Coexistence of UAVs and Terrestrial Users in Millimeter-Wave Urban Networks

Full text

Coexistence of UAVs and Terrestrial Users in
Millimeter-Wave Urban Networks
Seongjoon Kang†Marco Mezzavilla†Angel Lozano[Giovanni Geraci[
Sundeep Rangan†Vasilii Semkin]William Xia†Giuseppe Loianno†
†NYU Tandon School of Engineering, Brooklyn, NY, USA
[Univ. Pompeu Fabra, Barcelona, Spain
]VTT Technical Research Centre of Finland Ltd, Finland
Abstract—5G millimeter-wave (mmWave) cellular networks
are in the early phase of commercial deployments and present
a unique opportunity for robust, high-data-rate communication
to unmanned aerial vehicles (UAVs). A fundamental question
is whether and how mmWave networks designed for terrestrial
users should be modified to serve UAVs. The paper invokes
realistic cell layouts, antenna patterns, and channel models
trained from extensive ray tracing data to assess the performance
of various network alternatives. Importantly, the study considers
the addition of dedicated uptilted rooftop-mounted cells for aerial
coverage, as well as novel spectrum sharing modes between
terrestrial and aerial network operators. The effect of power
control and of multiuser multiple-input multiple-output are also
studied.
I. INTRODUCTION
The interest is growing rapidly in unmanned aerial vehi-
cles (UAVs) for applications such as monitoring, surveying,
precision agriculture, construction, remote sensing, or product
delivery [1]–[5]. Low-latency connectivity at high data rates
is required for many of these applications and the millimeter
wave (mmWave) frequency range offers bandwidths that can
be instrumental to meet these requirements [6]–[10]. Also,
links to UAVs are often line-of-sight (LOS), which is desirable
due to the limited diffraction at these frequencies [11], [12].
There has been extensive work in UAV wireless commu-
nication and its interaction with cellular networks for fre-
quencies below 6 GHz [13]–[18]. Likewise, mmWave com-
munication for terrestrial cellular networks has seen enormous
progress [19]–[21], particularly with the development of the
3rd Generation Partnership Project (3GPP) 5G standard [22].
However, the coordination in the use of spectrum between
aerial and terrestrial networks in the mmWave realm faces
unique challenges. Chiefly, due to the directional nature of
mmWave transmissions [23], the interference between aerial
and terrestrial links is intricate, particularly in dense urban
scenarios with substantial ground and building reflections.
S. Rangan, W. Xia, S. Kang, and M. Mezzavilla were supported by NSF
grants 1302336, 1564142, 1547332, and 1824434, SRC, and the industrial
affiliates of NYU WIRELESS. A. Lozano and G. Geraci were supported by
ERC grant 694974, by MINECO’s Projects RTI2018-101040 and PID2021-
123999OB-I00, by the “Ram´
on y Cajal” program, and by ICREA. The work
of V. Semkin was supported in part by the Academy of Finland.
In this paper, we investigate the coexistence of UAVs and
terrestrial users (UEs) at mmWave frequencies under different
spectrum sharing paradigms, thus bringing to the next level
previous work that only focused on coverage [24]. To this
end, we conduct extensive system-level simulations with 3GPP
channel models and antenna patterns for terrestrial users [25],
[26], and with data-driven aerial channel models along with
realistic antenna patterns for UAVs [27].
We begin by considering a standard mmWave deployment,
with a single operator providing connectivity to both UAVs and
UEs, and investigate the UAV-UE interplay under single- (SU)
and multiuser (MU) multiple-input multiple output (MIMO)
operation. This first study shows that:
•With SU-MIMO, UAV-to-UE interference is rather neg-
ligible because (i) urban LOS probability increases only
marginally with the UAV height, (ii) the arrays of stan-
dard base stations (BSs) are downtilted, and (iii) highly
directional mmWave transmissions provide considerable
cell isolation.
•With MU-MIMO, the data rates of both UAVs and UEs
improve substantially as the interference remains low
while multiple users are spatially multiplexed.
We further study a setup with two mobile network operators
(MNOs) sharing the same spectrum: a terrestrial MNO running
a standard mmWave network and an aerial MNO operating
dedicated rooftop-mounted cells exclusively for UAVs. For
this multi-MNO setup, we consider both closed access and
open access paradigms, respectively confining UAVs to the
aerial MNO and allowing UAVs to connect to whichever MNO
offers the best service. This second study reveals that:
•Ensuring UAV coverage under closed access requires high
densities of dedicated cells, yet this does not significantly
penalize UEs; the increased number of concurrent UAV
interferers is balanced by the fact that UAVs can afford
transmitting at lower power.
•Under open access, dedicated cells are not crucial for
coverage, but their addition improves the performance
of both UEs and UAVs. Indeed, UEs can now access
resources previously taken up by UAVs, and UAVs enjoy
shorter links and better antenna gains.
arXiv:2209.08964v2 [eess.SY] 20 Sep 2022

Fig. 1: Coexisting UAV and terrestrial UEs in a mmWave network.
TABLE I: Simulation parameters.
Parameter Value
Minimum UE-BS 2D distance (m)10
Minimum UAV-BS 3D distance (m)10
ISDs(m)200
Height of standard BSs (m)10
Height of dedicated BSs (m) [10,30]
Bandwidth (MHz)400
Frequency (GHz)28
BS noise figure (dB)6
UAV height (m)120
UAV and UE maximum transmit power (dBm)23
3GPP vertical half-power beamwidth (θ3dB)65◦
3GPP horizontal half-power beamwidth (φ3dB)65◦
Power control parameters, P0(dBm) and α-82,0.8
Number of associated users per cell 10
II. SY ST EM MO DE L
This work focuses on the uplink—the more data-hungry
direction for UAVs—with the terrestrial UEs, UAVs, and BSs,
all deployed in a 1km2×1km2area with wrap-around.
A. Deployment and Propagation Channel Features
Network Deployment: BSs are deployed following a homo-
geneous Poisson point process with a given average intersite
distance (ISD), and they have three sectors, each correspond-
ing to a cell and equipped with an 8×8uniform rectangular
array (URA). Two types of BSs are considered: a standard
deployment with antennas downtilted by −12◦, and a dedi-
cated deployment with rooftop-mounted antennas uptilted by
45◦, as illustrated in Fig. 1. In turn, UAVs and terrestrial users
feature 4×4URAs. Those mounted on UAVs are downtilted by
−90◦while those at terrestrial UEs are distributed uniformly
in azimuth with a fixed elevation of 0◦. All the parameters
are summarized in Table I, following the simulation settings
in [25]. The height of dedicated BSs, which is an additional
parameter, is uniformly random between 10 and 30 m.
UAV Radiation Pattern: In order to incorporate the effects
of the UAV frame on its radiation pattern, we simulated a
quarter-wave monopole antenna (with bottom hemisphere cov-
erage) placed under the UAV and used a software package [28]
that utilizes geometrical optics, uniform theory of diffraction,
and physical optics. Fig. 2 illustrates the effect of the UAV on
Fig. 2: Antenna radiation pattern accounting for the UAV frame.
the antenna radiation pattern. The fluctuations of the antenna
gain are due to protruding parts of the UAV (e.g., landing skid,
gimbal) that are within the field of view of the antenna [29].
Non-obviously, the effects of the UAV frame increase with the
antenna directivity. We apply the obtained element radiation
pattern to each UAV, whereas 3GPP antenna element patterns
are adopted for BSs and UEs [26] .
Propagation Model: The 3GPP channel model reported
in [25] is invoked to characterize the propagation between
UEs and BSs (for urban microcells). To represent the mmWave
propagation between UAVs and BSs, for which no calibrated
model is available, the data-driven channel model developed
in [27], [30] is invoked.
B. Cell Selection and Beamforming
Each transmitter, UE or UAV, connects to the strongest BS.
Both UAVs and UEs employ the open-loop power control
policy specified in [31]. UAVs and UEs apply long-term
transmit beamforming, aligning their beamforming vectors
with the maximum-eigenvalue eigenvector of their channel
covariance matrices [32]. After TX beamforming, we obtain a
SIMO channel h`k and signal to noise ration (SNR) SNR`k
from user kand BS `. With MU-MIMO, the minimum mean-
square error (MMSE) receive vector for user uat BS `is given
by [33]
w`u = Nu−1
X
k=0
SNR`k h`kh∗
`k
+ 1 + X
n6=`
Nn−1
X
k=0
SNRnk!I!−1
h`u,(1)
where the summation is only active users scheduled in a
particular time slot. The out-of-cell interference is represented
with the sum over n6=`where Nnis the number of active
users in cell n. As the channel matrices for other-cell users
are not estimated by BS `, their interference is regarded as
spatially white. Accordingly, the overall gain from user kto
BS `is
G`k =|w∗
`kh`k |2(2)

The interference experienced by user uis then
I`u =X
k6=u
Ptx
`k G`k X
m
Atx
`kmArx
`kmL−1
`km
+X
n6=`X
k
Ptx
nkGnk X
m
Atx
nkmArx
nkmL−1
nkm,(3)
where Ptx
`k is the transmit power of user kin cell `,Atx
`km and
Arx
`km are the transmit and receive antenna element gains on
path m, and L`km is the pathloss along path mbetween user
kand BS `. Similarly, the received power from user uis
Prx
`u =Ptx
`u G`u X
m
Atx
`umArx
`umL−1
`um.(4)
Altogether, the SINR between UE uand BS `equals
SINR`u =Prx
`u
I`u +N0Bkw`uk2,(5)
where N0is the noise power spectral density and Bthe
bandwidth.
III. UAV-UE SPE CTRUM SHARING WITH
STAND ARD MMWAVE CE LL S
We begin by considering a standard mmWave deployment.
This is handled by a single terrestrial MNO that provides
connectivity to both UAVs and terrestrial UEs. For this setup,
we investigate the coexistence and interplay between the two
populations of users under SU-MIMO and MU-MIMO. At this
stage, no dedicated BSs are assumed.
A. SU-MIMO mmWave Communication
To evaluate the amount of interference introduced by UAVs
depending on their penetration rate and power control, we
examine the three configurations reported in Table II, namely
with (1) no UAVs in the network, (2) a fraction of 50% of
UAVs employing fractional power control, and (3) a fraction of
50% UAVs transmitting at maximum power. Since the network
operates in SU-MIMO mode, each BS schedules a single user
at random on a given time-frequency physical resource block.
In configurations (2) and (3), half of the scheduled users on
average are UAVs.
TABLE II: UAV penetration rates and power control considered.
Configuration UAV fraction UAV power control
1 0% None
2 50% Open loop
3 50% Max (23 dBm)
Fig. 3 shows the distribution of SINR and interference-to-
noise ratio (INR) for UEs. Under SU-MIMO, the interference
observed is only of intercell origin. The low INR values
suggest that the network is predominantly noise-limited and
that, as a consequence, UAV power control improves the SINR
of UEs by only 1-2 dB relative to full-power transmission
(config 2 vs. config 3). The reasons why the interference
generated by UAVs onto UEs is rather negligible at mmWave
frequencies are as follows:
Fig. 3: SINR and INR distributions of UEs for the different UAV
penetration rates and power control configurations defined in Table II.
•LOS probability in urban scenarios increases only
marginally with the UAV altitude. Fig. 4 shows the LOS
probability of the aerial channel model developed in [27].
Since UAVs are surrounded by buildings, the BS is highly
visible by UAVs flying at 120 m only within a horizontal
distance of 200 m. Hence, the chance to cause severe
interference to UEs is low.
•The arrays of standard BSs are downtilted by −12◦,
mitigating the interference from UAVs due to the reduced
antenna element gains.
•Lastly, the increased pathloss and highly directional trans-
missions at mmWave frequencies result in a considerable
attenuation of the interference.
B. MU-MIMO mmWave Communication
There is a limited amount of prior work on exploiting MU-
MIMO in mmWave networks [34]–[36]. In particular, there is
no established research on spectrum sharing between UEs and
UAVs with MU-MIMO. In this section, we tackle this issue
by presenting the distribution of achievable data rates with
Fig. 4: LOS probability from a standard BS positioned at (0,0).

an increasing number of simultaneous scheduled users. We
consider Ncassociated users per cell, connected through the
control channel but not necessarily scheduled in a particular
time slot. In each time slot, the BS schedules Nuusers and
hence each user is scheduled in a fraction Nu/Ncof the time
slots. We simulated a scenario with 5UAVs and 5UEs per
BS, so the average value of Nc= 10. When scheduled, all
users obtain access to the full bandwidth Bresulting in an
average rate
R=Nu
Nc
Blog2(1 + SINR).(6)
Fig. 5 shows the data rate and SINR distributions of UAVs
and terrestrial UEs with increasing active users. Due to MU-
MIMO, the achievable data rate of both UAVs and UEs
improves substantially. Although the number of scheduled
users increases, the interference is still minimal. As shown in
Fig. 5, the SINR coverage for UEs and UAVs is still ensured
with a higher number of scheduled users (Nu= 4).
IV. UAV-UE SPECTR UM SHARING WITH
DED IC ATED AERIAL MMWAVE CELLS
In this section, we consider a hypothetical setup with MNOs
operating in the same frequency band, namely:
•A terrestrial operator, MNOT, running a standard
mmWave network as described in Section III.
•An aerial operator, MNOA, running a dedicated mmWave
network consisting of rooftop-mounted, uptilted BSs that
are reserved exclusively for UAV communication.
As the penetration of UAVs increases, terrestrial MNOT
may choose to share—under some leasing agreement—their
spectrum with another MNOAthat only intends to provide
aerial connectivity, giving rise to the above multi-MNO sce-
nario [18]. For this scenario, we examine the performance of
both UAVs and UEs under two spectrum sharing paradigms:
•Closed access, where UAVs are only allowed to con-
nect to dedicated BSs. On one hand, this paradigm
requires low-to-no coordination for radio resource alloca-
tion, since all scheduling, beamforming, and networking
decisions are performed individually by each MNO. On
the other hand, such simplification comes at the cost of a
suboptimal spectrum usage, owing to the restricted UAV
association and lack of synchronization.
•Open access, with UAVs allowed to connect to whichever
BS offers the best quality of service, standard or ded-
icated. This enables a more efficient use of spectrum
and network infrastructure. However, a high coordination
between the MNOs is required to jointly manage radio
resources for aerial and terrestrial users, possibly entail-
ing MNOTand MNOAto belong to the same network
provider.
Fig. 6 further summarizes the main features of the two
multi-MNO spectrum sharing paradigms as compared to a
single-MNO scenario. In the remainder of this section, we
present the performance results obtained under the closed
(a) UAVs
(b) UEs
Fig. 5: Performance under MU-MIMO with standard BSs, with
Nu= 1 corresponding to SU-MIMO.
access and open access paradigms. We consider different
dedicated BS densities by varying ISDd, and consider MU-
MIMO with Nu= 2.
Fig. 7 shows the SINRs and data rates attained by UAVs and
UEs under closed access, for various values of the dedicated
BSs ISD. Since UAVs can only connect to dedicated BSs, the
density of the latter greatly affects UAV coverage, with an ISD
of 200 m or less required to keep the UAV outage at bay. As
the density of dedicated BSs decreases, so does the number
of concurrent UAV interferers seen by each standard BS.
However, sparser dedicated deployments occasionally force
UAVs to connect to far-flung dedicated BSs, thereby increasing
their transmission power and generating stronger interference
to standard BSs, an instance of the near-far problem. Overall,
these two phenomena largely cancel one another, and the data
rate experienced by UEs only marginally shrinks with the
dedicated BS density.
Fig. 8 shows the SINRs and data rates achieved by UAVs
and UEs under open access, with UAVs allowed to connect

Single MNO Multi MNO — Closed Access Multi MNO — Open Access
Spectrum reuse
and efficiency
Low: No dedicated deployment; no
spectrum reuse between aerial and ter-
restrial networks.
Medium: Suboptimal spectrum and
infrastructure reuse due to restricted
association and lack of synchroniza-
tion.
High: Optimal spectrum and infras-
tructure reuse; seamless interference
and user association coordination.
Coordination for
resource allocation
None: Single operator running au-
tonomously, no need for coordination.
Low: Requires monitoring inter-
operator interference but scheduling,
beamforming, and networking
decisions are made individually.
High: Requires operators to manage
aerial and terrestrial users jointly.
Fig. 6: Spectrum sharing paradigms: (left) single MNO, (center) two MNOs in closed access, and (right) two MNOs in open access.
(a) UAVs
(b) UEs
Fig. 7: SINR and data rates for (a) UAVs and (b) terrestrial UEs
under a Closed Access multi-operator paradigm. Both BS tiers
feature MU-MIMO with Nu= 2.
(a) UAVs
(b) UEs
Fig. 8: SINR and data rates for (a) UAVs and (b) terrestrial UEs
under an Open Access multi-operator paradigm. Both BS tiers
feature MU-MIMO with Nu= 2.
to the BS providing the highest signal strength, thereby
circumventing the near-far effect. The figure considers three
values for the dedicated BSs ISD, as well as the baseline
case without dedicated deployment, labeled as ISDd=∞.
Despite introducing additional interference on the UEs—see
SINR curves in Fig. 8(b)—, an increase in the number of
dedicated BSs ultimately improves the data rates for both
UEs and UAVs. Indeed, UEs can now access resources that
were previously taken up by UAVs, as many of the latter are
offloaded to dedicated BSs. In turn, UAVs achieve higher data
rates because, when connecting to dedicated BSs, the link
distance is reduced, uptilted arrays provide higher element

gains, and resources are not shared with UEs.
V. CONCLUSION
The goal of this paper was to understand the effect of
mmWave spectrum sharing among UAVs and terrestrial users.
To this end, an extensive campaign of simulations was
launched, incorporating an accurate directional mmWave aerial
channel model and UAV antenna patterns alongside 3GPP-
based terrestrial mmWave channels and BS antenna patterns.
We considered multiple scenarios: (i) one with a single
terrestrial operator running a standard mmWave network with
SU-MIMO and MU-MIMO, (ii) another where a second aerial
operator reuses the same spectrum independently to support
UAVs via dedicated uptilted cells, (iii) and one last scenario
where the two operators jointly offer connectivity to UAVs.
Multiple insights have been drawn:
•The interference generated by UAVs onto terrestrial UEs
in standard mmWave networks is minimal, thanks to
downtilted cells and high directionality.
•Owing to the above intrinsic spatial separation, MU-
MIMO is highly effective and data rates improve despite
the additional concurrent interferers.
•Co-channel uptilted BSs run independently by an aerial
operator in closed access mode can provide satisfactory
UAV coverage, provided they are densely deployed.
•In open access—where UAVs can connect to either stan-
dard or dedicated uptilted cells—such uptilted cells are
no longer crucial for UAV coverage, but their addition
boosts the data rates of UAVs and UEs alike.
REFERENCES
[1] G. Geraci, A. Garcia-Rodriguez, M. M. Azari, A. Lozano, M. Mezzav-
illa, S. Chatzinotas, Y. Chen, S. Rangan, and M. Di Renzo, “What will
the future of UAV cellular communications be? A flight from 5G to
6G,” IEEE Communications Surveys & Tutorials, pp. 1–1, 2022.
[2] Y. Zeng, I. Guvenc, R. Zhang, G. Geraci, and D. W. Matolak (Eds.),
UAV Communications for 5G and Beyond. Wiley – IEEE Press, 2020.
[3] W. Saad, M. Bennis, M. Mozaffari, and X. Lin, Wireless Communi-
cations and Networking for Unmanned Aerial Vehicles. Cambridge
University Press, 2020.
[4] Y. Zeng, Q. Wu, and R. Zhang, “Accessing from the sky: A tutorial on
UAV communications for 5G and beyond,” Proceedings of the IEEE,
vol. 107, no. 12, pp. 2327–2375, 2019.
[5] N. Cherif, W. Jaafar, H. Yanikomeroglu, and A. Yongacoglu, “3D aerial
highway: The key enabler of the retail industry transformation,” IEEE
Communications Magazine, vol. 59, no. 9, pp. 65–71, 2020.
[6] 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, 2016.
[7] C. Zhang, W. Zhang, W. Wang, L. Yang, and W. Zhang, “Research
challenges and opportunities of UAV millimeter-wave communications,”
IEEE Wireless Communications, vol. 26, no. 1, pp. 58–62, 2019.
[8] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Communications
and control for wireless drone-based antenna array,” IEEE Transactions
on Communications, vol. 67, no. 1, pp. 820–834, Jan 2019.
[9] J. Lee et al., “Spectrum for 5G: Global status, challenges, and enabling
technologies,” IEEE Communications Magazine, vol. 56, no. 3, pp. 12–
18, 2018.
[10] M. Shafi, A. F. Molisch, P. J. Smith, T. Haustein, P. Zhu, P. De 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,
2017.
[11] W. Khawaja, O. Ozdemir, and I. Guvenc, “UAV air-to-ground channel
characterization for mmWave systems,” in Proc. IEEE VTC-Fall, 2017.
[12] 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. ITG WSA, March 2018, pp. 1–6.
[13] G. Geraci, A. Garcia Rodriguez, L. Galati Giordano, D. L´
opez-P´
erez,
and E. Bj¨
ornson, “Understanding UAV cellular communications: From
existing networks to massive MIMO,” IEEE Access, vol. 6, pp. 67 853–
67 865, Nov. 2018.
[14] X. Lin, V. Yajnanarayana, S. D. Muruganathan, S. Gao, H. Asplund, H.-
L. Maattanen, M. Bergstrom, S. Euler, and Y.-P. E. Wang, “The sky is
not the limit: LTE for unmanned aerial vehicles,” IEEE Communications
Magazine, vol. 56, no. 4, pp. 204–210, 2018.
[15] M. M. U. Chowdhury, I. Guvenc, W. Saad, and A. Bhuyan, “Ensuring
reliable connectivity to cellular-connected UAVs with uptilted antennas
and interference coordination,” arXiv:2108.05090, 2021.
[16] A. Garcia-Rodriguez, G. Geraci, D. L´
opez-P´
erez, L. Galati Giordano,
M. Ding, and E. Bj¨
ornson, “The essential guide to realizing 5G-
connected UAVs with massive MIMO,” IEEE Communications Mag-
azine, vol. 57, no. 12, pp. 84–90, Dec. 2019.
[17] M. Benzaghta, G. Geraci, R. Nikbakht, and D. L´
opez-P´
erez, “UAV
communications in integrated terrestrial and non-terrestrial networks,”
in Proc. IEEE Globecom, 2022.
[18] G. Geraci, D. L´
opez-P´
erez, M. Benzaghta, and S. Chatzinotas, “Inte-
grating terrestrial and non-terrestrial networks: 3D opportunities and
challenges,” arXiv:2207.10385, 2022.
[19] 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, March 2014.
[20] T. S. Rappaport et al., “Millimeter wave mobile communications for 5G
cellular: It will work!” IEEE Access, vol. 1, pp. 335–349, May 2013.
[21] M. Agiwal, A. Roy, and N. Saxena, “Next generation 5G wireless
networks: A comprehensive survey,” IEEE Communications Surveys &
Tutorials, vol. 18, no. 3, pp. 1617–1655, 2016.
[22] J. Andrews, S. Buzzi, W. Choi, S. Hanly, A. Lozano, A. Soong, and
J. Zhang, “What will 5G be?” IEEE Journal on Selected Areas in
Communications, vol. 32, no. 6, pp. 1065–1082, 2014.
[23] W. Roh, J.-Y. Seol, J. Park, B. Lee, J. Lee, Y. Kim, J. Cho, K. Cheun, 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,
2014.
[24] S. Kang, M. Mezzavilla, A. Lozano, G. Geraci, W. Xia, S. Rangan,
V. Semkin, and G. Loianno, “Millimeter-wave UAV coverage in urban
environments,” in Proc. IEEE Globecom, 2021, pp. 1–6.
[25] 3GPP TR 38.901, “Study on channel model for frequencies from 0.5 to
100 GHz (Release 16),” Dec. 2019.
[26] 3GPP TR 37.840, “Study of radio frequency (RF) and electromagnetic
compatibility (EMC) requirements for active antenna array system
(AAS) base station (Release 12),” Dec. 2013.
[27] W. Xia, S. Rangan, M. Mezzavilla, A. Lozano, G. Geraci, V. Semkin, and
G. Loianno, “Millimeter wave channel modeling via generative neural
networks,” in Proc. IEEE Globecom Workshops, 2020, pp. 1–6.
[28] “Wireless InSite 3D wireless prediction software,” available on-line at
https://www.remcom.com/wireless-insite-em-propagation-software.
[29] V. Semkin et al., “Lightweight UAV-based measurement system for air-
to-ground channels at 28 GHz,” in Proc. IEEE PIMRC, 2021.
[30] W. Xia, S. Rangan, M. Mezzavilla, A. Lozano, G. Geraci, V. Semkin,
and G. Loianno, “Generative neural network channel modeling for
millimeter-wave UAV communication,” IEEE Transactions on Wireless
Communications, to appear. Available as arXiv:2012.09133, 2022.
[31] 3GPP TS 36.213, “Technical specification group radio access network;
evolved universal terrestrial radio access (E-UTRA),” Sep. 2011.
[32] A. Lozano, “Long-term transmit beamforming for wireless multicast-
ing,” in Proc. IEEE ICASSP, vol. 3, 2007, pp. 417–420.
[33] R. W. Heath Jr. and A. Lozano, Foundations of MIMO Communication.
Cambridge University Press, 2018.
[34] T. Cuvelier and R. W. Heath Jr., “MmWave MU-MIMO for Aerial
Networks,” in Proc. ISWCS, Aug 2018, pp. 1–6.
[35] R. A. Stirling-Gallacher and M. S. Rahman, “Multi-user MIMO strate-
gies for a millimeter wave communication system using hybrid beam-
forming,” in Proc. IEEE ICC, 2015, pp. 2437–2443.

[36] F. G´
omez-Cuba and M. Zorzi, “Optimal link scheduling in millimeter
wave multi-hop networks with MU-MIMO radios,” IEEE Transactions
on Wireless Communications, vol. 19, no. 3, pp. 1839–1854, 2019.