Content uploaded by Minseok Kim
Author content
All content in this area was uploaded by Minseok Kim on Jul 16, 2022
Content may be subject to copyright.
A 24/60-GHz Dual-Band Double-Directional
Channel Sounder Using COTS Phased Arrays
Minseok Kim, Hibiki Tsukada, Keiichiro Kumakura,
Riku Takahashi, Naoya Suzuki
Graduate School of Science and Technology,
Niigata University, Niigata, Japan
Email: mskim@ieee.org
Hirokazu Sawada, Takeshi Matsumura
National Institute of Information and
Communications Technology (NICT),
Yokosuka, Japan
Email: {sawahiro, matsumura}@nict.go.jp
Abstract—The use of millimeter-wave (mm-wave) bands has
been expanding in new wireless communication systems such
as fifth-generation (5G) mobile systems and WiGig WLANs.
Since mm-waves have quasi-optical propagation characteristics,
it is necessary to develop a channel model that can cover site-
specific characteristics of various environments. For this reason,
a large amount of channel measurement data measured in a
wide range of frequency bands covering mm-wave in various
environments should still be required. In particular, it isnecessary
to obtain the angular channel characteristics on both sides of the
transmitter and receiver, but this requires considerable time and
cost. The authors’ previous study developed a 60-GHz double-
directional (D-D) channel sounder using commercial-off-the-shelf
(COTS) phased-array beamforming RFs, and it has currently
been extended to realize 24/60-GHz dual-band simultaneous
measurement. In this paper, we detailed the design of the dual-
band D-D channel sounder. As a demonstration of the operation,
some measurement results taken in typical urban macro-cell
(UMa)/micro-cell (UMi) scenarios were quickly reported.
Index Terms—millimeter-wave, channel sounder, angle-of-
arrival, beamforming, phased array, cluster, channel model
I. INTRODUCTION
In millimeter-wave (mm-wave) wireless systems the site-
specific nature of the propagation channels should be carefully
considered [1], [2]. To achieve an accurate channel model
reflecting the variety of the site-specific characteristics, a large
amount of measurement data in various scenarios should be
collected [3]. Double-directional (D-D) characteristics at both
the transmitter (Tx) and receiver (Rx) are essential information
for massive multiple-input-multiple-output (MIMO) antenna
arrays and beamforming-based mm-wave transmission tech-
niques. However, D-D channel acquisition is very time- and
cost-intensive task [4]–[8].
In many existing studies, a measurement method rotating a
high-gain antenna having a sharp directivity was common to
acquire the angular characteristics of millimeter-wave chan-
nels, but it takes an enormous amount of time for a D-
D full angle scan. Further, several studies [9], [10] showed
that high-resolution angular channel measurements can be
performed at high speed by adopting a beam switching method
based on a phased array. However, they require higher cost
and more time to develop a custom phased array antenna.
Therefore, in our previous study, a low-cost mm-wave channel
sounding system leveraging commercial off-the-shelf (COTS)
beamforming RF transceivers [11] has been developed, which
enables a fast D-D channel acquisition using electronic beam
steering in the 360◦full azimuth angle range using phased
array antennas [12]. Currently, it is extended to realize 24/60-
GHz dual-band simultaneous measurement. In this paper, we
detailed the design of the dual-band D-D channel sounder. To
demonstrate the operation, some measurement results taken
in typical urban macro-cell (UMa)/micro-cell (UMi) scenarios
are quickly reported.
II. 24/60-GHZ DUAL-BAND CHANNEL SOUNDER
A. Hardware configuration
Fig. 1(a) shows the developed system, which consists of
custom baseband processing units [13], [14] and COTS-
phased array antenna beamforming transceivers (EVK02001
for 24 GHz and EVK06002 for 60 GHz, Sivers IMA) [11].
In each transceiver, a narrow beam in the azimuth plane is
synthesized using a 2×8planar array and 16-element uniform
linear array (ULA) for the 24 and 60 GHz bands, respectively,
at both Tx and Rx [15]. The transmit power is approximately
32 and 41 dBm in terms of equivalent isotropic radiated power
(EIRP) for the 24 and 60 GHz respectively. The HPBWs of
the broadside beam patterns are approximately 15◦and 6◦for
the 24 and 60 GHz, respectively, in the azimuth plane. Further,
those in the elevation plane are 45◦for the 24 GHz phased
arrays, and 18◦for Rx and 45◦for Tx for the 60 GHz phased
arrays, respectively.
The scanning range of 90◦is covered by 5 Tx and 5 Rx
beams and 11 Tx and 12 Rx beams for 24 and 60 GHz,
respectively. By using four phased array antennas directing
toward −135◦,−45◦,+45◦and +135◦(an RF head), this
system is extended to the 360◦azimuth angle sweep. The
24-GHz RF heads were stacked on the 60-GHz RF heads
as shown in Fig. 1(b). Employing a dual 4×4MIMO time
division multiplexing (TDM) scheme to measure 32 (= 2·4·4)
channels in one scoop for rapid acquisition of full azimuth
angle range, as shown in Fig. 1(a). Fig. 2 shows the measured
beam patterns. The complete measurement consists of 132
MIMO measurement blocks for angle scanning with 11 Tx
and 12 Rx beams for both bands, which takes approximately
5 minutes [12].
(a) Schematic diagram
360 deg
camera
RF/Antenna
(24GHz)
RF/Antenn
a (60GHz)
3
6
0
d
e
g
c
a
m
e
r
a
R
F
/
A
n
t
e
n
n
a
(
2
4
G
H
z
)
R
F
/
A
n
t
e
n
n
a
(
6
0
G
H
z
)
(b) Phased arrays
Fig. 1. Dual-band channel sounder.
TABLE I
SYSTEM PARAMET ERS
Parameters 24-GHz band 60-GHz band
Carrier frequency 24.15 GHz 58.32 GHz
Signal bandwidth 200 MHz 400 MHz
Sounding signal Multitone Multitone
(N= 512) (N= 1024)
FFT points, Nf2048 2048
Sampling rates 800 Msps 800 Msps
Delay resolution 5.0ns 2.5ns
Delay span 2.56 µs2.56 µs
EIRP 32 dBm 41 dBm
HPBW Az : 15◦
El : 45◦
Az : 6◦
El @ Tx : 45◦
El @ Rx : 18◦
Polarization Vertical Vertical
The multitone allocation and the waveform of the sounding
signal are shown in Fig. 3 [13]. As shown in Table I, the
number of samples (Nf) for single waveform is 2,048 for
both frequencies, the numbers of tones are 512 in 200 MHz
and 1,024 in 400 MHz for the 24 GHz and 60 GHz bands,
respectively. As a result, the delay resolution is 5.0ns and
2.5ns for the 24 GHz and 60 GHz bands, respectively. The
maximum measurable delay is 2.56 µs.
As described in [16], the actual dynamic range of this
system is determined by the signal-to-noise-power-ratio (SNR)
in the channel impulse response (CIR), which ranges between
the over-the-air (OTA) reference level and the noise level
as DR = SNROTA +Gproc ≈51 dB where Gproc =
10 log10 N≈30, which indicates that the processing gain
of discrete Fourier transform (DFT) where Ndenotes the
number of tones. SNROTA denotes the SNR in the OTA
calibration which is approximately 21 dB. Here, decreasing
the transmit power by 20 dB with respect to the value used
in actual long-distance measurement, the OTA calibration for
outdoor measurement is carried out at the Tx-Rx distance
of 3m. This means the measurement distance must exceed
30 m. From above, the maximum measurable path loss is
PLmax = DR−PGOTA ≈148.5 dB where PGOTA =−97.5
dB denotes the path gain (PG) at the OTA calibration at the
equivalent distance of 30 m.
Regarding frequency and timing synchronization, as shown
(a) 24 GHz (Tx: 5 beams, Rx: 5 beams)
(b) 60 GHz (Tx: 11 beams, Rx: 12 beams)
Fig. 2. Beam patterns of phased arrays which are normalized with respect
to the maximum value of the center beam).
in Fig. 1(a), the GNSS (GPS) controlled 10-MHz Rubidium
clock reference at each side of Tx and Rx are used for timing
synchronization and reference clock signal generation (122.88
and 45 MHz for 24 and 60 GHz RFs, respectively). The trans-
mission and reception of the measurement block of snapshots
are executed based on the internally generated trigger pulses,
and those of Tx and Rx are synchronized in the calibration
procedure via cable connection before they will be physically
separated. In untethered configuration, the Rubidium clock
reference phase offset generates the trigger offset between Tx
and Rx. It notes that during the channel sounding campaigns
in several hours, the delay time is inevitably deviated gradually
at every snapshot. The frequency offset between the separate
frequency references results in the linear increase in the trigger
pulse offset. The slope of the pulse offset is about 0.1138
ns/min, which is within the range of the expected offset
value based on the precision of GPS synchronized Rubidium
(a) 24 GHz (W= 200 MHz, Nf= 512, Oversampling ratio= 4)
(b) 60 GHz (W= 400 MHz, Nf= 1024, Oversampling ratio= 2)
Fig. 3. Sounding signal.
oscillator but 10 times worse than Cesium standard [17].
During the 5-minute measurement duration of this system
the delay time deviation may be negligible. In addition, this
system does not require strict phase coherence among multiple
snapshots obtained within a measurement because the angle-
resolved power spectra are used in the post-processing. The
inter-measurement delay time deviation generated during the
measurement campaign can be simply fixed based on the linear
model [16] or simply by line-of-sight (LoS) component.
III. DUAL-BAND SIMULTANEOUS MEASUREMENT IN
URBAN CELLULAR SCENARIOS
A. Measurement scenarios
In order to demonstrate the operation of the developed
system, some measurement results taken in UMa and UMi
scenarios are presented. A measurement campaign was carried
out in typical urban environment around JR Kannai Station and
Yokohama World Porters, Yokohama, Kanagawa, Japan. The
aerial maps of two measurement environments are shown in
Fig. 5. The channel measurement was conducted in downlink
(BS: Tx, MS: Rx) in LoS condition. Regarding the UMa
scenario, the BS was installed on the roof of a 8-story building
(BS antenna height: 31.0m), and the MS was moved along
49 points on the sidewalk where it was LoS or obstructed
LoS (OLoS) condition and the distance between the BS and
MS antennas ranged between 40 ∼350 m. On the other
hand, in the UMi scenario, the height of the BS antenna
installed on the sidewalk was approximately 3m, and the
MS was moved along 25 points on the sidewalk it was LoS
condition in most of the MS points and the distance between
the BS and MS antennas ranged between 27 ∼200 m, which
(a) 24 GHz
(b) 60 GHz
Fig. 4. OTA calibration example (Tx1-Rx1)
corresponds to a typical street level mm-wave access link as
an ultra-high-rate hot spot, where a BS is typically mounted
on lamp posts and an MS is held by a person (MS antenna
height: 1.5m). The measurement data imperatively included
some influence of cars and pedestrians running around the
MS. In the measurement, the D-D angular delay channel
responses with respect to the azimuth angles of depature and
arrival, ϕTand ϕR, respectively, were obtained. A dual 4×4
MIMO measurement configuration was used to achieve the
360◦azimuth angle sweeps at both Tx and Rx for both 24
and 60 GHz.
B. Power spectra
In the measurements, the band-limited angle-resolved chan-
nel transfer functions (CTF) denoted as
Hk,nT,nR(1)
were obtained where NTand NRangular samples are acquired
over a given angle range by electronically steering the highly
directive beam pattern. Then, the CIR is obtained via discrete
inverse Fourier transform of the measured CTF as
hn,nT,nR=F−1{Hk,nT,nR}(2)
where ndenotes the delay time index of arrival
(DToA), and nTand nRdenote the pointing
angle indices. The delay tap is denoted by
{n∆τ|n= 0, ..., N −1}where ∆τ= 1/∆f. The pointing
angles are obtained by {nT∆nT|nT= 0, ..., NT−1},
{nR∆nR|nR= 0, ..., NR−1}where ∆nTand ∆nRdenote
certain (actually unequal) intervals of Tx and Rx beams,
(a) Urban macro-cell (UMa).
Google Earth
(b) Urban micro-cell (UMi).
Fig. 5. Topview of measurement environment.
0 500 1000
Delay [ns]
-150
-140
-130
-120
-110
-100
-90
Path Gain [dB]
24GHz
0 500 1000
Delay [ns]
-150
-140
-130
-120
-110
-100
-90
Path Gain [dB]
60GHz
Fig. 6. Synthesized PDP measured at NW16 (UMa)
(dashed lines: LoS delay and FSPG).
24 GHz
60 GHz
MS
(a) From BS side.
BS
60 GHz
24 GHz
(b) From MS side.
Fig. 7. Measured ADPS at NW16 (UMa).
respectively. The double-directional angle delay power spectra
(DDADPS) is obtained by
Pn,nT,nR=E{|hn,nT,nR|2}.(3)
The omnidirectional power delay profile (PDP) and the az-
imuth delay power spectrum (ADPS) are synthesized as
PDP(n) =
nT,nR
Pn,nT,nR,(4)
ADPSBS(n, nT) =
nR
Pn,nT,nR,(5)
ADPSMS(n, nR) =
nT
Pn,nT,nR.(6)
Fig. 6 shows the example omnidirectional PDPs measured
at the MS position of NW16 where the radial distance is
approximately 90 m as shown in Fig. 5(a). Here, it is seen that
the PG of the first arrival path is significantly lower than the
free-space path gain (FSPG) at both frequencies. One possible
reason is a small HPBW of the beam pattern in the vertical
plane, which degrades the PG in case of the radial distance
smaller than 200 m for the 60 GHz phased array. Moreover,
it is also reasonable that a partial blockage of the first Fresnel
zone should occur because there existed several buildings,
lamp posts and trees close to the LoS, and the fading by the
interference from the ground reflected (GR) wave which is
nonresolvable in most cases. Fig. 7 shows the ADPS from both
the BS and MS sides obtained at NW16. From these results, it
can be seen that the location of the LoS and the other dominant
paths which are possibly single- and double-bounce reflected
waves match well in both frequency bands. We can also see
that the specular reflection seems much more dominant in 60
GHz band. However, it should be noted that each system has
different resolution, namely, the delay resolutions are 5ns
and 2.5ns, and the HPBWs of the 24 and 60 GHz phased
arrays are 15◦and 6◦, respectively. The analysis should be
elaborated by removing such system characteristics through
the post-processing for more accurate comparison.
C. Path loss
As described above, the LoS path is visible in most BS
positions in both measurement scenarios. The behavior of the
50 100 150 200 250 300
Distance [m]
-140
-120
-100
-80
Path Gain [dB]
60 GHz
Two-ray
Urban Macro
50 100 150 200 250 300
Distance [m]
-140
-120
-100
-80
Path Gain [dB]
24 GHz
Two-ray
(a) UMa
102
Distance [m]
-150
-100
Path Gain [dB]
60 GHz
Two-ray
Urban Micro (Street)
102
Distance [m]
-150
-100
Path Gain [dB]
24 GHz
Two-ray
(b) UMi
Fig. 8. Path gains.
path loss characteristics is described using the power sum of
the DDADPS as PG = n,nT,nR
˜
Pn,nT,nRwhere ˜
Pdenotes
the noise filtered DDADPS. Fig. 8 shows the PG as a function
of the radial distance between the BS and MS antennas. As
a reference, the values obtained by the widely accepted two-
ray model are also shown. In the two-ray model, a simple
Gaussian beam with the same HPBW as in the measurement
was used as a beampattern in the vertical plane. Interestingly,
it is found that the PG is larger than the two-ray model in
the small distance region in the UMa scenario. It is because
the GR wave is resolvable in the delay domain. However, the
fading generated by the GR wave can be well observed in the
large distance region where the GR wave is nonresolvable.
From these results, it is seen that the GR wave is the most
dominant component in both scenarios at both frequencies.
IV. SUMMARY
In this paper, the design of a new channel sounder which
can perform a D-D channel sounding at both 24 and 60 GHz
simultaneously. The dual-band operation was verified from
some measurement results taken in two typical UMa and UMi
scenarios. Further, the PG analysis revealed that the GR wave
has significant influence on the PG in both scenarios at both
frequencies.
ACKNOWLEDGEMENT
This research has been conducted under the contract “R&D
for the realization of high-precision radio wave emulator in
cyberspace” (JPJ000254) made with the Ministry of Internal
Affairs and Communications of Japan.
REFERENCES
[1] 3GPP, Technical specification group radio access network; Study on
channel model for frequency spectrum above 6 GHz. TR 38.900, 3rd
Generation Partnership Project (3GPP), Jun. 2016.
[2] “Channel Models for IEEE 802.11ay,” IEEE Document 802.11-
15/1150r2, Sept. 2015.
[3] M. Shafi et al., “Microwave vs. Millimeter-Wave Propagation Channels:
Key Differences and Impact on 5G Cellular Systems,” IEEE Commun.
Mag., vol. 56, no. 12, pp. 14–20, Dec. 2018.
[4] M. Kim, S. Kishimoto, S. Yamakawa, K. Guan, “Millimeter-Wave Intra-
Cluster Channel Model for In-Room Access Scenarios,” IEEE Access,
vol. 8, pp. 82042-82053, May 2020.
[5] M. Kim, T. Iwata, S. Sasaki, J. Takada, “Millimeter-Wave Radio Channel
Characterization using Multi-Dimensional Sub-Grid CLEAN Algorithm,”
IEICE Trans. Commun., vol.E103-B, no.7, Jul. 2020.
[6] C. Ling, X. Yin, R. M¨
uller, S. H¨
afner, D. Dupleich, C. Schneider, J. Luo,
H. Yan, R. Thom¨
a, “Double-Directional Dual-Polarimetric Cluster-Based
Characterization of 70-77 GHz Indoor Channels,” IEEE Trans. Antennas
Propag., vol. 66, no. 2, pp. 857–870, Feb. 2018.
[7] X. Yin, C. Ling and M. Kim, “Experimental multipath-cluster characteris-
tics of 28-GHz propagation channel,” IEEE Access, vol. 3, pp.3138–3150,
2015.
[8] G. R. MacCartney, and T. S. Rappaport, “73 GHz Millimeter Wave
Propagation Measurements for Outdoor Urban Mobile and Backhaul
Communications in New York City,” Proc. 2014 IEEE Intl. Conf. Com-
mun. (ICC), pp. 4862–4867, Jun. 2014.
[9] D. Caudill, P. B. Papazian, C. Gentile, J. Chuang, N. Golmie, “Omnidi-
rectional Channel Sounder With Phased-ArrayAntennas for 5G Mobile
Communications,” IEEE Trans. Microwave Theory Tech., vol. 67, no. 7,
Jul. 2019.
[10] A. Chopra, A. Thornburg, O. Kanhere, A. Termos, S. Ghassemzadeh,
T. Rappaport, “Real-Time Millimeter Wave Omnidirectional Channel
Sounder Using Phased Array Antennas,” Proc. 2020 IEEE Global Com-
munications Conference, Taipei, Taiwan, 2020.
[11] EVK06002/00, SIVERS IMA, https://www.siversima.com/product/evk-
06002-00/
[12] M. Kim, S. Tang, K. Kumakura, “Fast Double-Directional Full Az-
imuth Sweep Channel Sounder Using Low-Cost COTS Beamforming RF
Transceivers,” IEEE Access, Vol.9, pp. 80288–80299, Jun. 2021.
[13] M. Kim, J. Takada, K. Saito, “Multi-Dimensional Radio Channel Mea-
surement, Analysis and Modeling for High Frequency Bands,” IEICE
Trans. Commun., vol.E101-B, no. 2, pp.293–308, Feb. 2018.
[14] M. Kim, H. Kinh Pham, Y. Chang, J. Takada, “Development of Low-
Cost 60-GHz Millimeter Wave MIMO Channel Sounding System,” 6th
Global Symposium of Millimeter Waves (GSMM 2013), Apr. 2013.
[15] I. Aziz, R. Dahlb¨
ack, E. ¨
Ojefors, K. S¨
ogren, A. Rydberg and D. Dancila,
“60 GHz Compact Broadband Antenna Arrays with Wide-Angle Beam
Steering,” J. Eng., vol. 2019, no. 8, pp. 5407–5414, Aug. 2019.
[16] M. Kim, K. Kumakura, S. Tang, H. Tsukada, “Multipath Clusters
Observed in Outdoor Open Square Environments at 60 GHz,” The 2021
IEEE 94th IEEE Vehicular Technology Conference (VTC2021), Sept.
2021.
[17] Y. Konishi, M. Kim, Y. Chang, J. Takada, “Versatile Radio Channel
Sounder for Double Directional and Multi-link MIMO Channel Mea-
surements at 11 GHz,” IEICE Trans. Electron., Vol.E97-C, No.10, pp.
994–1004, Oct. 2014.
