energy transmitters. It is believed that wireless power
transfer can be incorporated with mmWave communications
for UAV-assisted wireless networks, where further research is
needed.
2) MmWave Communications for Aerial Access Networks:
To achieve the 3D ubiquitous super-connectivity in the global
scale, the traditional terrestrial network architecture should be
fully integrated with various non-terrestrial platforms, includ-
ing UAVs, balloons, airships, and very low Earth orbit (VLEO)
satellites [206]–[208]. The integration of these platforms has
substantially boosted the prospect of implementing the aerial
access networks (AANs) by extending the spatial scale of
UAVs at low-altitude to higher altitudes. As the emerging
radio access platforms, the AANs has attracted growing at-
tentions from both academia and industry as a key enabler
for future sixth generation (6G) systems of the 2030s. In
IEEE ACCESS 36
addition, as a very promising way to sustain ultra-high-speed
transmission, we can also apply mmWave communications in
the AANs to make global access more robust and reliable
[22]. More investigation is needed in this direction to address
key technique challenges, e.g., 3D channel models, hybrid
altitude optimization, deployment of multiple heterogeneous
aerial platforms, cooperative access control, etc.
3) Network Functions Virtualization for UAV-Assisted Wire-
less Networks with MmWave Communications: It is generally
difficult to effectively manage different network services and
various emerging applications, by identifying specific service
for each networking entity, e.g., ground UEs and aerial BSs,
running on our considered network architecture. There is a
urgent need to build up a highly flexible network architecture
with enhanced intelligence and improved resilience for our
considered scenario [21], [209]–[211]. Network Functions
Virtualization (NFV) provides an efficient paradigm to opti-
mize network resource utility by virtualizing network services
[212]–[214]. Such virtualization enables deployment of new
network services and elastic network scaling to reduce mainte-
nance costs and make network more flexible, scalable and cost-
effective. It is firmly anticipated that the use of NFV in UAV-
assisted wireless networks with mmWave communications
will certainly make the network configuration interesting and
valuable for both academia and industry studies.
4) Integration of Far-Reaching Applications with MmWave
Communications for User-Centric UAV-Assisted Wireless Net-
works: As already stated, mmWave communications have
indeed paved the way into the widespread use of UAVs to en-
able wireless communications and networking, with the target
to provide ultra-high-speed transmission for 5G and beyond
applications. However, current achievements by linking UAV-
assisted wireless networks and mmWave communications to-
gether are primarily focused on conventional services and
applications. With the pervasiveness of Internet of Everything
(IoE), the time has come for an unprecedented proliferation of
newly emerging applications, e.g., boundless mobile eXtended
reality (XR), autonomous driving, haptics, etc [215]–[217],
to improve the QoE objective with the center of ground and
aerial users. Particularly, the original 5G vision of enabling
short-packet and sensing-based massive URLLC services will
be disrupted due to these IoE applications [215]. Further
research is especially warranted to address this challenging
issue of how to integrate the far-reaching applications with
5G mmWave communications for user-centric UAV-assisted
wireless networks.
5) MmWave Communications for UAV-Assisted Wireless
Networks with Machine Learning: MmWave communications
for UAV-assisted wireless networks not only have a substan-
tial potential to achieve higher transmission efficiency with
enhanced coverage and capacity, but also support a wide
variety of 5G and B5G wireless applications. These innovative
applications are expected to unleash a massive IoT ecosystem
with huge data throughput, high wireless bandwidth, super-
fast speeds, ultra-low latency, and increased connectivity. With
exciting opportunities, the design, control and optimization of
UAV-assisted wireless networks with mmWave communica-
tions become very challenging and complex. The significant
challenges require us to configure, manage, and control net-
works in a smarter and more agile manner. Inspired by the
trends, machine learning is needed to enable adaptive learning
and intelligent decision making, due to its capability to achieve
the convergence of computing power, algorithm improvement,
and data proliferation [218]–[221]. Future potential research
will be strongly expected to apply machine learning technique,
e.g., multi-agent deep reinforcement learning [222], federated
learning [223], concept drift learning [224], in our considered
scenario, to overcome the challenges of resource allocation,
cross-layer optimization, physical layer security, etc.
VI. CONCLUSION
In this paper, we have presented a comprehensive literature
survey focusing on the current state-of-the-art achievements
in the integration of 5G mmWave communications into UAV-
assisted wireless networks. By analyzing and reviewing the
existing research efforts, we first introduced a novel tax-
onomy to categorize the cutting-edge solutions from seven
aspects. We then reviewed existing articles related to UAV-
assisted wireless networks and communications as well as 5G
mmWave communications, and further compared our survey
with them. Subsequently, we provided a brief overview of
key technical advantages and challenges as well as potential
applications in 5G mmWave communications for UAV-assisted
wireless networks. With the devised taxonomy, we surveyed in
detail the state-of-the-art issues, solutions, and open challenges
of integrating mmWave communications into UAV-assisted
wireless networks. We also derived several lessons learned
from these research activities. Finally, we pinpointed several
open issues and outlined promising research directions. We
hope that this survey is timely and useful for the interested
researchers and practitioners to concentrate their research
activities on this area for the development of next generation
wireless networks.
REFERENCES
[1] S. Hayat, E. Yanmaz, and R. Muzaffar, “Survey on unmanned aerial
vehicle networks for civil applications: A communications viewpoint,”
IEEE Communications Surveys & Tutorials, vol. 18, no. 4, pp. 2624–
2661, Fourthquarter 2016.
[2] M. Erdelj, E. Natalizio, K. R. Chowdhury, and I. F. Akyildiz, “Help
from the sky: Leveraging UAVs for disaster management,” IEEE
Pervasive Computing, vol. 16, no. 1, pp. 24–32, Jan. 2017.
[3] H. Chao, Y. Cao, and Y. Chen, “Autopilots for small unmanned aerial
vehicles: A survey,” International Journal of Control, Automation and
Systems, vol. 8, no. 1, pp. 36–44, Feb. 2010.
[4] S. A. R. Naqvi, S. A. Hassan, H. Pervaiz, and Q. Ni, “Drone-aided
communication as a key enabler for 5G and resilient public safety
networks,” IEEE Communications Magazine, vol. 56, no. 1, pp. 36–
42, Jan. 2018.
[5] C. Zhang and J. M. Kovacs, “The application of small unmanned aerial
systems for precision agriculture: A review,” Precision Agriculture,
vol. 13, no. 6, pp. 693–712, Jul. 2012.
[6] N. H. Motlagh, T. Taleb, and O. Arouk, “Low-altitude unmanned aerial
vehicles-based internet of things services: Comprehensive survey and
future perspectives,” IEEE Internet of Things Journal, vol. 3, no. 6, pp.
899–922, Dec. 2016.
[7] Shipware, What the Amazon effect means for the shipping
industry, Nov. 2018. [Online]. Available: https://www.shipware.com/
what-the- amazon-effect-means-for-the- shipping-industry.
[8] M. Mozaffari, W. Saad, M. Bennis, Y. Nam, and M. Debbah, “A tutorial
on UAVs for wireless networks: Applications, challenges, and open
problems,” IEEE Communications Surveys & Tutorials, Mar. 2019,
Early Access. DOI: 10.1109/COMST.2019.2902862.
IEEE ACCESS 37
[9] Y. Zeng, R. Zhang, and T. J. Lim, “Wireless communications with
unmanned aerial vehicles: Opportunities and challenges,” IEEE Com-
munications Magazine, vol. 54, no. 5, pp. 36–42, May 2016.
[10] Z. Xiao, P. Xia, and X.-G. Xia, “Enabling UAV cellular with millimeter-
wave communication: Potentials and approaches,” IEEE Communica-
tions Magazine, vol. 54, no. 5, pp. 66–73, May 2016.
[11] A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano, A. García-
Rodríguez, and J. Yuan, “Survey on UAV cellular communications:
Practical aspects, standardization advancements, regulation, and securi-
ty challenges,” IEEE Communications Surveys & Tutorials, Mar. 2019,
Early Access. DOI: 10.1109/COMST.2019.2906228.
[12] S. Sekander, H. Tabassum, and E. Hossain, “Multi-tier drone architec-
ture for 5G/B5G cellular networks: Challenges, trends, and prospects,”
IEEE Communications Magazine, vol. 56, no. 3, pp. 96–103, Mar.
2018.
[13] Y. Zeng, Q. Wu, and R. Zhang, “Accessing from the sky: A tutorial on
UAV communications for 5G and beyond,” CoRR, vol. abs/1903.05289,
Mar. 2019. [Online]. Available: http://arxiv.org/abs/1903.05289.
[14] E. Vinogradov, H. Sallouha, S. D. Bast, M. M. Azari, and
S. Pollin, “Tutorial on UAV: A blue sky view on wireless
communication,” CoRR, vol. abs/1901.02306, Jan. 2019. [Online].
Available: http://arxiv.org/abs/1901.02306.
[15] H. Hellaoui, O. Bekkouche, M. Bagaa, and T. Taleb, “Aerial control
system for spectrum efficiency in UAV-to-cellular communications,”
IEEE Communications Magazine, vol. 56, no. 10, pp. 108–113, Oct.
2018.
[16] V. Sharma, K. Srinivasan, H.-C. Chao, K.-L. Hua, and W.-H. Cheng,
“Intelligent deployment of UAVs in 5G heterogeneous communication
environment for improved coverage,” Journal of Network and Comput-
er Applications, vol. 85, pp. 94–105, May 2017.
[17] H. Wu, X. Tao, N. Zhang, and X. Shen, “Cooperative UAV cluster-
assisted terrestrial cellular networks for ubiquitous coverage,” IEEE
Journal on Selected Areas in Communications, vol. 36, no. 9, pp. 2045–
2058, Sep. 2018.
[18] P. Yang, X. Cao, C. Yin, Z. Xiao, X. Xi, and D. Wu, “Proactive
drone-cell deployment: Overload relief for a cellular network under
flash crowd traffic,” IEEE Transactions on Intelligent Transportation
Systems, vol. 18, no. 10, pp. 2877–2892, Oct. 2017.
[19] H. Wang, J. Wang, G. Ding, L. Wang, T. A. Tsiftsis, and P. K. Sharma,
“Resource allocation for energy harvesting-powered D2D communica-
tion underlaying UAV-assisted networks,” IEEE Transactions on Green
Communications and Networking, vol. 2, no. 1, pp. 14–24, Mar. 2018.
[20] C.-M. Cheng, P.-H. Hsiao, H. T. Kung, and D. Vlah, “Maximizing
throughput of UAV-relaying networks with the load-carry-and-deliver
paradigm,” in Proc. IEEE Wireless Communications and Networking
Conference, Hong Kong, China, Mar. 2007.
[21] I. Bor-Yaliniz and H. Yanikomeroglu, “The new frontier in RAN het-
erogeneity: Multi-tier drone-cells,” IEEE Communications Magazine,
vol. 54, no. 11, pp. 48–55, Nov. 2016.
[22] Y. Huo, X. Dong, T. Lu, W. Xu, and M. Yuen, “Distributed and multi-
layer UAV network for the next-generation wireless communication,”
IEEE Internet of Things Journal, May 2019, Early Access. DOI:
10.1109/JIOT.2019.2914414.
[23] D. Kim, J. Lee, and T. Q. Quek, “Multi-layer unmanned
aerial vehicle networks: Modeling and performance analysis,”
CoRR, vol. abs/1904.01167, Apr. 2019. [Online]. Available: https:
//arxiv.org/abs/1904.01167
[24] ˙
I. Bekmezci, O. K. Sahingoz, and ¸S. Temel, “Flying ad-hoc networks
(FANETs): A survey,” Ad Hoc Networks, vol. 11, no. 3, pp. 1254–1270,
May 2013.
[25] L. Gupta, R. Jain, and G. Vaszkun, “Survey of important issues in UAV
communication networks,” IEEE Communications Surveys & Tutorials,
vol. 18, no. 2, pp. 1123–1152, Secondquarter 2016.
[26] S. Rosati, K. Kruzelecki, G. Heitz, D. Floreano, and B. Rimoldi,
“Dynamic routing for flying ad hoc networks,” IEEE Transactions on
Vehicular Technology, vol. 65, no. 3, pp. 1690–1700, Mar. 2016.
[27] M. Gharibi, R. Boutaba, and S. L. Waslander, “Internet of drones,”
IEEE Access, vol. 4, pp. 1148–1162, Mar. 2016.
[28] G. Choudhary, V. Sharma, and I. You, “Sustainable and secure trajec-
tories for the military internet of drones (IoD) through an efficient
medium access control (MAC) protocol,” Computers & Electrical
Engineering, vol. 74, pp. 59–73, Mar. 2019.
[29] J. Gong, T.-H. Chang, C. Shen, and X. Chen, “Flight time minimization
of UAV for data collection over wireless sensor networks,” IEEE
Journal on Selected Areas in Communications, vol. 36, no. 9, pp. 1942–
1954, Sep. 2018.
[30] W. Shi, H. Zhou, J. Li, W. Xu, N. Zhang, and X. Shen, “Drone assisted
vehicular networks: Architecture, challenges and opportunities,” IEEE
Network, vol. 32, no. 3, pp. 130–137, May 2018.
[31] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Mobile unmanned
aerial vehicles (UAVs) for energy-efficient internet of things commu-
nications,” IEEE Transactions on Wireless Communications, vol. 16,
no. 11, pp. 7574–7589, Nov. 2017.
[32] P. Zhou, X. Fang, Y. Fang, R. He, Y. Long, and G. Huang, “Beam
management and self-healing for mmWave UAV mesh networks,” IEEE
Transactions on Vehicular Technology, vol. 68, no. 2, pp. 1718–1732,
Feb. 2019.
[33] S. Yin, Y. Zhao, and L. Li, “UAV-assisted cooperative communications
with time-sharing SWIPT,” in Proc. IEEE International Conference on
Communications (ICC), Kansas City, MO, USA, May 2018.
[34] N. Zhao, F. Cheng, F. R. Yu, J. Tang, Y. Chen, G. Gui, and H. Sari,
“Caching UAV assisted secure transmission in hyper-dense networks
based on interference alignment,” IEEE Transactions on Communica-
tions, vol. 66, no. 5, pp. 2281–2294, May 2018.
[35] F. Zhou, Y. Wu, R. Q. Hu, and Y. Qian, “Computation rate max-
imization in UAV-enabled wireless-powered mobile-edge computing
systems,” IEEE Journal on Selected Areas in Communications, vol. 36,
no. 9, pp. 1927–1941, Sep. 2018.
[36] L. Zhang, Z. Zhao, Q. Wu, H. Zhao, H. Xu, and X. Wu, “Energy-aware
dynamic resource allocation in UAV assisted mobile edge computing
over social internet of vehicles,” IEEE Access, vol. 6, pp. 56 700–
56 715, Oct. 2018.
[37] M. Chen, M. Mozaffari, W. Saad, C. Yin, M. Debbah, and C. S. Hong,
“Caching in the sky: Proactive deployment of cache-enabled unmanned
aerial vehicles for optimized quality-of-experience,” IEEE Journal on
Selected Areas in Communications, vol. 35, no. 5, pp. 1046–1061, May
2017.
[38] S. Zhang and J. Liu, “Analysis and optimization of multiple unmanned
aerial vehicle-assisted communications in post-disaster areas,” IEEE
Transactions on Vehicular Technology, vol. 67, no. 12, pp. 12 049–
12 060, Dec. 2018.
[39] Y. Saleem, M. H. Rehmani, and S. Zeadally, “Integration of cognitive
radio technology with unmanned aerial vehicles: Issues, opportunities,
and future research challenges,” Journal of Network and Computer
Applications, vol. 50, pp. 15–31, Apr. 2015.
[40] L. Liang, W. Xu, and X. Dong, “Low-complexity hybrid precoding
in massive multiuser MIMO systems,” IEEE Wireless Communications
Letters, vol. 3, no. 6, pp. 653–656, Dec. 2014.
[41] X. Hong, J. Wang, C.-X. Wang, and J. Shi, “Cognitive radio in 5G:
A perspective on energy-spectral efficiency trade-off,” IEEE Commu-
nications Magazine, vol. 52, no. 7, pp. 46–53, Jul. 2014.
[42] N. Bhushan, J. Li, D. Malladi, R. Gilmore, D. Brenner, A. Damn-
janovic, R. Sukhavasi, C. Patel, and S. Geirhofer, “Network densi-
fication: The dominant theme for wireless evolution into 5G,” IEEE
Communications Magazine, vol. 52, no. 2, pp. 82–89, Feb. 2014.
[43] Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, and
V. K. Bhargava, “A survey on non-orthogonal multiple access for 5G
networks: Research challenges and future trends,” IEEE Journal on
Selected Areas in Communications, vol. 35, no. 10, pp. 2181–2195,
Oct. 2017.
[44] J. Wang, A. Jin, D. Shi, L. Wang, H. Shen, D. Wu, L. Hu, L. Gu, L. Lu,
Y. Chen, J. Wang, Y. Saito, A. Benjebbour, and Y. Kishiyama, “Spectral
efficiency improvement with 5G technologies: Results from field tests,”
IEEE Journal on Selected Areas in Communications, vol. 35, no. 8, pp.
1867–1875, Aug. 2017.
[45] E. Yeh and R. Berry, “Throughput optimal control of cooperative relay
networks,” IEEE Transactions on Information Theory, vol. 53, no. 10,
pp. 3827–3833, Oct. 2007.
[46] A. Ghosh, T. A. Thomas, M. C. Cudak, R. Ratasuk, P. Moorut,
F. W. Vook, T. S. Rappaport, G. R. MacCartney, S. Sun, and S. Nie,
“Millimeter-wave enhanced local area systems: A high-data-rate ap-
proach for future wireless networks,” IEEE Journal on Selected Areas
in Communications, vol. 32, no. 6, pp. 1152–1163, Jun. 2014.
[47] P. Wang, Y. Li, L. Song, and B. Vucetic, “Multi-gigabit millimeter
wave wireless communications for 5G: From fixed access to cellular
networks,” IEEE Communications Magazine, vol. 53, no. 1, pp. 168–
178, Jan. 2015.
[48] M. Xiao, S. Mumtaz, Y. Huang, L. Dai, Y. Li, M. Matthaiou, G. K.
Karagiannidis, E. Bjornson, K. Yang, C.-L. I, and A. Ghosh, “Millime-
ter wave communications for future mobile networks,” IEEE Journal
on Selected Areas in Communications, vol. 35, no. 9, pp. 1909–1935,
Sep. 2017.
IEEE ACCESS 38
[49] Y. Niu, Y. Li, D. Jin, L. Su, and A. V. Vasilakos, “A survey of
millimeter wave communications (mmWave) for 5G: Opportunities and
challenges,” Wireless Networks, vol. 21, no. 8, pp. 2657–2676, Apr.
2015.
[50] T. S. Rappaport, S. Sun, R. Mayzus, H. Zhao, Y. Azar, K. Wang, G. N.
Wong, J. K. Schulz, M. Samimi, and F. Gutierrez, “Millimeter wave
mobile communications for 5G cellular: It will work!” IEEE Access,
vol. 1, pp. 335–349, May 2013.
[51] W. Roh, J.-Y. Seol, J. Park, B. Lee, J. Lee, Y. Kim, J. Cho, K. Che-
un, and F. Aryanfar, “Millimeter-wave beamforming as an enabling
technology for 5G cellular communications: Theoretical feasibility and
prototype results,” IEEE Communications Magazine, vol. 52, no. 2, pp.
106–113, Feb. 2014.
[52] X. Wang, L. Kong, F. Kong, F. Qiu, M. Xia, S. Arnon, and G. Chen,
“Millimeter wave communication: A comprehensive survey,” IEEE
Communications Surveys & Tutorials, vol. 20, no. 3, pp. 1616–1653,
Thirdquarter 2018.
[53] I. R. Sector, “Use of multiple gigabit wireless systems in frequencies
around 60 GHz,” Mobile, radiodetermination, amateur and related
satellite services, pp. 1–30, Jan. 2018.
[54] S.-Y. Lien, S.-L. Shieh, Y. Huang, B. Su, Y.-L. Hsu, and H.-Y. Wei,
“5G new radio: Waveform, frame structure, multiple access, and initial
access,” IEEE Communications Magazine, vol. 55, no. 6, pp. 64–71,
Jun. 2017.
[55] M. Shafi, A. F. Molisch, P. J. Smith, T. Haustein, P. Zhu, P. D.
Silva, F. Tufvesson, A. Benjebbour, and G. Wunder, “5G: A tutorial
overview of standards, trials, challenges, deployment, and practice,”
IEEE Journal on Selected Areas in Communications, vol. 35, no. 6,
pp. 1201–1221, Jun. 2017.
[56] H. Ji, S. Park, J. Yeo, Y. Kim, J. Lee, and B. Shim, “Ultra-reliable and
low-latency communications in 5G downlink: Physical layer aspects,”
IEEE Wireless Communications, vol. 25, no. 3, pp. 124–130, Jun. 2018.
[57] W. Khawaja, O. Ozdemir, and I. Guvenc, “UAV air-to-ground channel
characterization for mmWave systems,” in Proc. IEEE 86th Vehicular
Technology Conference (VTC-Fall), Toronto, Canada, Sep. 2017.
[58] J. Zhao, F. Gao, L. Kuang, Q. Wu, and W. Jia, “Channel tracking
with flight control system for UAV mmWave MIMO communications,”
IEEE Communications Letters, vol. 22, no. 6, pp. 1224–1227, Jun.
2018.
[59] J. Bao, D. Sprinz, and H. Li, “Blockage of millimeter wave com-
munications on rotor UAVs: Demonstration and mitigation,” in Proc.
IEEE Military Communications Conference (MILCOM), Baltimore,
MD, USA, Oct. 2017, pp. 768–774.
[60] Z. Xiao, X.-G. Xia, D. Jin, and N. Ge, “Iterative eigenvalue decomposi-
tion and multipath-grouping Tx/Rx joint beamformings for millimeter-
wave communications,” IEEE Transactions on Wireless Communica-
tions, vol. 14, no. 3, pp. 1595–1607, Mar. 2015.
[61] Z. Xiao, P. Xia, and X.-G. Xia, “Full-duplex millimeter-wave commu-
nication,” IEEE Wireless Communications, vol. 24, no. 6, pp. 136–143,
Dec. 2017.
[62] Z. Xiao, “Suboptimal spatial diversity scheme for 60 GHz millimeter-
wave WLAN,” IEEE Communications Letters, vol. 17, no. 9, pp. 1790–
1793, Sep. 2013.
[63] M. Gapeyenko, A. Samuylov, M. Gerasimenko, D. Moltchanov, S. S-
ingh, E. Aryafar, S. ping Yeh, N. Himayat, S. Andreev, and Y. Kouch-
eryavy, “Analysis of human-body blockage in urban millimeter-wave
cellular communications,” in Proc. IEEE International Conference on
Communications (ICC), Kuala Lumpur, Malaysia, May 2016.
[64] M. Gapeyenko, I. Bor-Yaliniz, S. Andreev, H. Yanikomeroglu, and
Y. Koucheryavy, “Effects of blockage in deploying mmWave drone
base stations for 5G networks and beyond,” in Proc. IEEE International
Conference on Communications Workshops (ICC Workshops), Kansas
City, MO, USA, May 2018.
[65] M. Gapeyenko, V. Petrov, D. Moltchanov, S. Andreev, N. Himayat,
and Y. Koucheryavy, “Flexible and reliable UAV-assisted backhaul
operation in 5G mmWave cellular networks,” IEEE Journal on Selected
Areas in Communications, vol. 36, no. 11, pp. 2486–2496, Nov. 2018.
[66] B. Li, Z. Fei, and Y. Zhang, “UAV communications for 5G and beyond:
Recent advances and future trends,” IEEE Internet of Things Journal,
pp. 1–23, 2018, early access.
[67] H. Wang, H. Zhao, J. Zhang, D. Ma, J. Li, and J. Wei,
“Survey on unmanned aerial vehicle networks: A cyber physical
system perspective,” CoRR, vol. abs/1812.06821, Dec. 2018. [Online].
Available: http://arxiv.org/abs/1812.06821.
[68] J. Sánchez-García, J. García-Campos, M. Arzamendia, D. Reina,
S. Toral, and D. Gregor, “A survey on unmanned aerial and aquatic
vehicle multi-hop networks: Wireless communications, evaluation tools
and applications,” Computer Communications, vol. 119, pp. 43–65,
Apr. 2018.
[69] S. Rangan, T. S. Rappaport, and E. Erkip, “Millimeter-wave cellular
wireless networks: Potentials and challenges,” Proceedings of the IEEE,
vol. 102, no. 3, pp. 366–385, Mar. 2014.
[70] I. A. Hemadeh, K. Satyanarayana, M. El-Hajjar, and L. Hanzo,
“Millimeter-wave communications: Physical channel models, design
considerations, antenna constructions, and link-budget,” IEEE Com-
munications Surveys & Tutorials, vol. 20, no. 2, pp. 870–913, Sec-
ondquarter 2018.
[71] T. S. Rappaport, Y. Xing, G. R. MacCartney, A. F. Molisch, E. Mellios,
and J. Zhang, “Overview of millimeter wave communications for
fifth-generation (5G) wireless networks—with a focus on propagation
models,” IEEE Transactions on Antennas and Propagation, vol. 65,
no. 12, pp. 6213–6230, Dec. 2017.
[72] P. Zhou, K. Cheng, X. Han, X. Fang, Y. Fang, R. He, Y. Long, and
Y. Liu, “IEEE 802.11ay-based mmWave WLANs: Design challenges
and solutions,” IEEE Communications Surveys & Tutorials, vol. 20,
no. 3, pp. 1654–1681, Thirdquarter 2018.
[73] F. Jameel, S. Wyne, S. J. Nawaz, and Z. Chang, “Propagation channels
for mmWave vehicular communications: State-of-the-art and future
research directions,” IEEE Wireless Communications, vol. 26, no. 1,
pp. 144–150, Feb. 2019.
[74] J. Jiang and G. Han, “Routing protocols for unmanned aerial vehicles,”
IEEE Communications Magazine, vol. 56, no. 1, pp. 58–63, Jan. 2018.
[75] O. S. Oubbati, A. Lakas, F. Zhou, M. Gunes, and M. B. Yagoubi, “A
survey on position-based routing protocols for flying ad hoc networks
(FANETs),” Vehicular Communications, vol. 10, pp. 29–56, Oct. 2017.
[76] M. Y. Arafat and S. Moh, “A survey on cluster-based routing protocols
for unmanned aerial vehicle networks,” IEEE Access, vol. 7, pp. 498–
516, Jan. 2019.
[77] A. A. Khuwaja, Y. Chen, N. Zhao, M.-S. Alouini, and P. Dobbins,
“A survey of channel modeling for UAV communications,” IEEE
Communications Surveys & Tutorials, vol. 20, no. 4, pp. 2804–2821,
Fourthquarter 2018.
[78] H. Shakhatreh, A. Sawalmeh, A. I. Al-Fuqaha, Z. Dou, E. Almaita,
I. Khalil, N. S. Othman, A. Khreishah, and M. Guizani, “Unmanned
aerial vehicles: A survey on civil applications and key research chal-
lenges,” IEEE Access, vol. 7, pp. 48572–48 634, Apr. 2019.
[79] M. Lu, M. Bagheri, A. P. James, and T. Phung, “Wireless charging
techniques for UAVs: A review, reconceptualization, and extension,”
IEEE Access, vol. 6, pp. 29 865–29 884, May 2018.
[80] M. E. Mkiramweni, C. Yang, J. Li, and Z. Han, “Game-theoretic ap-
proaches for wireless communications with unmanned aerial vehicles,”
IEEE Wireless Communications, vol. 25, no. 6, pp. 104–112, Dec.
2018.
[81] P. Adhikari, “Understanding millimeter wave wireless communication,”
Loea Corporation, Jun. 2008.
[82] Huawei Technologies, “5G spectrum public policy position,” Tech.
Rep., 2017.
[83] H. Mehrpouyan, M. Khanzadi, M. Matthaiou, A. Sayeed, R. Schober,
and Y. Hua, “Improving bandwidth efficiency in E-band communication
systems,” IEEE Communications Magazine, vol. 52, no. 3, pp. 121–
128, Mar. 2014.
[84] S. Yost, “mmWave: Battle of the bands,” National Instruments Corpo-
ration (U.K.) Ltd, Tech. Rep., Mar. 2019.
[85] 5G Americas, “5G americas white paper on 5G spectrum recommen-
dations,” Tech. Rep., Apr. 2017.
[86] Qualcomm Technologies, “QTM052 mmwave antenna modules,”
Oct. 2018. [Online]. Available: https://www.qualcomm.com/products/
qtm052-mmwave-antenna- modules.
[87] J. Johnson and P. R. M, “A study on millimeter-wave vehicular com-
munication,” International Journal of Innovative Research in Science,
Engineering and Technology, vol. 7, no. 10, pp. 10 425–10 431, Oct.
2018.
[88] T. K. Vu, C.-F. Liu, M. Bennis, M. Debbah, M. Latva-aho, and C. S.
Hong, “Ultra-reliable and low latency communication in mmWave-
enabled massive MIMO networks,” IEEE Communications Letters,
vol. 21, no. 9, pp. 2041–2044, Sep. 2017.
[89] B. He and H. Jafarkhani, “Low-complexity reconfigurable MIMO for
millimeter wave communications,” IEEE Transactions on Communica-
tions, vol. 66, no. 11, pp. 5278–5291, Nov. 2018.
[90] L. Yang, Y. Zeng, and R. Zhang, “Channel estimation for millimeter-
wave MIMO communications with lens antenna arrays,” IEEE Trans-
actions on Vehicular Technology, vol. 67, no. 4, pp. 3239–3251, Apr.
2018.
IEEE ACCESS 39
[91] R. Sun, C. A. Gentile, J. Senic, P. Vouras, P. B. Papazian, N. T. Golmie,
and K. A. Remley, “Millimeter-wave radio channels vs. synthetic
beamwidth,” IEEE Communications Magazine, vol. 56, no. 12, pp. 53–
59, Dec. 2018.
[92] A. T. Nassar, A. I. Sulyman, and A. Alsanie, “Radio capacity estimation
for millimeter wave 5G cellular networks using narrow beamwidth
antennas at the base stations,” International Journal of Antennas and
Propagation, vol. 2015, pp. 1–6, Jun. 2015.
[93] Q. Xue, P. Zhou, X. Fang, and M. Xiao, “Performance analysis of
interference and eavesdropping immunity in narrow beam mmWave
networks,” IEEE Access, vol. 6, pp. 67 611–67 624, Oct. 2018.
[94] S. K. Islam and M. R. Haider, Sensors and Low Power Signal
Processing. Springer US, 2010.
[95] H. Friis, “A note on a simple transmission formula,” Proceedings of
the IRE, vol. 34, no. 5, pp. 254–256, May 1946.
[96] T. S. Rappaport, K. Blankenship, and H. Xu, “Propagation and radio
system design issues in mobile radio systems for the glomo project,”
Virginia Polytechnic Institute and State University, 1997.
[97] G. Yue, Z. Wang, L. Chen, L. Cheng, J. Tang, X. Zou, Y. Zeng, and
L. Li, “Demonstration of 60 GHz millimeter-wave short-range wireless
communication system at 3.5 Gbps over 5 m range,” Science China
Information Sciences, vol. 60, no. 8, Jun. 2017.
[98] I. F. Akyildiz, C. Han, and S. Nie, “Combating the distance problem
in the millimeter wave and terahertz frequency bands,” IEEE Commu-
nications Magazine, vol. 56, no. 6, pp. 102–108, Jun. 2018.
[99] A. Benjebbour, M. Iwabuchi, Y. Kishiyama, W. Guangjian, L. Gu,
Y. Cui, and T. Takada, “Outdoor experimental trials of long range
mobile communications using 39 GHz,” in Proc. IEEE 87th Vehicular
Technology Conference (VTC Spring), Porto, Portugal, Jun. 2018.
[100] M. B. Majed and T. A. Rahman, “Propagation path loss modeling
and coverage measurements in urban microcell in millimeter wave
frequency bands,” ARPN Journal of Engineering and Applied Sciences,
vol. 13, no. 8, pp. 2882–2887, Apr. 2018.
[101] G. A. Siles, J. M. Riera, and P. G. del Pino, “Atmospheric attenuation
in wireless communication systems at millimeter and THz frequencies
[wireless corner],” IEEE Antennas and Propagation Magazine, vol. 57,
no. 1, pp. 48–61, Feb. 2015.
[102] Y. Banday, G. M. Rather, and G. R. Begh, “Effect of atmospheric
absorption on millimetre wave frequencies for 5G cellular networks,”
IET Communications, vol. 13, no. 3, pp. 265–270, Feb. 2019.
[103] T. L. Frey, “The effects of the atmosphere and weather on the
performance of a mm-wave communication link,” Applied Microwave
and Wireless, vol. 11, pp. 76–81, 1999.
[104] X. Sun, H. Zhang, W. Meng, R. Zhang, K. Li, and T. Peng, “Primary
resonance analysis and vibration suppression for the harmonically
excited nonlinear suspension system using a pair of symmetric vis-
coelastic buffers,” Nonlinear Dynamics, vol. 94, no. 2, pp. 1243–1265,
Jun. 2018.
[105] P. Nagaraj, “Impact of atmospheric impairments on mmwave based
outdoor communication,” CoRR, vol. abs/1806.05176, Jun. 2018.
[Online]. Available: https://arxiv.org/abs/1806.05176
[106] D. Nandi and A. Maitra, “Study of rain attenuation effects for 5G mm-
wave cellular communication in tropical location,” IET Microwaves,
Antennas & Propagation, vol. 12, no. 9, pp. 1504–1507, Jul. 2018.
[107] I. K. Jain, R. Kumar, and S. S. Panwar, “The impact of mobile blockers
on millimeter wave cellular systems,” IEEE Journal on Selected Areas
in Communications, vol. 37, no. 4, pp. 854–868, Apr. 2019.
[108] Z. Pi and F. Khan, “An introduction to millimeter-wave mobile broad-
band systems,” IEEE Communications Magazine, vol. 49, no. 6, pp.
101–107, Jun. 2011.
[109] T. Bai and R. W. Heath, “Coverage and rate analysis for millimeter-
wave cellular networks,” IEEE Transactions on Wireless Communica-
tions, vol. 14, no. 2, pp. 1100–1114, Feb. 2015.
[110] C. Slezak, V. Semkin, S. Andreev, Y. Koucheryavy, and S. Rangan,
“Empirical effects of dynamic human-body blockage in 60 GHz
communications,” IEEE Communications Magazine, vol. 56, no. 12,
pp. 60–66, Dec. 2018.
[111] J. G. Andrews, T. Bai, M. Kulkarni, A. Alkhateeb, A. Gupta, and R. W.
Heath, “Modeling and analyzing millimeter wave cellular systems,”
IEEE Transactions on Communications, vol. 65, no. 1, pp. 403–430,
Jan. 2017.
[112] J. Wildman, P. H. J. Nardelli, M. Latva-aho, and S. Weber, “On the
joint impact of beamwidth and orientation error on throughput in
directional wireless poisson networks,” IEEE Transactions on Wireless
Communications, vol. 13, no. 12, pp. 7072–7085, Dec. 2014.
[113] X. Yu, J. Zhang, M. Haenggi, and K. B. Letaief, “Coverage analysis for
millimeter wave networks: The impact of directional antenna arrays,”
IEEE Journal on Selected Areas in Communications, vol. 35, no. 7,
pp. 1498–1512, Jul. 2017.
[114] M. Hashemi, A. Sabharwal, C. E. Koksal, and N. B. Shroff, “Effi-
cient beam alignment in millimeter wave systems using contextual
bandits,” in Proc. IEEE Conference on Computer Communications (
INFOCOM), Honolulu, HI, USA, Apr. 2018, pp. 2393–2401.
[115] H. Hassanieh, O. Abari, M. Rodriguez, M. Abdelghany, D. Katabi,
and P. Indyk, “Fast millimeter wave beam alignment,” in Proc. Con-
ference of the ACM Special Interest Group on Data Communication
(SIGCOMM), Budapest, Hungary, Aug. 2018, pp. 432–445.
[116] Z. Xiao, T. He, P. Xia, and X.-G. Xia, “Hierarchical codebook design
for beamforming training in millimeter-wave communication,” IEEE
Transactions on Wireless Communications, vol. 15, no. 5, pp. 3380–
3392, May 2016.
[117] L. Yang and W. Zhang, “Hierarchical codebook and beam alignment
for UAV communications,” in Proc. IEEE Globecom Workshops (GC
Wkshps), Abu Dhabi, United Arab Emirates, Dec. 2018.
[118] C. Liu, M. Li, S. V. Hanly, I. B. Collings, and P. Whiting, “Millimeter
wave beam alignment: Large deviations analysis and design insights,”
IEEE Journal on Selected Areas in Communications, vol. 35, no. 7,
pp. 1619 – –1631, Jul. 2017.
[119] A. Alkhateeb, J. Mo, N. Gonzalez-Prelcic, and R. W. Heath, “MIMO
precoding and combining solutions for millimeter-wave systems,” IEEE
Communications Magazine, vol. 52, no. 12, pp. 122–131, Dec. 2014.
[120] B. Gao, Z. Xiao, C. Zhang, L. Su, D. Jin, and L. Zeng, “Double-link
beam tracking against human blockage and device mobility for 60-GHz
WLAN,” in Proc. IEEE Wireless Communications and Networking
Conference (WCNC), Istanbul, Turkey, Apr. 2014.
[121] J. Lee, M.-D. Kim, J.-J. Park, and Y. J. Chong, “Field-measurement-
based received power analysis for directional beamforming millimeter-
wave systems: Effects of beamwidth and beam misalignment,” ETRI
Journal, vol. 40, no. 1, pp. 26–38, Feb. 2018.
[122] T. Nitsche, A. B. Flores, E. W. Knightly, and J. Widmer, “Steering with
eyes closed: Mm-wave beam steering without in-band measurement,”
in Proc. IEEE Conference on Computer Communications (INFOCOM),
Kowloon, Hong Kong, Apr. 2015, pp. 2416–2424.
[123] N. Rupasinghe, Y. Yapici, I. Guvenc, and Y. Kakishima, “Non-
orthogonal multiple access for mmWave drone networks with limited
feedback,” IEEE Transactions on Communications, vol. 67, no. 1, pp.
762–777, Jan. 2019.
[124] Y. Zhu, G. Zheng, and M. Fitch, “Secrecy rate analysis of UAV-
enabled mmWave networks using Matérn hardcore point processes,”
IEEE Journal on Selected Areas in Communications, vol. 36, no. 7,
pp. 1397–1409, Jul. 2018.
[125] R. Taori and A. Sridharan, “Point-to-multipoint in-band mmwave
backhaul for 5G networks,” IEEE Communications Magazine, vol. 53,
no. 1, pp. 195–201, Jan. 2015.
[126] FCC, “Use of spectrum bands above 24 GHz for mobile radio services,”
Tech. Rep. GN Docket No. 14-177, Feb. 2019.
[127] ——, “Revision of part 15 of the commission’s rules regarding oper-
ation in the 57-64 GHz band,” Tech. Rep. FCC-13-112, Aug. 2013.
[128] L. Zhu, J. Zhang, Z. Xiao, X. Cao, D. O. Wu, and X.-G. Xia,
“3-D beamforming for flexible coverage in millimeter-wave UAV
communications,” IEEE Wireless Communications Letters, vol. 8, no. 3,
pp. 837–840, Jun. 2019.
[129] W. Zhong, L. Xu, Q. Zhu, X. Chen, and J. Zhou, “Mmwave beam-
forming for UAV communications with unstable beam pointing,” China
Communications, vol. 16, no. 1, pp. 37–46, Jan. 2019.
[130] H. Tsuji, M. Oodo, R. Miura, and M. Suzuki, “The development of
intelligent beamforming antenna systems for stratospheric platforms in
the millimeter-wave band,” Wireless Personal Communications, vol. 32,
no. 3-4, pp. 237–255, Feb. 2005.
[131] S. S. Siddiq, G. S. Karthikeya, T. Tanjavur, and N. Agnihotri, “Mi-
crostrip dual band millimeter-wave antenna array for UAV applica-
tions,” in Proc. International Conference on Microwave, Radar and
Wireless Communications, Krakow, Poland, May 2016.
[132] Y. Huo and X. Dong, “Millimeter-wave for unmanned aerial vehicles
networks: Enabling multi-beam multi-stream communications,” CoRR,
vol. abs/1810.06923, Oct. 2018. [Online]. Available: http://arxiv.org/
abs/1810.06923.
[133] K. Heimann, J. Tiemann, S. Boecker, and C. Wietfeld, “On the potential
of 5G mmwave pencil beam antennas for UAV communications: An
experimental evaluation,” in Proc. International ITG Workshop on
Smart Antennas, Bochum, Germany, Mar. 2018.
[134] J. Li, Y. Sun, L. Xiao, S. Zhou, and C. E. Koksal, “Analog beam
tracking in linear antenna arrays: Convergence, optimality, and per-
IEEE ACCESS 40
formance,” in Proc. Asilomar Conference on Signals, Systems, and
Computers, Pacific Grove, CA, USA, Oct. 2017.
[135] ——, “Fast analog beam tracking in phased antenna arrays: Theory
and performance,” CoRR, vol. abs/1710.07873, Jun. 2017. [Online].
Available: http://arxiv.org/abs/1710.07873.
[136] W. Zhong, L. Xu, X. Lu, and L. Wang, “Research on millimeter
wave communication interference suppression of UAV based on beam
optimization,” in Machine Learning and Intelligent Communications.
Springer International Publishing, 2018, pp. 472–481.
[137] S. Kutty and D. Sen, “Beamforming for millimeter wave commu-
nications: An inclusive survey,” IEEE Communications Surveys &
Tutorials, vol. 18, no. 2, pp. 949–973, Secondquarter 2016.
[138] T. He and Z. Xiao, “Suboptimal beam search algorithm and codebook
design for millimeter-wave communications,” Mobile Networks and
Applications, vol. 20, no. 1, pp. 86–97, Jan. 2015.
[139] W. Khawaja, O. Ozdemir, and I. Guvenc, “Temporal and spatial
characteristics of mm wave propagation channels for UAVs,” in Proc.
IEEE Global Symposium on Millimeter Waves (GSMM), Boulder, CO,
USA, May 2018.
[140] M. R. Akdeniz, Y. Liu, M. K. Samimi, S. Sun, S. Rangan, T. S.
Rappaport, and E. Erkip, “Millimeter wave channel modeling and
cellular capacity evaluation,” IEEE Journal on Selected Areas in
Communications, vol. 32, no. 6, pp. 1164–1179, Jun. 2014.
[141] S. Meng, X. Su, Z. Wen, X. Dai, Y. Zhou, and W. Yang, “Robust drones
formation control in 5G wireless sensor network using mmWave,”
Wireless Communications and Mobile Computing, vol. 2018, pp. 1–
7, May 2018.
[142] J. Rodriguez-Fernandez, N. Gonzalez-Prelcic, and R. W. Heath,
“Position-aided compressive channel estimation and tracking for mil-
limeter wave multi-user MIMO air-to-air communications,” in Proc.
IEEE International Conference on Communications Workshops (ICC
Workshops), Kansas City, MO, USA, May 2018.
[143] J. Zhao and W. Jia, “Channel transmission strategy for mmWave hybrid
UAV communications with blockage,” Electronics Letters, vol. 54,
no. 2, pp. 74–76, Jan. 2018.
[144] N. Rupasinghe, Y. Yapici, and I. Guevenc, “Performance of limited
feedback based NOMA transmission in mmWave drone networks,” in
Proc. IEEE International Conference on Communications Workshops
(ICC Workshops), Kansas City, MO, USA, May 2018.
[145] N. Rupasinghe, Y. Yapici, I. Guvenc, and Y. Kakishima, “Non-
orthogonal multiple access for mmWave drones with multi-antenna
transmission,” in Proc. Asilomar Conference on Signals, Systems, and
Computers, Pacific Grove, CA, USA, Oct. 2017, pp. 958–963.
[146] ——, “Comparison of limited feedback schemes for NOMA transmis-
sion in mmWave drone networks,” in Proc. IEEE 19th International
Workshop on Signal Processing Advances in Wireless Communications
(SPAWC), Kalamata, Greece, Jun. 2018.
[147] J. Talvitie, M. Valkama, G. Destino, and H. Wymeersch, “Novel
algorithms for high-accuracy joint position and orientation estimation
in 5G mmWave systems,” in Proc. IEEE Globecom Workshops (GC
Wkshps), Singapore, Dec. 2017.
[148] J. Zhao and W. Jia, “Efficient channel tracking strategy for mmWave
UAV communications,” Electronics Letters, vol. 54, no. 21, pp. 1218–
1220, Oct. 2018.
[149] A. Prokes, J. Vychodil, T. Mikulasek, J. Blumenstein, E. Zochman-
n, H. Groll, C. F. Mecklenbrauker, M. Hofer, D. Loschenbrand,
L. Bernado, T. Zemen, S. Sangodoyin, and A. Molisch, “Time-domain
broadband 60 GHz channel sounder for vehicle-to-vehicle channel
measurement,” in Proc. IEEE Vehicular Networking Conference (VNC),
Taipei, Taiwan, Dec. 2018.
[150] S. M. Abuelenin, “On the similarity between Nakagami-m fading
distribution and the Gaussian ensembles of random matrix theory,”
CoRR, vol. abs/1803.08688, Apr. 2018. [Online]. Available: http:
//arxiv.org/abs/1803.08688.
[151] S. Panic, M. Stefanovic, J. Anastasov, and P. Spalevic, Fading and
Interference Mitigation in Wireless Communications. CRC Press, Dec.
2013.
[152] T. S. Rappaport, Wireless communications: Principles and practice,
2nd ed. Prentice Hall, 2002.
[153] T. S. Rappaport, F. Gutierrez, E. Ben-Dor, J. N. Murdock, Y. Qiao, and
J. I. Tamir, “Broadband millimeter-wave propagation measurements
and models using adaptive-beam antennas for outdoor urban cellular
communications,” IEEE Transactions on Antennas and Propagation,
vol. 61, no. 4, pp. 1850–1859, Apr. 2013.
[154] R.-H. Zhang, Z.-C. He, H.-W. Wang, F. You, and K.-N. Li, “Study
on self-tuning tyre friction control for developing main-servo loop
integrated chassis control system,” IEEE Access, vol. 5, pp. 6649–6660,
Feb. 2017.
[155] J. Li, Y. Sun, L. Xiao, S. Zhou, and A. Sabharwal, “How to mobilize
mmwave: A joint beam and channel tracking approach,” in Proc. IEEE
International Conference on Acoustics, Speech and Signal Processing
(ICASSP), Calgary, AB, Canada, Apr. 2018, pp. 3624–3628.
[156] H. Yan, S. Chaudhari, and D. Cabric, “Wideband channel tracking for
mmWave MIMO system with hybrid beamforming architecture: (invit-
ed paper),” in Proc. IEEE International Workshop on Computational
Advances in Multi-Sensor Adaptive Processing (CAMSAP), Curacao,
Dutch Antilles, Dec. 2017.
[157] L. Song, Y. Li, Z. Ding, and H. V. Poor, “Resource management in
non-orthogonal multiple access networks for 5G and beyond,” IEEE
Network, vol. 31, no. 4, pp. 8–14, Jul. 2017.
[158] Z. Ding, Y. Liu, J. Choi, Q. Sun, M. Elkashlan, C.-L. I, and H. V.
Poor, “Application of non-orthogonal multiple access in LTE and 5G
networks,” IEEE Communications Magazine, vol. 55, no. 2, pp. 185–
191, Feb. 2017.
[159] Z. Xiao, L. Zhu, J. Choi, P. Xia, and X.-G. Xia, “Joint power allocation
and beamforming for non-orthogonal multiple access (NOMA) in
5G millimeter wave communications,” IEEE Transactions on Wireless
Communications, vol. 17, no. 5, pp. 2961–2974, May 2018.
[160] W. Tan, S. Li, C. He, W. Tan, X. Gu, and Z. Shi, “Achievable rate
of UAV-based communication with uniform circular arrays in Ricean
fading,” Physical Communication, vol. 32, pp. 185–191, Feb. 2019.
[161] L. Kong, L. Ye, F. Wu, M. Tao, G. Chen, and A. V. Vasilakos,
“Autonomous relay for millimeter-wave wireless communications,”
IEEE Journal on Selected Areas in Communications, vol. 35, no. 9,
pp. 2127–2136, Sep. 2017.
[162] G. S. Thomas and C. S. Chandran, “A review on autonomous relay for
millimeter wave wireless communications,” IOP Conference Series:
Materials Science and Engineering, vol. 396, p. 012043, Aug. 2018.
[163] S. Fu, L. Zhao, Z. Su, and X. Jian, “UAV based relay for wireless
sensor networks in 5G systems,” Sensors, vol. 18, no. 8, p. 2413, Jul.
2018.
[164] H. Ghazzai, M. B. Ghorbel, A. Kassler, and M. J. Hossain, “Trajectory
optimization for cooperative dual-band UAV swarms,” in Proc. IEEE
Global Communications Conference (GLOBECOM), Abu Dhabi, Unit-
ed Arab Emirates, Dec. 2018.
[165] Z. Khosravi, M. Gerasimenko, S. Andreev, and Y. Koucheryavy, “Per-
formance evaluation of UAV-assisted mmWave operation in mobility-
enabled urban deployments,” in Proc. International Conference on
Telecommunications and Signal Processing (TSP), Athens, Greece, Jul.
2018, pp. 150–153.
[166] E. Castaneda, A. Silva, A. Gameiro, and M. Kountouris, “An overview
on resource allocation techniques for multi-user MIMO systems,” IEEE
Communications Surveys & Tutorials, vol. 19, no. 1, pp. 239–284,
Firstquarter 2017.
[167] N. C. Luong, P. Wang, D. Niyato, Y. Wen, and Z. Han, “Resource
management in cloud networking using economic analysis and pricing
models: A survey,” IEEE Communications Surveys & Tutorials, vol. 19,
no. 2, pp. 954–1001, Secondquarter 2017.
[168] B. Li, Z. Fei, X. Xu, and Z. Chu, “Resource allocations for secure
cognitive satellite-terrestrial networks,” IEEE Wireless Communications
Letters, vol. 7, no. 1, pp. 78–81, Feb. 2018.
[169] L. Zhang, W. Huang, Q. Wu, and W. Cao, “Differential game theo-
retic approach for distributed dynamic cooperative power control in
cognitive radio ad hoc networks,” KSII Transactions on Internet and
Information Systems, vol. 9, no. 10, pp. 3810–3830, Oct. 2015.
[170] S. Li, Q. Ni, Y. Sun, G. Min, and S. Al-Rubaye, “Energy-efficient
resource allocation for industrial cyber-physical IoT systems in 5G
era,” IEEE Transactions on Industrial Informatics, vol. 14, no. 6, pp.
2618–2628, Jul. 2018.
[171] G. Zhang, H. Zhang, Z. Han, and G. K. Karagiannidis, “Spectrum
allocation and power control in full-duplex ultra-dense heterogeneous
networks,” IEEE Transactions on Communications, vol. 67, no. 6, pp.
4365–4380, Jun. 2019.
[172] M. Liu, F. R. Yu, Y. Teng, V. C. M. Leung, and M. Song, “Distributed
resource allocation in blockchain-based video streaming systems with
mobile edge computing,” IEEE Transactions on Wireless Communica-
tions, vol. 18, no. 1, pp. 695–708, Jan. 2019.
[173] J. Chakareski, S. Naqvi, N. Mastronarde, J. Xu, F. Afghah, and A. Razi,
“An energy efficient framework for UAV-assisted millimeter wave
5G heterogeneous cellular networks,” IEEE Transactions on Green
Communications and Networking, vol. 3, no. 1, pp. 37–44, Mar. 2019.
[174] S. Naqvi, J. Chakareski, N. Mastronarde, J. Xu, F. Afghah, and A. Raz-
i, “Energy efficiency analysis of UAV-assisted mmWave HetNets,”
IEEE ACCESS 41
in Proc. IEEE International Conference on Communications (ICC),
Kansas City, MO, USA, May 2018.
[175] M. Chen, W. Saad, and C. Yin, “Echo-liquid state deep learning for
360° content transmission and caching in wireless VR networks with
cellular-connected UAVs,” IEEE Transactions on Communications,
May 2019, Early Access. DOI: 10.1109/TCOMM.2019.2917440.
[176] Y. Zou, J. Zhu, X. Wang, and L. Hanzo, “A survey on wireless security:
Technical challenges, recent advances, and future trends,” Proceedings
of the IEEE, vol. 104, no. 9, pp. 1727–1765, Sep. 2016.
[177] G. Zhang, Q. Wu, M. Cui, and R. Zhang, “Securing UAV communi-
cations via joint trajectory and power control,” IEEE Transactions on
Wireless Communications, vol. 18, no. 2, pp. 1376–1389, Feb. 2019.
[178] H. Liu, S.-J. Yoo, and K. S. Kwak, “Opportunistic relaying for low-
altitude UAV swarm secure communications with multiple eavesdrop-
pers,” Journal of Communications and Networks, vol. 20, no. 5, pp.
496–508, Oct. 2018.
[179] Z. Li, M. Chen, C. Pan, N. Huang, Z. Yang, and A. Nallanathan, “Joint
trajectory and communication design for secure UAV networks,” IEEE
Communications Letters, vol. 23, no. 4, pp. 636–639, Apr. 2019.
[180] C. Liu, T. Q. S. Quek, and J. Lee, “Secure UAV communication in
the presence of active eavesdropper (invited paper),” in Proc. Interna-
tional Conference on Wireless Communications and Signal Processing
(WCSP), Nanjing, China, Oct. 2017.
[181] F. Shu, Y. Qin, R. Chen, L. Xu, T. Shen, S. Wan, S. Jin, J. Wang, and
X. You, “Directional modulation: A secure solution to 5G and beyond
mobile networks,” CoRR, vol. abs/1803.09938, Mar. 2018. [Online].
Available: http://arxiv.org/abs/1803.09938.
[182] J. Hu, S. Yan, F. Shu, J. Wang, J. Li, and Y. Zhang, “Artificial-noise-
aided secure transmission with directional modulation based on random
frequency diverse arrays,” IEEE Access, vol. 5, pp. 1658–1667, Jan.
2017.
[183] F. Shu, X. Wu, J. Hu, J. Li, R. Chen, and J. Wang, “Secure and precise
wireless transmission for random-subcarrier-selection-based directional
modulation transmit antenna array,” IEEE Journal on Selected Areas
in Communications, vol. 36, no. 4, pp. 890–904, Apr. 2018.
[184] F. Shu, S. Wan, Y. Wu, S. Yan, R. Chen, J. Li, and J. Lu, “Secure
directional modulation to enhance physical layer security in IoT
networks,” CoRR, vol. abs/1712.02104, Jan. 2017. [Online]. Available:
http://arxiv.org/abs/1712.02104.
[185] Y. Ding and V. Fusco, “A review of directional modulation technology,”
International Journal of Microwave and Wireless Technologies, vol. 8,
no. 7, pp. 981–993, Jul. 2015.
[186] T. Xie, J. Zhu, and Y. Li, “Artificial-noise-aided zero-forcing synthesis
approach for secure multi-beam directional modulation,” IEEE Com-
munications Letters, vol. 22, no. 2, pp. 276–279, Feb. 2018.
[187] J. Xiong, S. Y. Nusenu, and W.-Q. Wang, “Directional modulation
using frequency diverse array for secure communications,” Wireless
Personal Communications, vol. 95, no. 3, pp. 2679–2689, Jan. 2017.
[188] W. Yi, Y. Liu, E. L. Bodanese, A. Nallanathan, and G. K.
Karagiannidis, “A unified spatial framework for UAV-aided mmwave
networks,” CoRR, vol. abs/1901.01432, Jan. 2019. [Online]. Available:
http://arxiv.org/abs/1901.01432.
[189] B. Galkin, J. Kibilda, and L. A. DaSilva, “Backhaul for low-altitude
UAVs in urban environments,” in Proc. IEEE International Conference
on Communications (ICC), Kansas City, MO, USA, May 2018.
[190] R. Kovalchukov, D. Moltchanov, A. Samuylov, A. Ometov, S. Andreev,
Y. Koucheryavy, and K. Samouylov, “Analyzing effects of directionality
and random heights in drone-based mmWave communication,” IEEE
Transactions on Vehicular Technology, vol. 67, no. 10, pp. 10 064–
10 069, Oct. 2018.
[191] D. S. Vasiliev, A. Abilov, I. A. Kaysina, and D. S. Meitis, “Simulation-
based assesment of quality of service in UAV-assisted mmWave system
in crowded area,” in Proc. IEEE Conference of Open Innovations
Association (FRUCT), Helsinki, Finland, Nov. 2018.
[192] V. Va, X. Zhang, and R. W. Heath, “Beam switching for millimeter
wave communication to support high speed trains,” in Proc. IEEE 82nd
Vehicular Technology Conference (VTC-Fall), Boston, MA, USA, Sep.
2015.
[193] L. Zhang, F. Zhuo, and H. Xu, “A cross-layer optimization framework
for congestion and power control in cognitive radio ad hoc networks
under predictable contact,” EURASIP Journal on Wireless Communi-
cations and Networking, vol. 2018, pp. 1–23, Mar. 2018.
[194] Y. Zeng, R. Zhang, and T. J. Lim, “Throughput maximization for UAV-
enabled mobile relaying systems,” IEEE Transactions on Communica-
tions, vol. 64, no. 12, pp. 4983–4996, Dec. 2016.
[195] M. F. Sohail, C. Y. Leow, and S. Won, “Non-orthogonal multiple access
for unmanned aerial vehicle assisted communication,” IEEE Access,
vol. 6, pp. 22 716–22 727, Apr. 2018.
[196] N. Cheng, W. Xu, W. Shi, Y. Zhou, N. Lu, H. Zhou, and X. Shen, “Air-
ground integrated mobile edge networks: Architecture, challenges, and
opportunities,” IEEE Communications Magazine, vol. 56, no. 8, pp.
26–32, Aug. 2018.
[197] X. Lin, J. Xia, and Z. Wang, “Probabilistic caching placement in UAV-
assisted heterogeneous wireless networks,” Physical Communication,
vol. 33, pp. 54–61, Apr. 2019.
[198] M. Chen, W. Saad, and C. Yin, “Liquid state machine learning for
resource allocation in a network of cache-enabled LTE-U UAVs,”
in Proc. IEEE Global Communications Conference (GLOBECOM),
Singapore, Dec. 2017.
[199] M. H. Tareque, M. S. Hossain, and M. Atiquzzaman, “On the routing in
flying ad hoc networks,” in Proc. Federated Conference on Computer
Science and Information Systems, Lodz, Poland, Oct. 2015.
[200] L. Zhang, F. Zhuo, W. Huang, C. Bai, and H. Xu, “Joint opportunistic
routing with autonomic forwarding angle adjustment and channel
assignment for throughput maximization in cognitive radio ad hoc
networks,” Ad Hoc & Sensor Wireless Networks, vol. 38, no. 1-4, pp.
21–50, Sep. 2017.
[201] H. Xiong, X. Zhu, and R. Zhang, “Energy recovery strategy numerical
simulation for dual axle drive pure electric vehicle based on motor loss
model and big data calculation,” Complexity, vol. 2018, pp. 1–14, Aug.
2018.
[202] S. Bi, C. K. Ho, and R. Zhang, “Wireless powered communication: Op-
portunities and challenges,” IEEE Communications Magazine, vol. 53,
no. 4, pp. 117–125, Apr. 2015.
[203] Y. Wu, L. Qiu, and J. Xu, “UAV-enabled wireless power transfer with
directional antenna: A two-user case (invited paper),” in Proc. Inter-
national Symposium on Wireless Communication Systems (ISWCS),
Lisbon, Portugal, Aug. 2018.
[204] J. Park, H. Lee, S. Eom, and I. Lee, “Wireless powered communication
networks aided by an unmanned aerial vehicle,” in Proc. IEEE 88th
Vehicular Technology Conference (VTC-Fall), Chicago, IL, USA, Aug.
2018.
[205] Y. Sun, D. W. K. Ng, D. Xu, L. Dai, and R. Schober, “Resource allo-
cation for solar powered UAV communication systems,” in Proc. IEEE
International Workshop on Signal Processing Advances in Wireless
Communications (SPAWC), Kalamata, Greece, Jun. 2018.
[206] H. Yanikomeroglu, “Integrated terrestrial/non-terrestrial 6G networks
for ubiquitous 3D super-connectivity,” in Proc. ACM International
Conference on Modeling, Analysis and Simulation of Wireless and
Mobile Systems. Montreal, QC, Canada: ACM Press, Oct. 2018.
[207] X. Cao, P. Yang, M. Alzenad, X. Xi, D. Wu, and H. Yanikomeroglu,
“Airborne communication networks: A survey,” IEEE Journal on
Selected Areas in Communications, vol. 36, no. 9, pp. 1907–1926, Sep.
2018.
[208] E. C. Strinati, S. Barbarossa, J. L. Gonzalez-Jimenez, D. Kténas,
N. Cassiau, and C. Dehos, “6G: The next frontier,” CoRR, vol.
abs/1901.03239, May 2019. [Online]. Available: https://arxiv.org/abs/
1901.03239
[209] J. Liu, Y. Shi, Z. M. Fadlullah, and N. Kato, “Space-air-ground inte-
grated network: A survey,” IEEE Communications Surveys & Tutorials,
vol. 20, no. 4, pp. 2714–2741, Fourthquarter 2018.
[210] K. J. S. White, E. Denney, M. D. Knudson, A. K. Mamerides,
and D. P. Pezaros, “A programmable SDN+NFV-based architecture
for UAV telemetry monitoring,” in Proc. IEEE Annual Consumer
Communications & Networking Conference (CCNC), Las Vegas, NV,
USA, Jan. 2017, pp. 522–527.
[211] C. Rametta and G. Schembra, “Designing a softwarized network
deployed on a fleet of drones for rural zone monitoring,” Future
Internet, vol. 9, no. 1, Mar. 2017.
[212] R. Mijumbi, J. Serrat, J.-L. Gorricho, N. Bouten, F. D. Turck, and
R. Boutaba, “Network function virtualization: State-of-the-art and re-
search challenges,” IEEE Communications Surveys & Tutorials, vol. 18,
no. 1, pp. 236–262, Firstquarter 2016.
[213] X. Cheng, Y. Wu, G. Min, and A. Y. Zomaya, “Network function
virtualization in dynamic networks: A stochastic perspective,” IEEE
Journal on Selected Areas in Communications, vol. 36, no. 10, pp.
2218–2232, Oct. 2018.
[214] B. Yi, X. Wang, K. Li, S. k. Das, and M. Huang, “A comprehensive
survey of network function virtualization,” Computer Networks, vol.
133, pp. 212–262, Mar. 2018.
[215] W. Saad, M. Bennis, and M. Chen, “A vision of 6G wireless
systems: Applications, trends, technologies, and open research
IEEE ACCESS 42
problems,” CoRR, vol. abs/1902.10265, Feb. 2019. [Online]. Available:
http://arxiv.org/abs/1902.10265
[216] F. Tariq, M. R. A. Khandaker, K. Wong, M. A. Imran, M. Bennis, and
M. Debbah, “A speculative study on 6G,” CoRR, vol. abs/1902.06700,
Feb. 2019. [Online]. Available: http://arxiv.org/abs/1902.06700
[217] T. Braud, F. H. Bijarbooneh, D. Chatzopoulos, and P. Hui, “Future
networking challenges: The case of mobile augmented reality,” in Proc.
IEEE International Conference on Distributed Computing Systems
(ICDCS), Atlanta, GA, USA, Jun. 2017.
[218] C. Jiang, H. Zhang, Y. Ren, Z. Han, K.-C. Chen, and L. Hanzo,
“Machine learning paradigms for next-generation wireless networks,”
IEEE Wireless Communications, vol. 24, no. 2, pp. 98–105, Apr. 2017.
[219] M. Chen, U. Challita, W. Saad, C. Yin, and M. Debbah, “Machine
learning for wireless networks with artificial intelligence: A tutorial
on neural networks,” CoRR, vol. abs/1710.02913, Oct. 2017. [Online].
Available: http://arxiv.org/abs/1710.02913
[220] Y. Sun, M. Peng, Y. Zhou, Y. Huang, and S. Mao, “Application
of machine learning in wireless networks: Key techniques and open
issues,” CoRR, vol. abs/1809.08707, Mar. 2018. [Online]. Available:
http://arxiv.org/abs/1809.08707
[221] C. Zhang, P. Patras, and H. Haddadi, “Deep learning in mo-
bile and wireless networking: A survey,” IEEE Communica-
tions Surveys & Tutorials, Mar. 2019, Early Access. DOI:
10.1109/COMST.2019.2904897.
[222] R. Lowe, Y. Wu, A. Tamar, J. Harb, O. P. Abbeel, and I. Mordatch,
“Multi-agent actor-critic for mixed cooperative-competitive environ-
ments,” Advances in Neural Information Processing Systems, pp. 6379–
6390, Dec. 2017.
[223] X. Wang, Y. Han, C. Wang, Q. Zhao, X. Chen, and M. Chen,
“In-edge AI: Intelligentizing mobile edge computing, caching and
communication by federated learning,” CoRR, vol. abs/1809.07857,
Sep. 2018. [Online]. Available: http://arxiv.org/abs/1809.07857
[224] Y. Xu, R. Xu, W. Yan, and P. Ardis, “Concept drift learning with
alternating learners,” in Proc. International Joint Conference on Neural
Networks (IJCNN), Anchorage, AK, USA, May 2017, pp. 2104 –2111.
