Large-Scale Parameters of Spatio-Temporal Short-Range Indoor Backhaul Channels at 140 GHz

Abstract

The use of above-100 GHz radio frequencies would be one of promising approaches to enhance the fifth-generation cellular further. Any air interface and cellular network designs require channel models, for which measured evidence of largescale parameters such as pathloss, delay and angular spreads, is crucial. This paper provides the evidence from quasi-static spatiotemporal channel sounding campaigns at two indoor hotspot (InH) scenarios at 140 GHz band, assuming short-range backhaul connectivity. The measured two InH sites are shopping mall and airport check-in hall. Our estimated omni-directional large-scale parameters from the measurements are found in good match with those of the Third Generation Partnership Project (3GPP) for new radios (NR) channel model in InH scenario, despite the difference of assumed link types and radio frequency range. The 3GPP NR channel model is meant for access links and said to be valid up to 100 GHz, while our measurements cover shortrange backhaul scenarios at 140 GHz. We found more deviation between our estimated large-scale parameters and those of the 3GPP NR channel model in the airport than in the shopping mall.

Full text

Large-Scale Parameters of Spatio-Temporal
Short-Range Indoor Backhaul Channels at 140 GHz
Sinh L. H. Nguyen†, Katsuyuki Haneda†, Jan J¨
arvel¨
ainen†, Aki Karttunen†and Jyri Putkonen∗
†Aalto University, School of Electrical and Engineering, Finland, katsuyuki.haneda@aalto.fi
∗Nokia Bell Labs, Finland
Abstract—The use of above-100 GHz radio frequencies would
be one of promising approaches to enhance the fifth-generation
cellular further. Any air interface and cellular network designs
require channel models, for which measured evidence of large-
scale parameters such as pathloss, delay and angular spreads, is
crucial. This paper provides the evidence from quasi-static spatio-
temporal channel sounding campaigns at two indoor hotspot
(InH) scenarios at 140 GHz band, assuming short-range backhaul
connectivity. The measured two InH sites are shopping mall and
airport check-in hall. Our estimated omni-directional large-scale
parameters from the measurements are found in good match
with those of the Third Generation Partnership Project (3GPP)
for new radios (NR) channel model in InH scenario, despite the
difference of assumed link types and radio frequency range. The
3GPP NR channel model is meant for access links and said to
be valid up to 100 GHz, while our measurements cover short-
range backhaul scenarios at 140 GHz. We found more deviation
between our estimated large-scale parameters and those of the
3GPP NR channel model in the airport than in the shopping
mall.
I. INTRODUCTION
The commercial deployment of fifth-generation (5G) cel-
lular wireless has already commenced in different parts of
the world. Conceptualization of future generation wireless,
probably called beyond-5G or sixth-generation, has already
been in progress. One of core technological elements for
future-generation wireless systems may be exploitation of
unexplored frequency bands such as above-100 GHz. It is not
yet known if such high frequency bands are useful for cellular
applications or not. It may be restricted to a short-range,
back/front-haul and extremely high-data-rate communications
such as data center applications due to physically small but
electrically large antennas and high diffraction losses. Small
cellular coverage for hotspots may also be possible.
Radio channel models lay foundation in feasibility study
of radio systems to be designed and deployed. The one
developed by Third Generation Partnership Project (3GPP)
is one of the most widely used models across industry for
standardization of cellular wireless. The latest channel model
for New Radio (NR) [1] is said to work up to 100 GHz,
while their channel model parameters were derived from real-
world measurements up to 83 GHz carrier frequency. Above
83 GHz, there are handful of works reporting wave propaga-
tion and wave-material interaction. For example, Piesiewicz
et al. [2] analyze reflections from materials commonly found
in living environments, e.g., plasterboard and wall papers
between 100 GHz and 1THz, taking into account their surface
roughness. Losses of wave penetration through materials in
our living spaces are reported in [3]–[6], clearly indicating
increase of the penetration loss as the frequency goes up to
10 THz. Rappaport et al. [7] provide mathematical modeling
of scattering from a drywall. Abbasi et al. [8] measure radar
cross sections of human body at 140 and 220 GHz, indicating
the greater cross section as the frequency increases.
A very few short-range multipath channel measurements
have also been reported for above-83 GHz frequencies, in-
cluding those by Hanssens et al. [9] analyzing specular and
diffusive multipath propagation mechanisms at 94 GHz and
reporting dominance of specular mechanism and hence small
delay spread up to 15 ns. Similar level of delay spread was
reported by Pometcu and D’Errico [10] for transmit-receive
(Tx-Rx) distances up to 11 m, according to their indoor
measurements covering 126-156 GHz frequencies. The work
also shows slightly larger pathloss exponent compared to
below-100 GHz frequencies at the same indoor site. Challita et
al. [11] study indoor spatio-temporal channels at 94 GHz based
on measurements, showing small-scale characterization such
as Rician K-factors and antenna correlation. Vitucci et al. [12]
compare angular profiles of indoor multipath propagation at
10,60 and 300 GHz, showing that observed power angular
spectrum are consistent across the tested radio frequencies
with clearer distinction of specular multipath components
against diffusive multipaths at higher frequencies. Guan et
al. [13] report multipath propagation channels between a train
wagon and outside at 300 GHz, observing a very limited
number of multipath. Cheng and Zaji´
c [14]–[16] study pathloss
and delay dispersion characteristics of short range radio links
at 300 GHz assuming applications to data center scenarios. The
works are based on solid measurements of up to 2.1m Tx-
Rx separation. Abbasi et al. [17] reports a short-range indoor
channel properties for 140-220 GHz, indicating dominance of
line-of-sight (LOS) signal in path gain. The existing papers
of indoor channel measurements for above-100 GHz cellular
applications, e.g., indoor hotspots, are still insufficient for
comprehensive analyses of such channels. We need to accu-
mulate experimental evidence at above-100 GHz frequencies
to study if the existing channel models can reproduce the
observed reality of multipath propagation.
The present paper continues analyses provided in earlier
publication of the same authors [18], where large- and small-
scale parameters of 28 and 140 GHz radio channels are
compared for a shopping mall scenario. The present paper
arXiv:2009.13209v1 [eess.SP] 28 Sep 2020

Rx horn antenna
(19 dBi) Tx biconical
antenna
(0 dBi)
Down
converter
Rotator
Up
converter
Signal
Generator
LO signal
IF signal
RF signal
141.1-145.1 GHz
10 MHz sync
Control
PC
Waveguide
E/O
O/E
Optical fiber
cable 200m
Splitter
VNA
Fig. 1. Schematic of channel sounder setup.
focuses on 140 GHz measurements and discuss additional
measurements from the same shopping mall and airport check-
in hall, which have not been analyzed yet. Both measurements
are conducted with short-range backhaul quasi-static scenarios
by setting antenna heights above human heights and hence
avoiding their influence on links. When we compare our esti-
mated omni-directional large-scale parameters with the indoor
channel model in 3GPP NR [1], good agreement was observed
despite differences of assumed link types and applicable radio
frequency range.
The rest of the paper is organized as follows: Section II
details our channel sounder and measurement sites. Section IV
provides analysis of large-scale parameters according to our
140 GHz indoor channel measurements, and finally we con-
clude our paper in Section V.
TABLE I
INSTRUMENTS AND COMPONENTS CONSISTING OF THE CHANNEL
SOUNDER
Instrument Model
VNA Keysight ENA E8363A
LO Rohde & Schwarz SMR 60
Rotator Diamond Engineering 6100
Up-/down-converter Virginia Diodes MixAMC 297/MixAMC 298
RF-over-fiber Miteq LBL-50K4P5G (IF channel)
Miteq SCML-100M18G (LO channel)
Optical fiber Milcon P/N 2060115
Calibration attenuator Elmika VA-02E
Tx antenna Mi-Wave OmniDirectional Antenna WR-6
Rx antenna Flann microwave custom horn antennas
II. CHANNEL SOUNDING
This section provides an overview of measurement appara-
tus for channel sounding, along with two indoor sites where
the measurements were performed.
A. Measurement apparatus and its instability
140 GHz spatio-temporal channel sounder in Aalto Uni-
versity, Finland, uses a vector network analyzer (VNA) with
up- and down-converters for frequency extension and radio
TABLE II
SPECIFICATIONS AND PARAMETER SETTINGS IN CHANNEL SOUNDING
Properties Values
IF, power 0.1-4.1GHz, 5dBm
LO frequency, power 11.75 GHz, −2dBm
RF, power to Tx antenna 141.1-145.1GHz, −7dBm
PDP dynamic range 125 dB
Tx / Rx height 1.9 / 1.9 m (shopping mall)
1.7 m on the second floor /
2.1 m on the third floor (airport)
Link distance range 3−65 m (shopping mall)
15 −51 m (airport)
Tx antenna pattern 0 dBi gain, 60◦elevation beamwidth
Rx antenna pattern 19 dBi gain, 10◦azimuth beamwidth
40◦elevation beamwidth
frequency (RF)-over-fiber system for range extension as il-
lustrated in Fig. 1 [18]. The sounder consists of precision
instruments and RF components summarized in Table I. The
up- and down-converters include ×12 frequency multiplier for
local oscillator (LO) signals that are mixed with intermediate
frequency (IF) signals from the VNA. The RF-over-fiber
system consists of two sets of laser diodes as electric-to-
optical (E/O) converter and photo diodes for optical-to-electric
(O/E) conversion. The two sets are intended for sending IF
and LO signals, which are multiplexed in a single multi-
mode optical fiber of 200 m length. The RF-over-fiber system
improves a dynamic range of the sounder by 30 dB at 10 m
distance between the transmit (Tx) and receive (Rx) antennas,
compared to using RF cables to feed the IF and LO signals to
the up-converter. We chose an omni-directional bicone and
directional horn antennas as the Tx and Rx antennas. The
Rx horn antenna was rotated on the azimuth plane to obtain
angularly resolved channel impulse responses.
As discussed in [18], transfer functions of the sounder is
unstable over time just after instruments are turned on. The
instability originates from the LO channel of RF-over-fiber
system, which is then magnified at the frequency multiplier in
the up- and down-converters. It turned out during laboratory
tests that transfer functions of the channel sounder become
stable after the instruments are on for two hours, giving
uncertainty of 0.2dB at median level and of 2dB in the worst
case of a peak gain of channel impulse responses, according
to back-to-back calibration measurements of the sounder. The
measurements were performed by connecting the RF output of
the up-converter and RF input of the down-converter through
an attenuator to avoid overloading the down-converter. The
calibration measurements were performed right before and
after field measurements, in addition to laboratory tests, in
order to estimate transfer functions of the sounder.
B. Measurement sites
1) Shopping mall: The first measurement site was a shop-
ping mall “Sello” in Espoo, Finland. It is a modern, four-story
building with approximate dimensions of 120 ×70 m2and
has a large open space in the middle as seen in Fig. 2(a).
The floor plan of the measurements are shown in Fig. 3(a).
In total, 18 Tx-Rx links were measured, with the Rx antenna

Rx1
Tx1
(a)
Tx16; link is
obstructed by check-
in kiosks
(b)
Fig. 2. Photos of quasi-static channel measurements without moving people
interfering the link. (a) Shopping mall and (b) airport.
fixed at a single location. The Tx antenna was moved along
the corridor and around the open space. The antenna locations
were chosen so that time-varying nature of the channels is
minimized, such as link blockage due to a human body. We
furthermore performed the measurements during there was no
moving people or objects interfering the measured channels,
leading to quasi-static channels. Both Tx and Rx antennas
were elevated to 1.9m high above the floor, and the Tx-
Rx distance ranged from 3.9to 65.2m. At two Tx antenna
locations, LOS to Rx antenna was obstructed by static objects,
i.e., pillar or escalator. In each Tx-Rx link measurement, the
Rx horn antenna was rotated on the azimuth plane with 5◦
step across 360◦to obtain angularly resolved channel impulse
responses. While our earlier publication [18] reports analysis
of measurements performed for only 8links, i.e., Tx11-Tx18,
the present paper is based on measurements of 18 links, i.e.,
Tx1-Tx18.
2) Airport check-in hall: The second measurement site
was a check-in hall of Helsinki Vantaa Airport, Terminal
2. The hall looks like Fig. 2(b), while the floor plan of
the site along with Tx and Rx antenna locations are shown
in Fig. 3(b). Similarly to the shopping mall measurements,
the measurements were performed during absence of moving
people or objects interfering the measured channels, leading
to quasi-static channels. Altogether 11 Tx antenna locations
were considered across the hall, among which Tx1 was at
a terrace overlooking the hall. The Rx antenna was fixed at
the same terrace throughout the measurements. The terrace is
3.6m higher than the hall, making Tx1 and Rx antennas higher
than the Tx antennas on the hall by 4.0m. Antenna heights are
summarized in Table II. The Tx-Rx distance ranged from 15 to
51 m. All the measured links have LOS between the Tx and Rx
antennas, except for Tx16 where the LOS link is obstructed
by a self-check-in machine. Similarly to the shopping mall
measurements, Rx horn antenna was rotated across azimuth
angles with 5◦steps for 360◦range for Tx1-Rx link. In other
links, the angular range is limited to 0◦-20◦and 240◦-360◦
as the other azimuth angles point the horn antenna to a wall
just behind the Rx location. The azimuth angles are defined
according to the x-ycoordinate system of Fig. 3(b).
III. CHARACTERIZATION OF MULTIPATH CHANNELS
Three omni-directional large-scale parameters were ana-
lyzed based on our indoor multipath measurements at 140 GHz
band, i.e., pathloss, delay spread and azimuth angular spread.
To this end, we first approximate the measured channels using
band- and antenna aperture unlimited expression of the same,
defined as
h(φ, τ ) =
L
X
l=1
pGaαlδ(φ−φl)δ(τ−τl),(1)
where Ga,α,φ,τare combined gains of the Tx and
Rx antennas, path amplitude, azimuth angle of arrival and
propagation delay; subscript (·)ldenotes an l-th multipath,
1≤l≤L. There is an apparent difference between (1)
and measured channels, exemplified in Fig. 4(a) for Tx7-Rx
link from the airport measurement. What we can measure
and observe is always band- and antenna aperture limited
form that we call a power angular delay profile (PADP)
hereinafter. The approximation of band- and antenna aperture
limited measurements by the mathematically versatile band-
and antenna aperture unlimited formula is performed by find-
ing local maxima of the measured PADP, and assuming that the
maxima correspond to physical multipaths, i.e., Lmultipaths.
Our earlier publication [18] elaborates a simple but robust
method to estimate multipath parameters, i.e., αl,φl,τl, for
1≤l≤Lthrough peak search of the PADP. The combined
gains of the Tx and Rx antennas, Ga= 19 dBi, are known for
antenna gain de-embedding of estimated multipaths assuming
that multipaths are detected at the broadside of the Tx and Rx
antennas.
The large-scale parameters of our interests are derived
from the estimated multipath parameters. We use an ordinal
approach [19] with 1) only analyzing multipaths with high-
enough gain, i.e., between maxl|αl|2
dB and maxl|αl|2
dB −
30 dB and 2) ensuring circular continuity for the azimuth
angles when calculating the azimuth spread. However, con-
cerning 1), since some links typically with a longer Tx-
Rx distance than 30 m have a smaller dynamic range of
PADP than 30 dB, we use the observed dynamic range of
the measurements in the large-scale parameter calculation if
it is less than 30 dB. Here the dynamic range is defined by a
difference between the strongest signal and noise floor levels.

8
9
10
7
6
118
Rx
17
16
15
2
5
4
11
3
12
13
14
Red: LOS Tx
antenna index
Blue: NLOS Tx
antenna index
(a) (b)
Fig. 3. Maps of the Tx and Rx positions on (a) third floor of a shopping mall and (b) airport check-in hall for short-range backhaul channel sounding.
(a) (b) (c)
Fig. 4. Observed channel profiles in the furthest link between Tx and Rx in airport. (a) PADP, (b) PDP and (c) PAS. In (a), no data exist between 20◦and
240◦of azimuth angles.
IV. RES ULT S AN D DISCUSSIONS
This section first summarizes observations from our mea-
surements in terms of PADP, PDP and PAS, along with
our estimates of omni-directional large-scale parameters, i.e.,
pathloss, delay and azimuth angular spreads, after antenna gain
de-embedding. We finally compare the obtained large-scale
parameter estimates from those of a reference channel model,
i.e. 3GPP NR InH model.
A. Observations of multipath channels
For illustration of measured 140 GHz indoor multipath
channels, Figs. 4(b) and 4(c) show power delay profiles (PDP)
and power angular spectrum (PAS). They are obtained by
marginal integration of the PADP, i.e., Fig. 4(a), over azimuth
and delay domains, respectively. As noise power is integrated
as well as multipath signal powers, the noise floor of PADP,
PDP and PAS are different as apparent from Fig. 4. The highest
and lowest noise floors of −110 and −132 dB are observed
in the PAS and PADP, among the three plots.
Figure 4 shows PADP, PDP and PAS of Tx7-Rx link
at airport, with the 45.3m link separation. The PADP is
overlaid by detected multipaths as significant local maxima,
colored according to its gain. The plot shows more paths
than LOS, indicating possibilities to deliver energy from one
link end to another through multiple physical paths for spatial
multiplexing and when LOS is blocked by, e.g., a human body.
There are particularly strong paths at 250 and 275 ns even
though they have much longer delays than other multipaths,
clearly because of specular reflections from large and flat and
metallic side walls of the airport. The dynamic range of the
PADP is 26 dB. Figures 4(b) and 4(c) include reproduced
profile and spectrum according to our multipath estimates
indicated as red lines. They are band- and antenna aperture-
limited form derived by convolving a sinc function and antenna
field pattern with (1). The red curves show successful detection
of many distinct local maxima of the measured PDP and PAS,
while they show more deviation from the measurement at
low levels since they represent mainly noise. The measured
curves have high noise level as a result of accumulating noise
power in deriving the PDP and PAS from the PADP, while the
reproduced curves are free from noise. The overall agreement
between measured and reproduced curves for multipath signals
prove efficacy of the multipath estimation method.

1 5 10 50
d3D [m]
70
80
90
100
110
120
130
140
150
Omni-pathloss [dB]
Shopping mall LOS
Shopping mall OLOS/NLOS
3GPP InH LOS
3GPP InH NLOS
3GPP InH NLOS optional
(a)
1 5 10 50
d3D [m]
0
10
20
30
40
50
Delay spread [ns]
Shopping mall LOS
Shopping mall OLOS/NLOS
3GPP InH LOS
3GPP InH NLOS
(b)
1 5 10 50
d3D [m]
0
10
20
30
40
50
Azimuth angular spread [deg]
Shopping mall LOS
Shopping mall O/NLOS
3GPP InH AOD LOS
3GPP InH AOD NLOS
3GPP InH AOA LOS
3GPP InH AOA NLOS
(c)
(d) (e) (f)
Fig. 5. Large-scale omni-directional channel parameters at 140 GHz. (a) pathloss (b) delay spread and (c) azimuth spread at Rx in shopping mall and (d)
pathloss (e) delay spread and (f) azimuth spread at Rx in airport.
B. Large-Scale Parameters
The estimated large-scale parameters shown in Fig. 5. Each
plot in the figure is overlaid by reference models of 3GPP
NR for the indoor hotspot (InH) scenario [1]. The model
shows that all the studied large-scale parameters in this paper
have frequency dependency. Though the model is verified only
up to 83 GHz measurements, we substitute 140 GHz to the
frequency term. It must also be noted that the 3GPP NR
models are for access links, while our measurements assume
short-range backhaul links. The major differences between the
two types of links are antenna heights, where the former may
have different antenna heights at link ends, while the latter
may have the same elevated antenna heights than human, as
performed in our channel sounding.
1) Shopping mall: For shopping mall scenario, the mea-
sured large-scale parameter estimates at 140 GHz follow the
3GPP InH LOS model parameters well. The mean differences
between the measurements and LOS models are +0.9dB in
pathloss, +1.3ns in delay spread and +18.0◦and −0.2◦in
azimuth angle-of-departure (AoD) and angle-of-arrival (AoA)
spreads. The sign “+” indicates that the measured values are
greater than those in the model. Standard deviations are 3.6dB
in pathloss, 8.5ns in delay spread and 11.6◦in azimuth spread.
The largest differences of delay and angular spreads between
measurements and models are at the two longest Tx-Rx links,
indicating possible influence of a limited dynamic range.
2) Airport: For airport check-in hall, the measured pathloss
follows the 3GPP NR InH LOS model parameters well. The
mean difference between the measurements and LOS model
is −0.9dB, with standard deviation of 3.6dB. However,
delay spread of the LOS model is too small to represent our
measurements at 140 GHz, showing a mean difference and
standard deviation of 54 and 40 ns because of long-delayed
multipaths exemplified in Fig. 4(b). Finally, the measured az-
imuth spread is always smaller than the LOS models probably
because only a limited angular range is scanned at the wall-
side Rx. Two links Tx17-Rx and Tx6-Rx show significantly
smaller values than the LOS models because most multipaths
arrive from a limited angular range. It must be noted again that
azimuth angles of multipaths are estimated at the Rx on the
terrace, which can resemble a base station overlooking the hall.
The mean differences between measurements and LOS models
are −13.4◦and 4.8◦for AoD and AoA, while the standard
deviation is 9.1◦. The agreement between measurements and
3GPP InH LOS models is generally worse than shopping mall
scenario because the airport hall may be bigger than ordinal
InH scenario. More long-delayed multipaths provides greater
delay spread, while longer LOS link results in the reduced
angular range of multipaths.
V. CONCLUDING RE MA RK S
Large-scale parameters of 140 GHz InH links were reported
based on spatio-temporal channel sounding at two sites, i.e.,

shopping mall and airport check-in hall, which covered 18 and
11 measured links respectively. The measurements assumed
short-range backhaul scenarios and were performed during
the absence of moving people so that quasi-static channel
conditions were ensured for the use our channel sounder.
Multipath parameters of the channels, i.e., path magnitude,
AoA and propagation delay, were estimated through peak
detection of the PADP. Tx and Rx antenna gains were de-
embedded from the path magnitude estimates. Our large-
scale parameter estimates at 140 GHz, i.e., pathloss, delay
and azimuth angular spreads, were finally compared with a
reference channel model, i.e., the 3GPP NR InH LOS channel
model in this paper, which was derived from measurements
with RF up to 83 GHz and nominally valid up to 100 GHz.
The comparison showed good agreement in the shopping
mall, despite that the 3GPP NR InH channel model is for
access links while our measurements resemble short-range
backhaul links. The same comparison for airport showed more
deviation between our measurements and the 3GPP NR model.
The present evidence from channel sounding suggests further
comparison between measured reality of multipath channels
and reference channel models, strengthened by more channel
sounding campaigns. Additionally, small-scale multipath char-
acteristics, e.g., clusters, will have to be addressed at 140 GHz
to see if existing reference channel models can represent
them well. Finally, validity of other possible reference channel
models will also be tested against our measurements, e.g., ray-
based site-specific channel models.
ACK NOW LE DG EM EN T
The authors would like to thank Mr. Usman Virk, Dr.
Mamadou Balde and Mr. Bic¸er Sena for their help in chan-
nel sounding and laboratory tests. K. Haneda acknowledges
financial support of European Union Horizon 2020 project
Artificial Intelligence Aided D-band Network for 5G Long
Term Evolution (ARIADNE), proposal #871464.
REFERENCES
[1] 3GPP, “3GPP TR 38.901 v14.3.0: Study on channel model for frequen-
cies from 0.5 to 100 GHz,” Dec. 2017.
[2] R. Piesiewicz, C. Jansen, D. Mittleman, T. Kleine-Ostmann, M. Koch,
and T. K¨
urner, “Scattering analysis for the modeling of THz communi-
cation systems,” IEEE Trans. Ant. Prop., vol. 55, no. 11, pp. 3002–3009,
2007.
[3] J. Kokkoniemi, J. Lehtom¨
aki, and M. Juntti, “Measurements on pene-
tration loss in terahertz band,” in Proc. 10th European Conf. Ant. Prop.
(EuCAP 2016), 2016, pp. 1–5.
[4] J. Kokkoniemi, J. Lehtom¨
aki, V. Petrov, D. Moltchanov, and M. Juntti,
“Frequency domain penetration loss in the terahertz band,” in Proc. 2016
Global Symp. Millimeter Waves (GSMM) and ESA Works. Millimetre-
Wave Tech. Appl., 2016, pp. 1–4.
[5] Y. Xing and T. S. Rappaport, “Propagation measurement system and
approach at 140 GHz-moving to 6G and above 100 GHz,” in Proc.
2018 IEEE Global Commun. Conf. (GLOBECOM), 2018, pp. 1–6.
[6] V. Petrov, J. M. Eckhardt, D. Moltchanov, Y. Koucheryavy, and
T. K¨
urner, “Measurements of reflection and penetration losses in low
terahertz band vehicular communications,” in Proc. 14th European Conf.
Ant. Prop. (EuCAP 2020), 2020, pp. 1–5.
[7] T. S. Rappaport, Y. Xing, O. Kanhere, S. Ju, A. Madanayake, S. Mandal,
A. Alkhateeb, and G. C. Trichopoulos, “Wireless communications and
applications above 100 GHz: Opportunities and challenges for 6G and
beyond,” IEEE Access, vol. 7, pp. 78 729–78 757, 2019.
[8] N. A. Abbasi, A. F. Molisch, and J. C. Zhang, “Measurement of
directionally resolved radar cross section of human body for 140 and
220 GHz bands,” in Proc. 2020 IEEE Wireless Commun. Networ. Conf.
Works. (WCNCW), 2020, pp. 1–4.
[9] B. Hanssens, M. Mart´
ınez-Ingl´
es, E. Tanghe, D. Plets, J. Molina-Garc´
ıa-
Pardo, C. Oestges, L. Martens, and W. Joseph, “Measurement-based
analysis of specular and dense multipath components at 94GHz in an
indoor environment,” IET Microw. Ant. Prop., vol. 12, no. 4, pp. 509–
515, 2018.
[10] L. Pometcu and R. D’Errico, “Channel model characteristics in D-band
for NLOS indoor scenarios,” in Proc. 2019 13th European Conf. Ant.
Prop. (EuCAP 2019), Krakow, Poland, 2019, pp. 1–4.
[11] F. Challita, M. Martinez-Ingles, M. Li´
enard, J. Molina-Garc´
ıa-Pardo, and
D. P. Gaillot, “Line-of-sight massive MIMO channel characteristics in
an indoor scenario at 94 GHz,” IEEE Access, vol. 6, pp. 62361–62 370,
2018.
[12] E. M. Vitucci, M. Zoli, F. Fuschini, M. Barbiroli, V. Degli-Esposti,
K. Guan, B. Peng, and T. K¨
urner, “Tri-band Mm-wave directional
channel measurements in indoor environment,” in Proc. 2018 IEEE 29th
Annual Int. Symp. Personal, Indoor Mobile Radio Commun. (PIMRC
2018), Bologna, Italy, 2018, pp. 205–209.
[13] K. Guan, B. Peng, D. He, J. M. Eckhardt, S. Rey, B. Ai, Z. Zhong, and
T. Krner, “Measurement, simulation, and characterization of train-to-
infrastructure inside-station channel at the terahertz band,” IEEE Trans.
Terahertz Sci. Tech., vol. 9, no. 3, pp. 291–306, 2019.
[14] C. Cheng and A. Zaji´
c, “Characterization of propagation phenomena
relevant for 300 GHz wireless data center links,” IEEE Trans. Ant. Prop.,
vol. 68, no. 2, pp. 1074–1087, 2020.
[15] C. Cheng, S. Sangodoyin, and A. Zaji´
c, “THz cluster-based modeling
and propagation characterization in a data center environment,” IEEE
Access, vol. 8, pp. 56 544–56 558, 2020.
[16] ——, “THz MIMO channel characterization for wireless data center-like
environment,” in Proc. 2019 IEEE Int. Symp. Ant. Prop. USNC-URSI
Radio Sci. Meeting, 2019, pp. 2145–2146.
[17] N. A. Abbasi, A. Hariharan, A. M. Nair, and A. F. Molisch, “Channel
measurements and path loss modeling for indoor THz communication,”
in Proc. 14th European Conf. Ant. Prop. (EuCAP2020), Copenhagen,
Denmark, 2020, pp. 1–5.
[18] S. L. H. Nguyen, J. J¨
arvel¨
ainen, A. Karttunen, K. Haneda, and J. Putko-
nen, “Comparing radio propagation channels between 28 and 140 GHz
bands in a shopping mall,” in Proc. 12th European Conf. Ant. Prop.
(EuCAP 2018), London, UK, 2018, pp. 1–5.
[19] A. F. Molisch, Wireless Communications. Second Ed., Section 6, John
Wiley and Sons, Ltd., Dec. 2010.