Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.

28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

Millimeter Wave Spatial Statistical Channel Model for

High-altitude Military UAV Battlefield Situation Awareness
Jinsheng Yang1 , Huan Xi1 , and Zhou Pan2
1
Tianjin University
2
Tianjin Navigation Instrument Research Institute, Tianjin, China

February 27, 2023

Abstract
A high-altitude military UAV air-to-ground scenario model is established for the battlefield scenario without base station, the
MIMO antenna array technology is used to improve the channel communication performance, and the spatial statistical channel
model (SSCM) of this scenario is constructed in this paper. According to the simulation results, the dependence of the channel
on T-R separation distance, frequency, rain rate and antenna HPBW parameters was investigated, the large-scale fading and
small-scale fading characteristics was analyzed. By calculating and analyzing the condition number and rank of the channel
transmission matrix, the spatial multiplexing under different MIMO antenna arrays was investigated. The simulation results
can provide a theoretical basis for future high-altitude military UAV battlefield situation awareness frequency selection and
antenna design.

Hosted file
956915_0_art_file_10729651_rqfnkr.docx available at https://authorea.com/users/589412/
articles/626259-millimeter-wave-spatial-statistical-channel-model-for-high-altitude-
military-uav-battlefield-situation-awareness

1
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

2
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

3
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

4
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

5
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

6
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

7
Posted on 27 Feb 2023 — The copyright holder is the author/funder. All rights reserved. No reuse without permission. — https://doi.org/10.22541/essoar.167751582.28233263/v1 — This a preprint and has not been peer reviewed. Data may be preliminary.

8
 Radio Science

1 Millimeter Wave Spatial Statistical Channel Model for High-altitude Military UAV
2 Battlefield Situation Awareness
3 Jinsheng Yang1, Huan Xi1, and Zhou Pan2

1
4 School of Microelectronics, Tianjin University,Tianjin 300072, China
2
5 Tianjin Navigation Instrument Research Institute, Tianjin, China
6 Corresponding author: Zhou Pan(panli_an@163.com)

7 Key Points:
8 • A high-altitude military UAV air-to-ground scenario model is established for the
9 battlefield scenario without base station
10 • The MIMO antenna array technology and the spatial statistical channel model of
11 battlefield situation awareness scenario is constructed
12 • The simulation results can provide a theoretical basis for future high-altitude military
13 UAV frequency selection and antenna design
 Radio Science

14 Abstract
15 A high-altitude military UAV air-to-ground scenario model is established for the battlefield
16 scenario without base station, the MIMO antenna array technology is used to improve the
17 channel communication performance, and the spatial statistical channel model (SSCM) of this
18 scenario is constructed in this paper. According to the simulation results, the dependence of the
19 channel on T-R separation distance, frequency, rain rate and antenna HPBW parameters was
20 investigated, the large-scale fading and small-scale fading characteristics was analyzed. By
21 calculating and analyzing the condition number and rank of the channel transmission matrix, the
22 spatial multiplexing under different MIMO antenna arrays was investigated. The simulation
23 results can provide a theoretical basis for future high-altitude military UAV battlefield situation
24 awareness frequency selection and antenna design.

25 1 Introduction
26 With the continuous advancement of global informatization process, the Fifth-Generation
27 (5G) wireless communication technology has developed rapidly and been continuously applied.
28 As shown in Figure 1, all kinds of UAVs have become an important component and effective
29 supplement to realize the 6G air-space-ground-sea integrated communication system (Xiao et al.,
30 2022), and have become an important part of building SAGIN (space-air-ground integrated
31 network) (Bai et al., 2021). Full-spectrum resources will be further exploited, such as sub-6GHz,
32 millimeter wave (mmWave), terahertz, and optical bands (You et al., 2020). Among them,
33 millimeter wave bands provide new facilities for 5G mobile communication networks, as do
34 broadband wireless communication links, to meet the growing demand for higher data rates (Al-
35 Shammari et al., 2021).

36

37 Figure 1. Millimeter-wave band UAV application scenarios.

38 The modern battlefield environment represented by the electromagnetic environment has
39 become more and more prominent in its extensiveness, foundation, pillar, and commonality
 Radio Science

40 between the enemy and ourselves. In the military field, UAVs, as a new type of combat unit
41 relying on electromagnetic signals, can be widely used in war zones, spies, combat aircraft,
42 attack and missile launches, border surveillance, etc. (Bajracharya et al., 2022). Compared with
43 manned aircraft, the use of UAVs can save costs to a greater extent, but it will also bring a series
44 of problems such as UAV control and communication. Since base stations are vulnerable to
45 jamming and destruction in battlefield scenarios, the use of high-altitude military UAVs for
46 battlefield situation awareness through electromagnetic environment and electromagnetic actions
47 is an effective means of response.
48 Battlefield support from UAVs is inseparable from fortified battles and street battles in
49 modern cities. Among them, micro/small UAVs play a key role. However, UAVs that use
50 millimeter wave bands for communication at close range are prone to signal interference, while
51 military reconnaissance and combat integrated UAVs are usually at a higher altitude and far from
52 the target. It can deal with enemy omnidirectional jammers with insufficient power when the
53 enemy cannot determine its own position effectively. Meanwhile, it can carry out
54 electromagnetic perception, take the lead on the battlefield, and form an optimal battlefield
55 electromagnetic situation. Battlefield situation awareness through electromagnetism is an
56 important and fundamental part of Electromagnetic Battle Management (EMBM), through which
57 single or multiple combat operations can be closed-loop and efficient coordination. More
58 importantly, military UAVs can enhance Line-of-sight (LOS) wireless channels (Zhang, Wang,
59 & Poor, 2022; Li et al., 2020) and have clear advantages in terms of link budget and latency
60 compared to multi-hop communication (Tozer et al., 2000). The targets in the battlefield
61 situation awareness scenario are scattered in the complex natural and electromagnetic
62 environment including complex terrain, sea conditions, meteorological conditions, ground clutter,
63 sea clutter and various passive/active anthropogenic disturbances (Yang, Song, Xu et al., 2022),
64 as shown in Figure 2. In the battlefield, the communication function of the base station may be
65 stopped due to interference, and the reconnaissance and combat integrated UAV flying at high
66 altitude directly communicates information with the ground target for UAV control and
67 battlefield situation awareness. This type of reconnaissance and combat integrated UAV needs to
68 have a wider reconnaissance range, so its flight altitude is higher than that of civilian UAVs. For
69 example, the maximum flight altitude of the US MQ-9 UAV can reach 12,192 meters. China's
70 CH-5 UAV cruises at an altitude of about 15,000 meters and a combat altitude of about 8,000
71 meters. Therefore, the use of UAVs for battlefield situation awareness will be more extensive,
72 and its air-to-ground channel characteristics are of great significance in military scenarios.

73

74 Figure 2. The complex environment of situation awareness.

75
 Radio Science

76 Meanwhile, long-distance communication means that millimeter waves face high losses
77 during propagation, which usually need to be solved by using highly directional antennas or
78 using antenna arrays (Xiao et al., 2022; Bian et al., 2021; Wang et al., 2021). In the battlefield
79 scene without base station, the highly directional antenna is not suitable for the task of finding
80 targets in high-altitude military UAVs. From the point of view of UAV control, omnidirectional
81 antennas have better communication performance during movement, while directional antennas
82 perform poorly during movement (Khawaja et al, 2019), so the use of MIMO antenna arrays can
83 better meet the needs of battlefield situation awareness. MIMO antenna arrays can improve
84 channel reliability and channel capacity (Liao et al., 2020), and multiple antennas in a MIMO
85 system can be utilized in different ways effectively. The most common of these are spatial
86 multiplexing (SM) and beamforming (BF) (Sun et al., 2014).
87 In more scenarios as mentioned, high-altitude military UAVs and low-altitude UAVs can
88 work together on the battlefield, and micro/small UAV swarms are usually used at low altitudes
89 for close target strikes. It has obvious advantages such as indestructibility, low cost, and function
90 distribution in complex battlefield environments with powerful electronic jamming and anti-
91 aircraft firepower, which involves non-terrestrial networks (NTN) communication problems
92 (Zhu, 2020). In the 3GPP specification, it is proposed that NTN can perform high-capacity
93 transmission in millimeter wave, which requires huge bandwidth and can achieve high-speed
94 connection through high-directional antennas (Traspadini et al., 2022). Therefore, after receiving
95 electromagnetic signals, military UAVs can transmit electromagnetic signals to UAV swarms to
96 complete efficient cooperative operations (Bajracharya et al., 2022; Zhang, Zhu, & Poor, 2022a).
97 Meanwhile, high-altitude UAVs have greater coverage and smaller path loss than low-altitude
98 UAVs compared to high-altitude platforms.
99 In recent years, the channel modeling researches of millimeter wave and MIMO antenna
100 arrays mainly focus on indoor short-range communication scenarios, outdoor device-to-device
101 (D2D) communication scenarios and low-altitude civil UAV-to-ground communication scenarios.
102 The research on the application of ray tracing method to assist indoor positioning in millimeter
103 band map was analyzed in reference (Kanhere et al., 2019). Satellite-to-UAV MIMO
104 communication channel for LOS scenarios at terahertz frequencies was analyzed in reference
105 (Geraci et al., 2022). Model of the communication channel of the UAV with the offshore unit at
106 sub-6GHz was analyzed in reference (Liu, Wang, Chang et al., 2021; Wang et al., 2020). V2V
107 channel model for mmWave MIMO channels was analyzed in reference (Bian et al., 2021;
108 Huang et al., 2020). The above research results provide a theoretical basis and feasibility support
109 for mmWave MIMO channels. A linear relationship between rain rate and attenuation for Ku/C
110 band is presented in reference (Tian & Shi, 2020).
111 At present, in the study of UAV air-to-ground channel, an UAV trajectory tracking
112 control with base station Non-line-of-sight (NLOS) scene was analyzed in reference (Zhang, Zhu,
113 & Poor, 2022b), the Offload problem of UAV as an edge server in the construction of Internet of
114 Vehicles was analyzed in reference (Liu, Liu, Qu et al., 2021), the three-dimensional
115 nonstationary model based on geometry and the GBSM channel model to investigate low-
116 altitude UAV ground channel modeling were analyzed in reference (Chang et al., 2021; Liu et al.,
117 2019), a shaded double scattering channel to investigate the dynamic propagation conditions of
118 LOS scene was analyzed in reference (Bithas et al., 2020).
119 In terms of battlefield situation awareness, the evolution trend of battlefield situation
120 awareness of unmanned platforms was summarized in reference (Yang, Yang, Zhang et al.,
 Radio Science

121 2021), a new kind of vehicle-mounted battlefield perception system design and implementation
122 scheme is presented in reference (Gao et al., 2021). The signal propagation loss model proposed
123 by ITUR P.1546 is applied and implemented in reference (Shen et al., 2017). The ITU-R P.1546
124 model is a ground-based service node-to-surface prediction method suitable for the prediction
125 range of 30 MHz to 3000 MHz frequency. It is suitable for effective transmission of tropospheric
126 radio circuits on land, sea and/or land-sea hybrid paths with a small antenna height of 3000 m
127 and a path length of 1 to 1000 km, and has wide applicability.
128 The above research results focus more on millimeter-wave ground equipment
129 communication, indoor communication or low-altitude UAV (below 150 meters) ground
130 communication, and lack the research on air-to-ground channel characteristics of high-altitude
131 military UAV in the scenario of no base station. At present, the studies supports the feasibility of
132 MIMO antenna array modeling of millimeter wave of UAVs, but lacks channel modeling for
133 long-distance communication application scenarios.
134 In order to fill this gap, this paper uses the spatial statistical channel model modeling
135 method to establish the air-to-ground communication channel of high-altitude military UAV, and
136 simulates the scenario of receiving control signals and conducting electromagnetic situation
137 perception in the case of base station failure in the battlefield, and gets a large amount of
138 statistics through simulation. The channel parameters such as AOA power spectrum, AOD power
139 spectrum, omnidirectional and directional received power, path loss, small-scale power delay
140 spectrum, rms delay spread, Rice K-factor, path loss index, shadow fading standard deviation
141 and other channel parameters were statistically analyzed to comprehensively characterize the
142 channel characteristics. The analysis results provide a reference for the battlefield situation
143 awareness communication link design of high-altitude military UAVs in the future, and
144 supplement the communication of high-altitude UAVs in the air-space-ground-sea integration
145 under the 6G framework effectively. The main research contents of this paper are as follows:
146 • In this paper, a high-altitude military UAV air-to-ground scenario model is established
147 for the battlefield scenario without base station, and a MIMO antenna array is used to
148 improve the channel communication performance, and the spatial statistical channel
149 model of this scenario is constructed. The channel was simulated to get statistics, the
150 dependence of the channel on T-R distance, frequency, rain rate, antenna HPBW
151 parameters, and conduct large-scale fading and small-scale fading statistical analysis was
152 investigated to get channel characteristics.
153 • According to the statistics derived by simulation, the spatial multiplexing of channels in
154 the air-to-ground scenario of high-altitude military UAVs in the air-to-ground scene of
155 high-altitude military UAVs under different MIMO antenna arrays is investigated by
156 calculating and analyzing the condition number and rank of the channel transmission
157 matrix.
158 The remainder of this study is organized as follows. Section 2 introduces the relevant
159 theoretical basis, including channel fading model parameters, MIMO antenna array technology,
160 beamforming and half-power-beamwidth (HPBW). Section 3 models the high-altitude military
161 UAV air-to-ground communication channel without base station, characterizes modeling
162 scenarios, and analyzes the channel fading characteristics. Section 4 introduces the configuration
163 of various simulation parameters, and analyzes the simulation results of the channel model
164 parameters. Section 5 provides a conclusion.
 Raadio Science

165 2 Releva
ant Theoretiical Basis

166 2.1 Channel Fading

F Modeel Parameterrs
167 Inn addition too direct radiaation, radio waves
w encouunter an obsttacle, accordding to their oown
168 waveleng gth and the relative
r size of the obstaccle to reflectt, diffract, sccatter, whichh will have a
169 certain lo
oss of radio wave
w energy n signal fadiing. Accordiing to the intterval of the field
y, resulting in
170 strength change
c of th
he received signal,
s radio waves can bbe divided innto large-scaale fading annd
171 small-scaale fading. The
T size of th he interval heere means thhe relationshhip between tthe distance we
172 observe the
t movemen nt of the mo obile station and
a the wavvelength of thhe signal itseelf. The
173 classificaation of wireeless fading channels
c can
n be summarrized in the F Figure 3.

174

175 Figure 3. Wireless ch

hannel fadin
ng classificattion.
176 According
A he characterisstics of largee-scale fadinng are descriibed by path loss
to Figure 2, th
177 and shad
dow fading. Large-scale
L average
a path
h loss measuures the averaage fading oof the signal
178 between the transmittter and receiiver, definedd as the diffeerence betweeen the effecctive transmiit
179 power annd the averag
ge received power.
p Shaddow fading reefers to the fformation off a shadow aarea
180 behind th
he obstructio
on due to thee obstructionn of obstacless such as terrrain undulattions or tall
181 building groups when n radio signaals are transm
mitted in chaannels at meesoscale distances, resultting
182 in random
m changes inn the averagee power of th he received signal. Its faading characcteristics
183 approxim
mately followw a lognormaal distributio on (Rappapoort, 1996).
184 Small-scale fading
fa refers to the rapid change in aamplitude, phhase, or multtipath delay of a
185 radio sign nal after a sh
hort time or distance (sevveral waveleengths). Thiss fading is caaused by thee
186 same tran nsmission signal travelin ng along diffferent paths, and the signnals arrivingg at the receivver
187 at differeent moments (or phases) are superim mposed on eaach other. Thhe signals thaat arrive at thhese
188 different paths are caalled multipaath signals, and
a multipathh signals incclude direct ppaths and
189 multiple path signals such as refllection, diffrraction and sscattering gennerated by thhe presence of
190 scattererss. Small-scalle channel modeling
m maiinly consideers the modeling of time dispersion
191 parameteers, frequenccy dispersion n parameters and spatial dispersion pparameters (C Chetlur &
192 Dhillon, 2017).
193 The
T type of sm mall-scale faading depend ds on the chharacteristicss of the transsmitted signaal
194 (signal baandwidth an
nd symbol peeriod) and ch hannel charaacteristics (ddelay spread and Dopplerr
195 spread). The
T relationship between n signal paraameters and channel parrameters deteermines thatt
196 different transmitted signals will undergo diffferent fadingg characterisstics. Accordding to the ddelay
 Radio Science

197 spread of the channel, the channel can be divided into flat fading channel and frequency selective
198 fading channel. According to the Doppler expansion of the channel, the channel can be divided
199 into fast fading channel and slow fading channel.
200 In the actual scene propagation process, the signal will experience both large-scale fading
201 and small-scale fading, large-scale fading generally affects the network coverage ability of the
202 wireless system, and small-scale fading affects the communication quality of the communication
203 system (Khuwaja et al., 2018).
204 For UAV air-to-ground channel analysis, this paper performs channel modeling for the
205 simulation measurement of path loss, received power, shadow fading, AOA angle, AOD angle,
206 RMS delay spread, antenna gain, pass loss exponent and other parameters. The symbols and
207 descriptions of each parameter are shown in Table 1.

208 Table 1. The Symbols and Descriptions of Channel Parameters

Parameter Symbol Description
The separation distance
T-R Separation
from the transmitter to the
Distance (m)
receiver.
The time it takes for an
electromagnetic or optical
Time Delay (absolute
signal to travel a certain
propagation time) (ns)
distance in the transmission
medium.
Standard deviation of power
delay distribution, the most
RMS Delay Spread (ns)
commonly used temporal
dispersion parameter.
Received Power (dBm) The power RX received.
The loss caused by the
propagation of radio waves
in space. It is caused by the
Path Loss (dB) radiated diffusion of the
transmitted power and the
propagation characteristics
of the channel.
The path loss exponent
ranges from 2 to 6, with 2
Path Loss Exponent representing free space and
6 representing severe
obstruction.
Shadow Fading Obstacles attenuate signal
Standard Deviation power by absorption,
(dB) reflection, scattering and
 Radio Science

diffraction, causing shadow

fading. The range is from
5dB to 12dB, and the typical
value is 8dB.
An important parameter to
characterize the degree of
channel fading, and has an
Rician K-factor important impact on link
budget, transmit diversity
power allocation, and
adaptive receiver design.
An editable parameter
denoting the
azimuth/elevation half-
TX Ant. HPBW
power-beamwidth (HPBW)
of the TX antenna (array) in
degrees.
TX Ant. Gain (dBi) TX antenna gain.
An editable parameter
denoting the
azimuth/elevation half-
RX Ant. HPBW
power-beamwidth (HPBW)
of the RX antenna (array) in
degrees.
RX Ant. Gain (dBi) RX antenna gain.
209 In the lower corner mark of n and σ, ‘omni’ represents omnidirectional, ‘dir’ represents
210 directional, ‘dir − best’ represents the direction with the strongest received power.
211 In the table above, the omnidirectional received power can be described as
(1)

212 Where i, j, k, m denotes unique pointing directions indices in azimuth and elevation at the
213 TX and RX, respectively. AZ , , EL , , AZ , , EL , represents the TX azimuth and elevation
214 angles, and the RX azimuth and elevation angles, respectively (Mou et al., 2019).
215 Path loss can be described as
(2)

216 Where FSPL implies free space path loss, AT is related to the attenuation factor, X is a
217 zero-mean Gaussian random variable with a standard deviation σ in dB, d signifies the free
218 space reference distance (Teixeira et al., 2021).
219 According to the Friis transmission equation, free space path loss can be described as
 Radio Science

(3)

220 Where c represents the electromagnetic wave propagation speed (approximate speed of
221 light). The unit of d is km, the unit of f is MHz (Jawhly & Chandra, 2021).
222 From another point of view, the directional path loss is equal to the transmit power plus
223 the TX and RX antenna gains, minus the directional received power (Sun et al., 2017), it can be
224 written as
(4)

225 In addition, the relationship between time delay and T-R separation distance can be
226 described as
(5)

227 The time delay of the direct path is T-R separation distance divided by the
228 electromagnetic wave propagation speed. In the presence of multipath components, the multipath
229 component has a slightly larger time delay than the direct path.
230 Power Delay Profile (PDP) describes the dispersion of a channel over time, which is the
231 expectation of received power at a certain delay. It is calculated by averaging the channel
232 impulse response CIR (Channel Impulse Response) in the time domain and squaring it (Yang, Li,
233 & Xu, 2021; Yang, & Yuan, 2022).
234 The parameters used to describe time expansion are mean excess delay, RMS delay
235 spread and X dB, which are all related to PDP.
236 Mean excess delay is the first moment of PDP. It is defined as

(6)

237 Where p is the n multipath power, τ is the n extra delay.

238 RMS Delay Spread is an important latency domain parameter used to quantify the
239 dispersion effect caused by radio waves propagating in the delay domain due to the sensitivity of
240 the communication system, and describes the delay statistics of the multipath effect of the
241 channel, as shown in Equation 7,

(7)

242 The Rician K-factor can reflect the effect of direct paths and multipath components on
243 the channel, defined as the ratio of the principal signal power (the strongest) to the variance of
244 the multipath components, as shown in Equation 8,
(8)
 Radio Science

245 2.2 MIMO Antenna Array Techology

246 MIMO technology can exponentially increase the capacity and spectrum utilization of
247 communication systems without increasing bandwidth, and more importantly, improve long-
248 distance communication performance, which is suitable for the channel requirements of the
249 scenarios investigated in this paper.
250 In the arrangement of MIMO antennas, there are usually four common ways of ULA
251 (uniform linear array), URA (uniform rectangular array), UCA (uniform circular array), CCA
252 (cylindrical conformal array) (Xiao et al., 2022; Wang et al., 2021). This paper uses ULA and
253 URA arrangement to simulate and measure channel characteristics, and further describes the
254 channel characteristics in the form of HPBW azimuth and elevation.
255 As shown in Figure 4, according to the number of antennas, it can be divided into four
256 types: SISO, SIMO, MISO and MIMO. Adding antennas on the transceiver can provide
257 multiplexing ideally.

258

259 Figure 4. Four types of transmission divided by the number of receivers and transmitters, (a)
260 SISO; (b) SIMO; (c) MISO; (d) MIMO.
261 Space division multiplexing is a method that uses multiple antennas to multiplex different
262 transmission paths in space to send multiple copies of different data in parallel to increase
263 capacity. Figure 4(d) shows a 2×2 MIMO channel formed by two antennas on both transceiver
264 and receiver, and four transmission paths can be established ideally, reaching 4 times the
265 capacity of SISO. However, in reality, due to the same fading and interference of multiple paths,
266 due to spatial correlation, it may not be possible to transmit different signals in multiple ways,
267 resulting in poor spatial multiplexing effect.
268 In Figure 4(d), the transmitter data is recorded as X , X , the receiver data is recorded as
269 Y , Y , in order to judge the independence of the transmission path, the transmission channel
270 matrix H is established as

(9)
 Radio Science

271 So there is

(10)

272 When referring to output Y, input X is pre-encoded, the above equation can be varied to

(11)

273 λ and λ are transfer coefficients. If neither is 0, the matrix rank is 2, which means that
274 the system has two relatively independent spatial channels and can send and receive two data
275 channels at the same time; If one of them is 0, the matrix rank is 1, which means that the
276 transmission space of this 2×2 MIMO system is very correlated, and it has degenerated from
277 MIMO to SISO or SIMO, and can only send and receive data at the same time. To determine its
278 channel capacity, its condition number was determined as
(12)

279 When the condition number is 1, the transmission coefficients of the two channels are
280 equal and the channel independence is high. When the condition number is greater than 1, the
281 transmission quality of the two channels is different, the capacity of this 2×2 MIMO system is
282 between 1 and 2 times that of the SISO system.
283 Orthogonal Frequency Division Multiplexing (OFDM) is a specific implementation of
284 multi-carrier modulation (MCM), which utilizes MCM to reduce the transmission rate, increase
285 the symbol period, effectively reduce and eliminate the influence of inter-symbol interference
286 and frequency-selective weakening, increasing the utilization of the spectrum and improving the
287 performance of systems (Jia et al., 2020).
288 Take ULAs at both the transmitter and receiver for example, the equation for generating
289 such a channel coefficient is
(13)

290 H , (f) is expressed as the channel coefficient between the m transmitting antenna and
291 the k receiving antenna when the subcarrier is f. p represents the p resolvable multipath
292 component. α is the amplitude of the channel gain, Φ denotes the phase of the multipath
293 component, τ represents the time delay , d and d are the antenna element spacing at the
294 transmitter and receiver, respectively. ф and φ denote the azimuth angle of departure and angle
295 of arrival, respectively. In a MIMO-OFDM system, each subcarrier f corresponds to a channel
296 transmission matrix H with the product of the number of transmit antennas and the number of
297 received antennas (Sun et al., 2017; Ji et al., 2022).

298 2.3 Beamforming and HPBW

299 The purpose of beamforming is to concentrate the energy radiated by the antenna. It can
300 form specific beams and transmit data through the beams. Beamforming is mainly based on
301 MIMO antenna arrays, and antenna beam patterns can be obtained by beamforming. Antenna
302 beam pattern indicate that the power or amplitude of a signal changes with angle. Antenna beam
 Radio Science

303 pattern is generally a function of antenna shape, size, and frequency. The angles here are
304 expressed in azimuth angle and elevation angle. These angles are shown in the Figure 5.

305

306 Figure 5. Azimuth Angle and Elevation Angle.

307 The beam pattern usually consists of a main lobe, some side lobes, and some nulls, as
308 shown in Figure 6. The y-axis in the Figure 8 is the signal amplitude normalized to the maximum
309 value of the main lobe. HPBW is a measure of beam width and is important for beamforming
310 performance analysis. The half-power beamwidth of the main lobe on a given main section is an
311 angular region in the direction of maximum radiation in which the relative radiated power of the
312 antenna is greater than one-half. In this example, HPBW is the angular distance (azimuth angle
313 or elevation angle) between two antenna pattern points, where the power becomes half of its
314 maximum. HPBW can be found by drawing the line y = .
√

315

316 Figure 6. Antenna beam pattern.

317 3 Channel Scenario Model

318 In this application scenario, this paper builds a scene model as shown in Figure 7 , mainly
319 investigates the air-to-ground communication mode of high-altitude military UAV in the open
320 scene without base station, and considers the foliage loss caused by vegetation.
 Radio Science

321

322 Figure 7. Air-to-ground simulation scenario of high-altitude military UAV, considering foliage
323 loss.

324 3.1 Modeling Scenario Characterization

325 In the basic scene construction, this paper uses a free space propagation model, which is
326 closer to the type in the battlefield situation. This is a common large-scale loss model, where the
327 PLE (path loss exponent) parameter provides an insight into the path loss based on the
328 environment, with a PLE value of 2 for free space and a PLE value of 4 for the asymptotic
329 bidirectional ground reflection propagation model (Sun et al., 2017). The reference distance
330 normalized to 1 meter makes it easy to compare measurements and models, provides a standard
331 definition for PLE, while enabling intuitive and fast calculation of path losses. Compared with
332 the existing alpha-beta-gamma (ABG) path loss model used in the 3GPP/ITU channel model,
333 this model has better model parameter stability, better prediction performance over a wide range
334 of microwave and millimeter wave frequencies, distances, and scenarios, and fewer parameters
335 (Sun et al., 2017).
336 From the fading causes and the model constructed in this paper, this channel has both
337 large-scale fading and small-scale fading, and the large-scale fading characteristics are more
338 obvious. In order to solve the problem of high propagation loss, it is common to use a highly
339 directional antenna or use an antenna array. When searching for targets, the problem of setting
340 up highly directional antennas is difficult to solve, so the using of antenna arrays is a better
341 choice. In the selection of MIMO antenna, since the reconnaissance and combat integrated UAV
342 needs to have the function of cruising strike at the same time, the results of the omnidirectional
343 channel model and the directional channel model should be considered. Military UAVs require
344 highly directional antennas to ensure the huge bandwidth to reach such ultra-high data rates and
345 meet massive wireless data traffic demands (Zhang et al., 2021). Meanwhile, the directional
346 channel model is of great significance for the correct implementation of MIMO systems, so this
347 paper further investigates the channel characteristics of the direction with the strongest energy
348 received at the receiver.
 Radio Science

349 3.2 Theoretical Analysis of Channel Fading Characteristics

350 In this section, the large-scale fading and small-scale fading characteristic parameters of
351 the channel was analyzed through formulas, which provides a theoretical basis and analysis
352 source for subsequent simulation work.

353 3.2.1 Large-scale Fading

354 According to Equation 3, the free space path loss can be calculated by T-R separation
355 distance and the frequency. For example, when d is 8.2399 km, f is 28000 MHz, the
356 theoretical value of FSPL will be -139.7 dB. According to Equation 4, if both the TX power and
357 the antenna array are determined, the sum of the directional path loss and the received power will
358 be a constant value.
359 Since in the actual transmission scenario, the factors that cause the random attenuation of
360 the signal are generally unknown, so only statistical models can be used to characterize this
361 random attenuation, the most commonly used model is the lognormal shadow fading model,
362 which has been confirmed by measured data and can be used to model the change of received
363 power in outdoor and indoor wireless propagation environments.
364 The lognormal shadow fading model assumes the ratio of transmit power (P ) and
365 received power (P ) as a random variable with a lognormal distribution. Set the ratio to K , it can
366 be written as
(14)

367 The lognormal distribution can be written as

(15)

368 Where ξ = . A large number of outdoor channel measurements show that the shadow
369 fading standard deviation σ range from 4 dB to 13 dB.

370 3.2.2 Small-scale Fading

371 According to Equation 5, when d is 8239.9 m, the time delay of the directional path
372 will be 2.749×104 ns. As described in Section 2.1, power delay profile describes the dispersion
373 of a channel over time, which is the expectation of received power at a certain delay. It refers to
374 the phenomenon of signal time diffusion due to multipath propagation. The reason for this is that
375 the time it takes for the transmitted signal to travel through different paths to the receiving point
376 varies.
377 The small-scale PDP generally follows an exponential distribution. The PDP distribution
378 can be written as
(16)

379 Where T is the average of the multipath time delay. The theoretical correlation between
380 time delay (or call it absolute propagation time) and received power can be can be represented by
381 Figure 8.
 Radio Science

382

383 Figure 8. The theoretical Power Delay Profile.

384 Under the law of negative exponential, according to Equation 6 and 7, the mean excess
385 delay and RMS delay spread can be derived as
(17)

386 and

(18)

387 Both the mean excess delay (τ) and RMS delay spread (σ ) are equal to the average of
388 the multipath time delay (T). In addition, the multipath delay is mainly distributed in the range of
389 0 to 2T.

390 4 Channel Characterization and Simulation Analysis

391 The current wireless channel modeling methods mainly include statistical modeling
392 methods, deterministic modeling methods and semi-deterministic modeling methods. The
393 statistical modeling method relies on channel measurement, summarizes the statistical
394 characteristics and empirical formulas of channels through a large number of measured statistics,
395 and extends to other scenarios with similar structures. Spatial statistical channel model is a
396 common statistical channel model. SSCM has high prediction accuracy and fast processing speed
397 in applicable environments, so it has become the preferred signal propagation loss model for
398 real-time systems (Shen et al., 2017).
399 In this paper, some parameters are used as variables, such as T-R separation distance,
400 frequency, rain rate, number of MIMO antennas, azimuth HPBWs and elevation HPBWs of RX
401 antennas. Based on a large number of simulation statistics, the channel of the model is
402 comprehensively characterized by using SSCM.

403 4.1 Simulation Parameter Configuration

404 In this paper, the channel parameters of high-altitude military UAVs are modeled by
405 using the open-source software NYU Simulator (NYUSIM) for millimeter wave channel
406 modeling. This is a simulation software for SSCM that uses OFDM modulation. Its latest version,
407 V3.1, contains three important channel characteristics: spatial consistency, human blockage
 Radio Science

408 parameters, and outdoor-to-indoor (O2I) penetration loss. Meanwhile, it contains the calculation
409 ability of path loss and shadow fading models and RMS delay propagation parameters, which
410 can be well applied to the simulation of mmWave MIMO channels. For more information about
411 NYUSIM, see reference (Sun et al., 2017) and https://wireless.engineering.nyu.edu/nyusim/.
412 Due to its lack of data analysis capabilities and partial visualization output, MATLAB is
413 used to visualize the simulation statistics generated by its multiple runs for the investigate and
414 analysis of channel characteristics.
415 Set the distance between transmitter TX and receiver RX to be in the range of 500 meters
416 to 10,000 meters, corresponding to possible UAV and ground unit distances. The RF bandwidth
417 is set to 800 MHz to meet the needs of military UAV use, the TX power is set to 50 dBm, the
418 atmospheric pressure is 1013.25 mbar, the humidity is 50%, the temperature is 20°C, considering
419 the foliage loss caused by vegetation, the total distance that the transmitted signal travels within
420 foliage is set to 0.1 m, and the foliage attenuation is 0.4 dB/m.

421 4.2 T-R Separation Distance

422 At a frequency of 28 GHz, according to reference (Rappaport et al., 2013, 2015), set the
423 HPBW of the MIMO antenna as a definite value. Set the azimuth and elevation HPBW of the
424 antenna to 10.9◦ and 8.6◦ at both TX and RX, respectively. The antenna spacing is 0.5 times the
425 wavelength, the polarization direction is Co-Pol(co-polarization), the rain rate is 0 mm/hr.
426 Considering that the target is outdoors, ignore the O2I loss. Assume that the MIMO antennas are
427 all URA 2×2 arrangement, forming a 4×4 MIMO antenna array. In this way, channel
428 characteristics with T-R separation distances of 500 to 10,000 meters can be simulated.
429 Taking the T-R separation distance at 8239.9 meters and 2990.6 meters as an example,
430 the omnidirectional power delay profile can be derived as shown in Figure 9(a) and 9(d). The X-
431 axis is the absolute propagation time (ns), the Y-axis is the received power (dBm), each line in
432 the figure represents a multipath component, and also corresponds to the multipath component of
433 the arrival threshold shown in the 3-D AOA power spectrum in Figure 9(b) and 9(e), respectively.
434 It can be seen that this channel has more multipath components, such as 29 at a distance of
435 8239.9 meters, 64 at a distance of 2990.6 meters, and the power the receiver received decreases
436 exponentially with the increase of propagation time. This result ties well with the theoretical
437 result in Section 3.2.2. Figure 9(c) and 9(f) shows the directional PDP with strongest power,
438 which is based on omnidirectional PDP, derived from HPBW and antenna gain. Its multipath
439 case is similar to omnidirectional, and its characteristics can be used to guide the design of
440 directional antennas at specific altitudes.
 Radio Science

441

442 Figure 9. Omnidirectional power delay profile, 3-D AOA power spectrum, directional PDP with
443 strongest power of 8239.9 m and 2990.6 m. (a) omnidirectional power delay profile, 8239.9 m;
444 (b) 3-D AOA power spectrum, 8239.9 m; (c) directional PDP with strongest power, 8239.9 m; (d)
445 omnidirectional power delay profile, 2990.6 m; (e) 3-D AOA power spectrum, 2990.6 m; (f)
446 directional PDP with strongest power, 2990.6 m.
447 In addition, the small scale PDPs of each array of the receiving antenna under the model
448 can be derived as shown in Figure 10, where the third dimension is the interval of the antenna in
449 the antenna array. Figure 10 shows the PDP of the four antennas, and the PDP is basically the
450 same, but there are some differences in delay. At a distance of 8239.9 meters, according to
451 Section 3.2.2, the theoretical value of direct path delay in this scenario is about 2.749×104 ns.
 Radio Science

452

453 Figure 10. Small Scale PDPs of each array of the receiving antenna.
454 The total propagation delay is distributed from about 2.747×104 ns to 2.763×104 ns, and
455 the smallest propagation delay component corresponds to the direct path. This result ties well
456 with the theoretical value. In terms of trends, the decay situation is also the same as above. In
457 addition, the omnidirectional and directional path loss value scatter plot and fitted plot line are
458 generated by NYUSIM as shown in Figure 11. Due to the presence of multipath components, RX
459 receives more multipath components and energy, and the directional path loss and PLE will
460 always be greater than the omnidirectional path loss and PLE. PLE will reach 2.7 in this scenario.

461
 Radio Science

462 Figure 11. Scatterplots and fitted plot lines of path loss values generated by NYUSIM.
463 The omnidirectional received power and omnidirectional path loss at different distances
464 are shown in Figure 12, and they are quadratically fitted, where R square is 0.7508 and the
465 residual value is 53.37, which can get a relatively good fitting effect.

466

467 Figure 12. The omnidirectional receiving power and omnidirectional path loss varies with
468 distance.
469 The empirical cumulative distribution function (Empirical CDF) of the RMS delay spread
470 derived from the simulation is shown in Figure 13(a). Its values are distributed from about 0 to
471 27 ns, of which 90% of the RMS delay spread value is less than 22 ns, and the median value is
472 about 17.26 ns. It can be seen from the figure that the RMS delay spread is large due to the
473 influence of high-altitude distance in this scenario. Moreover, the higher RMS delay spread part
474 accounts for a relatively large part, which indicates that there is an obvious multipath effect other
475 than the direct component in this channel, so it is necessary to pay attention to the small-scale
476 fading of the channel.
477 The empirical cumulative distribution function of the omnidirectional Rician K-factor
478 derived from the simulation is shown in Figure 13(b). Its values range from -8 to 12 dB, with 90%
479 of the Rician K-factor values less than 3.35 dB and the median value being about -1.74 dB. The
480 Rician K-factor value is small, which means that this channel can maintain a large channel
481 capacity because it is not completely dominated by direct paths, and the multipath component
482 has a large contribution. In addition, the simulation shows that the small-scale fading
483 characteristics of the channel conform to the Gaussian distribution.
 Radio Science

484

485 Figure 13. The Empirical CDF of RMS Delay Spread (ns) and Rician K-factor (dB). (a) The
486 Empirical CDF of RMS Delay Spread (ns); (b) The Empirical CDF of Rician K-factor (dB).

487 4.3 Frequency

488 From the point of view of UAV controlling, in order to prevent inter-signal interference,
489 the frequency is selected from 28 GHz to 55 GHz. Considering the frequency dependence at the
490 distance of 8239.9 m at the transmitter and receiver, it can be derived by simulation that the
491 omnidirectional RMS delay spread is 17.9 ns, the directional RMS delay spread is 3.1 ns, and the
492 antenna gain of both the transmitter and receiver is 24.9 dBi. Its omnidirectional received power
493 and omnidirectional path loss are shown in Figure 14(a), and directional receive power and
494 directional path loss are shown in Figure 14(b). It can be concluded that when the frequency is
495 less than 50 GHz, the two curves are linear; when the frequency is greater than 50 GHz, the
496 omnidirectional received power (dBm) decreases rapidly, and the omnidirectional path loss (dB)
497 increases rapidly. Therefore, the frequency selection of high-altitude military UAVs should be as
498 much as possible not greater than 50 GHz to maintain a linear relationship and avoid excessive
499 channel fading.

500
 Radio Science

501 Figure 14. Omnidirectional/Directional received power and path loss at 28 to 55 GHz, (a)
502 omnidirectional; (b) directional.
503 At a distance of 8239.9 m, the frequency dependence of omnidirectional and directional
504 PLE is shown in Figure 15(a), which can ensure good PLE stability below about 49 GHz. The
505 omnidirectional PLE is about 2.2 and the directional PLE is about 2.4 in this range.
506 By fitting at a distance of 500 to 10,000 meters, considering the path loss exponent and
507 the shadow fading standard deviation, the curve is derived as shown in Figure 15(b) and 15(c). It
508 can be seen that in the research of high-altitude UAV-to-ground channels, frequencies below
509 about 49GHz can ensure better channel stability, and the loss above 52 GHz will be significantly
510 improved.

511

512 Figure 15. PLE and shadow fading standard deviation varies with frequency, (a) omnidirectional
513 and directional PLE at 28 to 55 GHz of 8239.9 m; (b) PLE at 28 to 55 GHz of 500 m to 10000 m;
514 (c) shadow fading standard deviation at 28 to 55 GHz of 500 m to 10000 m.

515 4.4 Rain Rate

516 The passage of radio waves through the rain area will produce attenuation, and when the
517 frequency of the radio wave is higher, the wavelength will be shorter. High-band microwaves
518 such as Ku and Ka with frequencies above 10 GHz have a wavelength of only 10 to 30
519 millimeters, similar to raindrops with a diameter of several millimeters, and the polarization
520 angle of electromagnetic waves is also changed by rainfall.
521 Assuming that at the 28 GHz frequency, other simulation parameters are the same as
522 Section 4.1 and 4.2, and the rain rate is between 0 and 20 mm/hr. According to Figure 16, at
523 8239.9 m, the omnidirectional receiving power (dBm) decreases linearly, the omnidirectional
524 path loss (dB) increases linearly, and the directional received power and path loss also have the
525 same change characteristics.
 Radio Science

526

527 Figure 16. Omnidirectional/Directional received power and path loss at 0 to 20 mm/hr.
528 According to Figure 17(a), rain rate has a greater impact on the PLE of high-altitude to
529 ground channel, so the impact of current rain rate cannot be ignored in the battlefield scenario.
530 By fitting at a distance of 500 to 10,000 meters, considering the path loss exponent and the
531 shadow fading standard deviation, the curve is derived as shown in Figure 17(b) and 17(c).
532 Therefore, in the rainfall scenario, it is necessary to appropriately reduce the signal frequency or
533 increase the number of system antennas to ensure that the path loss exponent and shadow fading
534 standard deviation are not too high, the channel capacity should be increased.

535

536 Figure 17. PLE and shadow fading standard deviation varies with rain rate. (a) omnidirectional
537 and directional PLE at 0 to 20 mm/hr of 8239.9 m; (b) PLE at 0 to 20 mm/hr of 500 m to 10000
538 m; (c) shadow fading standard deviation at 0 to 20 mm/hr of 500 m to 10000 m.

539 4.5 MIMO Antennas

540 According to Section 2.2, the condition number of the MIMO channel and the rank of the
541 channel transmission matrix can be investigated to judge its spatial multiplexing application.
542 Taking a 4×4 MIMO channel with both receiver and transmitter antenna numbers of URA 2×2 as
543 an example, an empirical CDF plot of the channel condition number and its 95% confidence
544 bound can be derived, as shown in Figure 18(a). Similarly, the MIMO channels of 9×9 and
545 16×16 are modeled, and the condition number empirical CDFs are derived as shown in Figure
546 18(b). The MIMO channels of 2×2, 3×3, 4×4, and 5×5 are modeled, and the empirical CDFs
 Radio Science

547 with the condition number are derived as shown in Figure 18(c). The rank distributions of each
548 channel transmission matrix under each main simulation condition are shown in Table 2.

549

550 Figure 18. The Empirical CDF of Condition Number, (a) 4×4 MIMO; (b) 4×4, 9×9, 16×16
551 MIMO; (c) 2×2, 3×3, 4×4, 5×5 MIMO.

552 Table 2. The Rank Distribution of Channel Transmission Matrix

MIMO Rank=1 Rank=2 Rank=3 Rank=4 Rank=5 Rank=6 Rank=7 Rank=8 Rank=9

2×2 1805 78245

3×3 1612 19348 59090
4×4 1602 7813 37286 33349
5×5 1601 4118 29789 33105 11437
9×9 1601 30 9768 21291 16556 20386 7529 1311 1578

553 As can be seen from the Figure 18 and the Table 2, for MIMO channels below 9×9, each
554 antenna is added to the transmitter and receiver, and the condition number is increased by about
555 12dB. In the 2×2 MIMO channel, 97.7% of the channel transmission matrix is in the full-rank
556 condition, which can realize the spatial multiplexing of two channels. In the 3×3 MIMO channel,
557 73.8% of the channel transmission matrix is in the full-rank condition, which can realize the
558 spatial multiplexing of three channels. In contrast, the proportion of full rank in MIMO channels
559 decreases as the number of antennas increases, and there is not even a full rank situation in
560 MIMO channels of 16x16 (the maximum rank is 15). This provides a basis for selecting the
561 number of antennas in the case of multiplexing.

562 4.6 HPBW Simulation

563 Suppose there is a signal transmitter at 50 dBm power, azimuth HPBW 10°, elevation
564 HPBW 10°, and other relevant parameter settings are the same as Section 4.1 and 4.2.
 Radio Science

565 Figure 19 shows the effects of the azimuth of the receiving antenna on directional
566 received power and path loss, receiver antenna gain, path loss exponent and shadow fading
567 standard deviation, respectively. Figure 20 shows the effects of the elevation of the receiving
568 antenna on directional received power and path loss, receiver antenna gain, path loss exponent
569 and shadow fading standard deviation, respectively. This provides a basis for the HPBW design
570 of directional antennas in the air-to-ground channel of high-altitude military UAVs.

571
572 Figure 19. The effects of the azimuth of the receiving antenna, (a) directional received power
573 and path loss; (b) receiver antenna gain; (c) path loss exponent; (d) shadow fading standard
574 deviation.
 Radio Science

575

576 Figure 20. The effects of the elevation of the receiving antenna, (a) directional received power
577 and path loss; (b) receiver antenna gain; (c) path loss exponent; (d) shadow fading standard
578 deviation.

579 5 Conclusions
580 In this paper, the spatial statistical channel model of high-altitude military UAV air-to-
581 ground scenario is constructed. By using NYUSIM, a multi-dimensional model of the high-
582 altitude military UAV air-to-ground channel without base station was built, and the channel was
583 simulated to get useful statistics. The dependence of the channel on T-R separation distance,
584 frequency, rain rate and antenna HPBW parameters have been analyzed, and large-scale fading
585 and small-scale fading statistical analyses were carried out to get channel characteristics. In
586 addition, by calculating and analyzing the condition number and rank of the channel transmission
587 matrix, the spatial multiplexing of the high-altitude military UAV air-to-ground channel without
588 base station scenario under different MIMO antenna arrays have been analyzed.
589 According to the channel analysis results, under this model, the linearity of the millimeter
590 band below 49 GHz is better, which is suitable for this battlefield situation awareness scenario.
591 The use of 2×2 and 3×3 MIMO antenna arrays can realize spatial multiplexing effectively.
592 Moreover, this communication channel should also take the rain rate and distance conditions into
 Radio Science

593 account, adjust the communication frequency and MIMO antenna parameters flexibly. The result
594 can provide a theoretical basis for future high-altitude military UAV frequency selection and
595 antenna design.

596

597 References
598 Al-Shammari, B. K. J., Hburi, I., Idan, H. R., & Khazaal, H. F. (2021). An Overview of
599 mmWave Communications for 5G. 2021 International Conference on Communication &
600 Information Technology (ICICT) (pp. 133-139).
601 https://doi.org/10.1109/ICICT52195.2021.9568459
602 Bai, L., Huang, Z., Zhang, X., & Cheng, X. (2021). A Non-Stationary 3D Model for 6G Massive
603 MIMO mmWave UAV Channels. IEEE Transactions on Wireless Communications, 21(6), 4325-
604 4339. https://doi.org/10.1109/TWC.2021.3128970
605 Bajracharya, R., Shrestha, R., Kim, S., & Jung, H. (2022). 6G NR-U Based Wireless
606 Infrastructure UAV: Standardization, Opportunities, Challenges and Future Scopes. IEEE Access,
607 10, 30536-30555. https://doi.org/10.1109/ACCESS.2022.3159698
608 Bian, J., Wang, C. -X., Gao, X., You, X., & Zhang, M. (2021). A General 3D Non-Stationary
609 Wireless Channel Model for 5G and Beyond. IEEE Transactions on Wireless Communications,
610 20(5), 3211-3224. https://doi.org/10.1109/TWC.2020.3047973
611 Bithas, P. S., Nikolaidis, V., Kanatas, A. G., & Karagiannidis, G. K. (2020). UAV-to-Ground
612 Communications: Channel Modeling and UAV Selection. IEEE Transactions on
613 Communications, 68(8), 5135-5144. https://doi.org/10.1109/TCOMM.2020.2992040
614 Chang, H., Wang, C. -X., Liu, Y., Huang, J., Sun, J., Zhang, W., & Gao, X. (2021). A Novel
615 Nonstationary 6G UAV-to-Ground Wireless Channel Model With 3-D Arbitrary Trajectory
616 Changes. IEEE Internet of Things Journal, 8(12), 9865-9877.
617 https://doi.org/10.1109/JIOT.2020.3018479
618 Chetlur, V. V., & Dhillon, H. S. (2017). Downlink Coverage Analysis for a Finite 3-D Wireless
619 Network of Unmanned Aerial Vehicles. IEEE Transactions on Communications, 65(10), 4543-
620 4558. https://doi.org/10.1109/TCOMM.2017.2722500
621 Gao, Q., Wan, Z. -N., Zhang, M. -X., & Ma, W. -Q. (2021). Design and implementation of
622 panorama vehicle-mounted battlefield perception system based on multi-light fusion imaging
623 technology. 2021 2nd International Conference on Computer Communication and Network
624 Security (CCNS) (pp. 80-83). https://doi.org/10.1109/CCNS53852.2021.00024
625 Geraci, G., Garcia-Rodriguez, A., Azari, M. M., Lozano, A., Mezzavilla, M., Chatzinotas, S., et
626 al. (2022). What Will the Future of UAV Cellular Communications Be? A Flight From 5G to 6G.
627 IEEE Communications Surveys & Tutorials, 24(3), 1304-1335.
628 https://doi.org/10.1109/COMST.2022.3171135
629 Huang, J., Wang, C. -X., Chang, H., Sun, J., & Gao, X. (2020). Multi-Frequency Multi-Scenario
630 Millimeter Wave MIMO Channel Measurements and Modeling for B5G Wireless
631 Communication Systems. IEEE Journal on Selected Areas in Communications, 38(9), 2020-
632 2025. https://doi.org/10.1109/JSAC.2020.3000839
 Radio Science

633 Jawhly, T., & Chandra Tiwari, R. (2021). Loss exponent modeling for the hilly forested region in
634 the VHF band III. Radio Science, 56, e2020RS007201. https://doi.org/10.1029/2020RS007201
635 Ji, Z., Zhang, Y., He, Z., Yeoh, P.L., Li, B., Yin, H., et al. (2022). Wireless Secret Key
636 Generation for Distributed Antenna Systems: A Joint Space-Time-Frequency Perspective. IEEE
637 Internet of Things Journal, 9(1), 633-647. https://doi.org/10.1109/JIOT.2021.3084361
638 Jia, Y., Tu, X., & Yan, W. (2020). An UAV wireless communication noise suppression method
639 based on OFDM modulation and demodulation. Radio Science, 55, e2019RS006959. https://doi.
640 org/10.1029/2019RS006959
641 Kanhere, O., Ju, S., Xing, Y., & Rappaport, T.S. (2019). Map-Assisted Millimeter Wave
642 Localization for Accurate Position Location. 2019 IEEE Global Communications Conference
643 (GLOBECOM) (pp. 1-6). https://doi.org/10.1109/GLOBECOM38437.2019.9013365
644 Khawaja, W., Guvenc, I., Matolak, D. W., Fiebig, U. -C., & Schneckenburger, N. (2019). A
645 Survey of Air-to-Ground Propagation Channel Modeling for Unmanned Aerial Vehicles. IEEE
646 Communications Surveys & Tutorials, 21(3), 2361-2391.
647 https://doi.org/10.1109/COMST.2019.2915069
648 Khuwaja, A. A., Chen, Y., Zhao, N., Alouini, M. -S., & Dobbins, P. (2018). A Survey of
649 Channel Modeling for UAV Communications. IEEE Communications Surveys & Tutorials,
650 20(4), 2804-2821. https://doi.org/10.1109/COMST.2018.2856587
651 Li, X., Feng, W., Chen, Y., Wang, C. -X., & Ge, N. (2020). Maritime Coverage Enhancement
652 Using UAVs Coordinated With Hybrid Satellite-Terrestrial Networks. IEEE Transactions on
653 Communications, 68(4), 2355-2369. https://doi.org/10.1109/TCOMM.2020.2966715
654 Liao, C., Xu, K., Xie, W., & Xia, X. (2020). 3‐D massive MIMO channel model for high‐speed
655 railway wireless communication. Radio Science, 55, e2020RS007070. https://doi.org/
656 10.1029/2020RS007070
657 Liu, R., Liu, A., Qu, Z., & Xiong, N. N. (2021). An UAV-Enabled Intelligent Connected
658 Transportation System With 6G Communications for Internet of Vehicles. IEEE Transactions on
659 Intelligent Transportation Systems (pp. 1-15). https://doi.org/10.1109/TITS.2021.3122567
660 Liu, Y., Wang, C. -X., Chang, H., He, Y., & Bian, J. (2021). A Novel Non-Stationary 6G UAV
661 Channel Model for Maritime Communications. IEEE Journal on Selected Areas in
662 Communications, 39(10), 2992-3005. https://doi.org/10.1109/JSAC.2021.3088664
663 Liu, Y., Wang, C. -X., Lopez, C. F., Goussetis, G., Yang, Y., & Karagiannidis, G. K. (2019). 3D
664 Non-Stationary Wideband Tunnel Channel Models for 5G High-Speed Train Wireless
665 Communications. IEEE Transactions on Intelligent Transportation Systems, 21(1), 259-272.
666 https://doi.org/10.1109/TITS.2019.2890992
667 Mou, M. A., Mowla, M. M., & Aftabi Momo, S. H. (2019). Statistical Channel Model at
668 mmWave Band Inside High Speed Water Vehicle. 2019 3rd International Conference on
669 Electrical, Computer & Telecommunication Engineering (ICECTE) (pp. 109-112).
670 https://doi.org/10.1109/ICECTE48615.2019.9303530
671 NYUSIM. Accessed: Jun. 10, 2019. [Online]. Available:
672 https://wireless.engineering.nyu.edu/nyusim/
673 Rappaport, T. S. (1996). Wireless communications: Principles and practice. Prentice Hall PTR
 Radio Science

674 Rappaport, T. S., MacCartney, G. R., Samimi, M. K., & Sun, S. (2015). Wideband millimeter-
675 wave propagation measurements and channel models for future wireless communication system
676 design. IEEE Transactions on Communications, 63(9), 3029-3056.
677 https://doi.org/10.1109/TCOMM.2015.2434384
678 Rappaport, T. S., Sun, S., Mayzus, R., Zhao, H., Azar, Y., Wang, K., et al.(2013). Millimeter
679 wave mobile communications for 5G cellular: It will work!. IEEE Access, 1, 335-349.
680 https://doi.org/10.1109/ACCESS.2013.2260813
681 Shen, D., Jiang, B., Liu, T., Guo, J., & Qi, S. (2017). Realizing of a battlefield electromagnetic
682 situation system. 2017 36th Chinese Control Conference (CCC) (pp. 10304-10309).
683 https://doi.org/10.23919/ChiCC.2017.8028994
684 Sun, S., MacCartney, G. R., & Rappaport, T. S. (2017). A novel millimeter-wave channel
685 simulator and applications for 5G wireless communications. 2017 IEEE International
686 Conference on Communications (ICC) (pp. 1-7). https://doi.org/10.1109/ICC.2017.7996792
687 Sun, S., Rappaport, T.S., Heath, R.W., Nix, A., & Rangan, S. (2014). Mimo for millimeter-wave
688 wireless communications: beamforming, spatial multiplexing, or both?. IEEE Communications
689 Magazine, 52(12),110-121. https://doi.org/10.1109/MCOM.2014.6979962
690 Teixeira, E., Sousa, S., Velez, F. J., & Peha, J. M. (2021). Impact of the propagation model on
691 the capacity in small-cell networks: Comparison between the UHF/SHF and the millimeter
692 wavebands. Radio Science, 56, e2020RS007150. https://doi. org/10.1029/2020RS007150
693 Tian, J., & Shi, J. (2020). A new method for analyzing the attenuation characteristics of satellite
694 radar altimeter signals due to rainfall based on a multilayer medium model. Radio Science, 55,
695 e2019RS006962. https://doi.org/10.1029/2019RS006962
696 Tozer, T., Grace, D., Thompson, J., & Baynham, P. (2000). UAVs and HAPs-potential
697 convergence for military communications. IEE Colloquium on Military Satellite
698 Communications (pp. 10/1-10/6). https://doi.org/10.1049/ic:20000130
699 Traspadini, A., Giordani, M., & Zorzi, M. (2022). UAV/HAP-Assisted Vehicular Edge
700 Computing in 6G: Where and What to Offload?. 2022 Joint European Conference on Networks
701 and Communications & 6G Summit (EuCNC/6G Summit) (pp. 178-183).
702 https://doi.org/10.1109/EuCNC/6GSummit54941.2022.9815734
703 Wang, C. -X., Huang, J., Wang, H., Gao, X., You, X., & Hao, Y. (2020). 6G Wireless Channel
704 Measurements and Models: Trends and Challenges. IEEE Vehicular Technology Magazine,
705 15(4), 22-32. https://doi.org/10.1109/MVT.2020.3018436
706 Wang, J., Wang, C. -X., Huang, J., Wang, H., & Gao, X. (2021). A General 3D Space-Time-
707 Frequency Non-Stationary THz Channel Model for 6G Ultra-Massive MIMO Wireless
708 Communication Systems. IEEE Journal on Selected Areas in Communications, 39(6), 1576-
709 1589. https://doi.org/10.1109/JSAC.2021.3071850
710 Xiao, Z., Zhu, L., Liu, Y., Yi, P., Zhang, R., Xia, X., & Schober, R. (2022). A Survey on
711 Millimeter-Wave Beamforming Enabled UAV Communications and Networking. IEEE
712 Communications Surveys & Tutorials, 24(1), 557-610.
713 https://doi.org/10.1109/COMST.2021.3124512
 Radio Science

714 Yang, H., Yang, J., Zhang, B., & Wang, C. (2021). Visualization Analysis of Research on
715 Unmanned-Platform Based Battlefield Situation Awareness. 2021 IEEE International
716 Conference on Artificial Intelligence and Computer Applications (ICAICA) (pp. 334-338).
717 https://doi.org/10.1109/ICAICA52286.2021.9497890
718 Yang, J., Li, H., & Xu, Z. (2021). Analysis of channel characteristics between satellite and space
719 station in terahertz band based on ray tracing. Radio Science, 56, e2021RS007290.
720 https://doi.org/10.1029/2021RS007290
721 Yang, X., Song, C., Xu, C., Hao, M., Tan, J., Lou, X., et al. (2022). A survey of the estimation
722 and fusion methods for battlefield situation awareness. Seventh Asia Pacific Conference on
723 Optics Manufacture and 2021 International Forum of Young Scientists on Advanced Optical
724 Manufacturing (APCOM and YSAOM 2021), 1216633. https://doi.org/10.1117/12.2616097
725 Yang, J.,& Yuan, L. (2022). Analysis of channel characteristics for inter-satellite terahertz
726 communication based on Lambertian model. Journal of Signal Processing, 38(2), 232-240.
727 https://doi.org/10.16798/j.issn.1003-0530.2022.02.002
728 You, X., Wang, C. -X., Huang, J., Gao, X., Zhang, Z., Wang, M., et al. (2020). Towards 6G
729 wireless communication networks:Vision,enabling technologies, and new pradigm shifts. Sci
730 China Inf Sci, 64, 110301. https://doi.org/10.1007/s11432-020-2955-6
731 Zhang, X., Wang, J., & Poor, H. V. (2022). Statistical QoS-Driven Beamforming and Trajectory
732 Optimizations in UAV/IRS-Based 6G Wireless Networks in the Non-Asymptotic Regime. 2022
733 IEEE International Symposium on Information Theory (ISIT) (pp. 3333-3338).
734 https://doi.org/10.1109/ISIT50566.2022.9834715
735 Zhang, X., Zhu, Q., & Poor, H. V. (2022a). Massive-MIMO Channel Capacity Modeling for
736 mURLLC Over 6G UAV Mobile Wireless Networks. 2022 56th Annual Conference on
737 Information Sciences and Systems (CISS) (pp. 49-54).
738 https://doi.org/10.1109/CISS53076.2022.9751152
739 Zhang, X., Zhu, Q., & Poor, H. V. (2022b). Multiple-Access Based UAV Communications and
740 Trajectory Tracking Over 6G Mobile Wireless Networks. 2022 IEEE Wireless Communications
741 and Networking Conference (WCNC) (pp. 2429-2434).
742 https://doi.org/10.1109/WCNC51071.2022.9771943
743 Zhang, Y., Zhao, L., & He, Z. (2021). A 3-D hybrid dynamic channel model for indoor THz
744 communications. China communications, 18(5), 50-65.
745 https://doi.org/10.23919/JCC.2021.05.004
746 Zhu, X. (2020). Analysis of military application of UAV swarm technology. 2020 3rd
747 International Conference on Unmanned Systems (ICUS) (pp. 1200-1204).
748 https://doi.org/10.1109/ICUS50048.2020.9274974
749
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Figure 9.
Figure 10.
Figure 11.
Figure 12.
Figure 13.
Figure 14.
Figure 15.
Figure 16.
Figure 17.
Figure 18.
Figure 19.
Figure 20.