Investigation of Channel Properties for 28 GHz Band in Urban Street Microcell Environments Minoru Inomata, Tetsuro Imai, Koshiro Kitao, and Yukihiko Okumura NTT DOCOMO, INC. 3-6 Hikari-no-oka, Yokosuka-shi, Kanagawa, 239-8536, Japan Email: [email protected]Abstract - In Japan, field trials on the fifth generation mobile communications system (5G) were started in order to create a new market by actualizing 5G. The 28 GHz band is the most promising candidate for commercial 5G systems. When designing the service area, site-specific channel properties in urban areas are required. Therefore, in this paper, we investigate site-specific channel properties for the 28 GHz band in an urban street microcell environments and clarify that the paths affect the channel properties. Index Terms — 5G mobile communication system, High frequency bands, Channel properties, Street microcell. 1. Introduction Development of the fifth generation mobile communications system (5G) has been actively investigated [1]-[5]. In Japan, 5G field trials were started in order to create a new market by actualizing 5G [6]. In 5G, high frequency bands higher than 6 GHz will be used in order to provide attractive higher bit rates, and the 28 GHz band is the most promising candidate for commercial 5G systems in Japan. It is assumed that the main service areas for 5G using high frequency bands will be in urban areas. In designing these service areas, since the propagation characteristics is difference for every environment, it is important to understand site-specific channel properties for the 28 GHz band in urban areas. Regarding the channel properties in high frequency bands, based on the 5G system requirements, standardized models have been reported by the 3GPP [4] and ITU-R [5]. These models are useful in evaluating the performance of technical specifications. However, when designing the service area, the site-specific channel properties is required, and since the standardized models are generated by statistical processing the measurement data which is acquired in various sites, site specific channel properties are not clear in standardized models. Therefore, we investigate site-specific channel properties for the 28 GHz band in urban street microcell environments. In a 5G field trial we study on the channel models for path loss, Doppler characteristics, delay profiles and angular profiles. Therefore, in this paper, we report the path loss and Doppler characteristics. We first report on the verification results of the prediction accuracy based on a comparison between measurements and prediction results using the representative model, Rep. ITU-R M.2412 [5]. On the basis of the obtained results, we clarify that the paths affect the channel properties. 2. Measurement Campaign The channel properties were measured in an urban area nearby Tokyo Station, Tokyo, Japan. The streets in the area are surrounded by tall buildings (approximately ten stories or 40 m). In this paper, we report on the path loss and Doppler characteristics. The path loss measurements are taken along routes 1 and 2, as shown in Fig. 1. Route 1 includes non-line of sight (NLOS) and line of sight (LOS) areas. Route 2 is only a NLOS street. The Doppler measurements are recoded along route 3. Route 3 is circular route of 2590 m. The measurements are taken in the 28 GHz bands for a commercial 5G system. The transmission (Tx) antennas are set at a height of 10 m. We used them to transmit continuous waves. A receiver (Rx) antenna is fixed on the roof of a measurement car whose height is 2.3 m. The Tx and Rx antenna radiation patterns are omni-directional. Fig. 1. Measurement routes. 3. Measurement Results Fig. 2 shows comparison results between the measurement results on routes 1 and 2, and the predictions for the ITU-R M.2412 channel model. We calculated the path loss using channel model B in an urban microcell (UMi). Fig. 2 shows that the ITU-R model predictions in the LOS environment on route 1 are relatively accurate. A quantitative evaluation is given where median prediction errors are calculated. The error between the prediction results and measurement results is approximately 2 dB. However, in the NLOS environments on routes 1 and 2, Fig. 2 shows that the measured path loss becomes smaller than that for the prediction results. In this case, the difference is approximately 7.0 dB lower on route 1, and 10 dB lower on route 2. In order to the analyze the path which affects the path loss, on route 3 including the street on routes 1 and 2, we analyzed the propagation path based on Doppler characteristics. The Doppler frequency shift, Δf, is calculated using the following equation ∆ = cos = cos (1) Tx 240 m 150 m 221 m 196 m Route 1 Route 2 877 m 396 m 926 m Route 3 ① ② ③ ④ Rx [WeB1-1] 2018 International Symposium on Antennas and Propagation (ISAP 2018) October 23~26, 2018 / Paradise Hotel Busan, Busan, Korea 11
2
Embed
Investigation of Channel Properties for 28 GHz Band in Urban Street Microcell … · 2018-10-19 · Investigation of Channel Properties for 28 GHz Band in Urban Street Microcell Environments
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Investigation of Channel Properties for 28 GHz
Band in Urban Street Microcell Environments
Minoru Inomata, Tetsuro Imai, Koshiro Kitao, and Yukihiko Okumura NTT DOCOMO, INC. 3-6 Hikari-no-oka, Yokosuka-shi, Kanagawa, 239-8536, Japan
Abstract - In Japan, field trials on the fifth generation mobile communications system (5G) were started in order to create a new market by actualizing 5G. The 28 GHz band is the most promising candidate for commercial 5G systems. When designing the service area, site-specific channel properties in urban areas are required. Therefore, in this paper, we investigate site-specific channel properties for the 28 GHz band in an urban street microcell environments and clarify that the paths affect the channel properties.
Index Terms — 5G mobile communication system, High frequency bands, Channel properties, Street microcell.
1. Introduction
Development of the fifth generation mobile
communications system (5G) has been actively investigated
[1]-[5]. In Japan, 5G field trials were started in order to
create a new market by actualizing 5G [6]. In 5G, high
frequency bands higher than 6 GHz will be used in order to
provide attractive higher bit rates, and the 28 GHz band is
the most promising candidate for commercial 5G systems in
Japan. It is assumed that the main service areas for 5G using
high frequency bands will be in urban areas. In designing
these service areas, since the propagation characteristics is
difference for every environment, it is important to
understand site-specific channel properties for the 28 GHz
band in urban areas. Regarding the channel properties in high
frequency bands, based on the 5G system requirements,
standardized models have been reported by the 3GPP [4] and
ITU-R [5]. These models are useful in evaluating the
performance of technical specifications. However, when
designing the service area, the site-specific channel
properties is required, and since the standardized models are
generated by statistical processing the measurement data
which is acquired in various sites, site specific channel
properties are not clear in standardized models. Therefore,
we investigate site-specific channel properties for the 28
GHz band in urban street microcell environments. In a 5G
field trial we study on the channel models for path loss,
Doppler characteristics, delay profiles and angular profiles.
Therefore, in this paper, we report the path loss and Doppler
characteristics. We first report on the verification results of
the prediction accuracy based on a comparison between
measurements and prediction results using the representative
model, Rep. ITU-R M.2412 [5]. On the basis of the obtained
results, we clarify that the paths affect the channel properties.
2. Measurement Campaign
The channel properties were measured in an urban area
nearby Tokyo Station, Tokyo, Japan. The streets in the area
are surrounded by tall buildings (approximately ten stories or
40 m). In this paper, we report on the path loss and Doppler
characteristics. The path loss measurements are taken along
routes 1 and 2, as shown in Fig. 1. Route 1 includes non-line
of sight (NLOS) and line of sight (LOS) areas. Route 2 is
only a NLOS street. The Doppler measurements are recoded
along route 3. Route 3 is circular route of 2590 m. The
measurements are taken in the 28 GHz bands for a
commercial 5G system. The transmission (Tx) antennas are
set at a height of 10 m. We used them to transmit continuous
waves. A receiver (Rx) antenna is fixed on the roof of a
measurement car whose height is 2.3 m. The Tx and Rx
antenna radiation patterns are omni-directional.
Fig. 1. Measurement routes.
3. Measurement Results
Fig. 2 shows comparison results between the measurement
results on routes 1 and 2, and the predictions for the ITU-R
M.2412 channel model. We calculated the path loss using
channel model B in an urban microcell (UMi). Fig. 2 shows
that the ITU-R model predictions in the LOS environment on
route 1 are relatively accurate. A quantitative evaluation is
given where median prediction errors are calculated. The
error between the prediction results and measurement results
is approximately 2 dB. However, in the NLOS environments
on routes 1 and 2, Fig. 2 shows that the measured path loss
becomes smaller than that for the prediction results. In this
case, the difference is approximately 7.0 dB lower on route 1,
and 10 dB lower on route 2. In order to the analyze the path
which affects the path loss, on route 3 including the street on
routes 1 and 2, we analyzed the propagation path based on
Doppler characteristics. The Doppler frequency shift, Δf, is
calculated using the following equation
∆𝑓 =𝑣
𝜆cos 𝜃 = 𝑓𝐷 cos 𝜃 (1)
Tx240 m 150 m
221 m196 m
Route 1 Route 2
877 m
396 m
926 m
Route 3
①
② ③
④Rx
[WeB1-1] 2018 International Symposium on Antennas and Propagation (ISAP 2018)October 23~26, 2018 / Paradise Hotel Busan, Busan, Korea
11
where the speed of the mobile station is v (m/s), the
wavelength is λ, θ is the angle of the arrival path, and fD is
the maximum Doppler frequency. From (1), the frequency of
paths from the front of the mobile station is + fD. On the
other hand, the frequency of paths from the back of the
mobile station is - fD. Therefore, the direction of the arriving
path from the front or back of the mobile station can be
obtained based on Doppler characteristics. In particular,
when the mobile station passes through the radio wave
source including the transmitter or scattering objects, the
Doppler spectrum shifts from + fD to - fD. Thus, from the
position of the frequency shift Δf = 0, the position of the
source can be obtained. In the following, in order to clarify
the site-specific channel property, we analyze the
propagation paths in an urban area based on the Doppler
characteristics. Fig. 3 shows the measured Doppler spectrum
and the speed of the measurement car. The measurement car
passed by the base station, moved to ①, ②, ③, and ④ as
shown in Fig. 1(b), and returned to the start position. In the
section up to ①, the maximum peak of the Doppler spectrum
varies from + fD to - fD at the moving distance of
approximately 80 m. This path indicates the direct wave
from the base station. Also, around the base station, there are
many paths that vary from + fD to - fD. Since the frequency
shift is continuous, it is assumed that these paths are
scattered by the objects along the street. Furthermore, we
find that the frequency shift of the paths is greater than the
maximum Doppler frequency, fD, and the power of those
paths is greater than the noise level. It is assumed that
because these paths arrive after multiple scattering by a
moving vehicle on the street, the frequency of the paths is
Doppler shifted multiple times. In the section from ① to ②
and from ③ to ④, the peak of the Doppler spectrum is the
same as fD. Therefore, we find that the dominant path in
these sections is a path that propagates along the street from
intersection ① or ④. From these results, in Fig. 2, the reason
why measurements in the NLOS environment are lower than
the predictions is due to the building shape along the street.
In particular, it is assumed that the building shape in the
intersection significantly affects the path loss as described in
[7]. In the section from ② to ③, the peak of the Doppler
spectrum is not the same as fD. This indicates that the arrival
paths do not propagate from intersection ② or ③, and it is
assumed that the dominant path is the path over buildings or
between buildings.
Fig. 2. Comparison results of path loss.
Fig. 3. Measured Doppler spectrum on route 3.
4. Conclusion
This paper investigated site-specific channel properties for
the 28 GHz band in an urban area. We clarified that the paths
affect the channel properties based on the measured path loss
and Doppler spectrum. We find that direct wave and
scattering from fixed or moving objects affect the channel in
a LOS street, paths along the street that propagate from an
intersection affect the channel in a NLOS cross street, and
paths over buildings or between buildings affect the channel
in a parallel NLOS street. Further measurement using
channel sounder and modeling channel properties base on
these results will be future subjects.
Acknowledgment
This paper includes a part of the results from "The
research and examine the technological requirements for 5th
generation wireless systems that can realize a data
communication speed exceeding 10 Gbps in densely
populated areas" commissioned by The Ministry of Internal
Affairs and Communications, Japan.
References
[1] NTT DOCOMO, INC. “DOCOMO 5G white paper, 5G radio access:
Requirements, concept and technologies,” July 2014.