micromachines

Article
Full-Azimuth Beam Steering MIMO Antenna
Arranged in a Daisy Chain Array Structure
Kazuhiro Honda 1, *, Taiki Fukushima 2 and Koichi Ogawa 1
1 Graduate School of Engineering, Toyama University, 3190 Gofuku, Toyama 930-8555, Japan;
ogawa@eng.u-toyama.ac.jp
2 Panasonic System Networks R&D Lab. Co., Ltd., Technology Center, 2-5 Akedori, Izumi-ku, Sendai,
Miyagi 981-3206, Japan; fukushima.taiki@jp.panasonic.com
* Correspondence: hondak@eng.u-toyama.ac.jp; Tel.: +81-76-445-6759




Received: 28 August 2020; Accepted: 18 September 2020; Published: 19 September 2020


Abstract: This paper presents a multiple-input, multiple-output (MIMO) antenna system with the
ability to perform full-azimuth beam steering, and with the aim of realizing greater than 20 Gbps
vehicular communications. The MIMO antenna described in this paper comprises 64 elements
arranged in a daisy chain array structure, where 32 subarrays are formed by pairing elements in
each subarray; the antenna yields 32 independent subchannels for MIMO transmission, and covers
all communication targets regardless of their position relative to the array. Analytical results reveal
that the proposed antenna system can provide a channel capacity of more than 200 bits/s/Hz at a
signal-to-noise power ratio (SNR) of 30 dB over the whole azimuth, which is equivalent to 20 Gbps
for a bandwidth of 100 MHz. This remarkably high channel capacity is shown to be due to two
significant factors; the improved directivity created by the optimum in-phase excitation and the
low correlation between the subarrays due to the orthogonal alignment of the array with respect
to the incident waves. Over-the-air (OTA) experiments confirm the increase in channel capacity;
the proposed antenna can maintain a constant transmission rate over all azimuth angles.

Keywords: daisy chain multiple-input multiple-output (MIMO) antenna; beam steering array;
large-scale MIMO; over-the-air (OTA) testing; Monte Carlo simulation; connected car

1. Introduction
One of the most straightforward approaches for improving the capacity of multiple-input
multiple-output (MIMO) systems is to use a large number of antenna elements. The concept of
large-scale MIMO or massive MIMO systems with more than 100 antenna elements has been proposed
for both fifth and sixth generation (5G and 6G) mobile communications [1,2].
Figure 1 shows a conceptual illustration of a massive MIMO system. In massive MIMO systems,
investigations are mostly based on the development of a large-scale antenna at a base station, in which
a planar array antenna with a number of patch antennas arranged in a two-dimensional manner on
a column-row alignment basis is commonly developed [3–6], as depicted in Figure 1. The primary
objective of these antennas is, using a beam forming technique, to illuminate respective mobile
stations distributed in a specific confined region of the service area in a cell, known as a hot-spot.
Thus, a large-scale MIMO antenna used for massive MIMO systems is capable of communicating with
mobile stations in the hemispherical spatial region perpendicular to the surface of the patch array.
With regard to a mobile station antenna, on the other hand, one of the most important performance goals
in MIMO systems is the ability to communicate with the target over the full azimuth, i.e., full-azimuth
beam steering, as illustrated in Figure 1.

Micromachines 2020, 11, 871; doi:10.3390/mi11090871 www.mdpi.com/journal/micromachines

Micromachines 2020, 11, 871 2 of 25
Micromachines 2020, 11, x 2 of 25

Large-scale MIMO antenna

MIMO transmission
Patch array Mobile station

Full-azimuth beam steering

Base station

Figure
Figure 1.
1. Conceptual
Conceptual illustration
illustration of
of aa massive
massive MIMO
MIMO system.
system.

One of the possible solutions to fulfilling this requirement is the use of a circular array in which
a number of patch antennas are arranged in a cylindrical manner [7,8]. However, However, there
there is a certain
drawback to this configuration in that each patch antenna radiates in a direction normal to the surface
of the
the patch,
patch,resulting
resultinginina situation
a situation in which
in which not not
all ofallthe
ofradiation
the radiation
beamsbeams contribute
contribute to MIMO to MIMO
spatial
spatial multiplexing when an incident wave comes from a particular
multiplexing when an incident wave comes from a particular azimuthal direction. Consequently, azimuthal direction.
Consequently,
we cannot obtain wefull-rank
cannot obtain
channelfull-rank
matriceschannel matrices
corresponding tocorresponding to the number
the number of patches of patches
due to the limited
due to the limited
availability availability
of the received of thewhich
signals, received
leadssignals, which leads
to a reduction to a reduction
in channel capacity.in channel capacity.
There has been a great deal of interest in the development of connected car systems with greater
than a gigabit transmission rates [9]. ItIt is is anticipated
anticipated that that the MIMO antennas implemented implemented in
automobiles will will be
beused
usedininsystems
systemswithwitha relatively
a relatively small cell radius involving
small cell radius involving a street a street microcell
microcell [10].
[10].
In suchIn such an environment,
an environment, the the various
various objects
objects in in
thethesurrounding
surroundingarea—such
area—suchas as trees,
trees, signboards,
signboards,
buildings, and numerous vehicles—result in complex radio radio propagation
propagation phenomena.
phenomena. In In aa street
microcell environment, it is known that the spatial angular power spectrum (APS) is small compared
thatin
with that inthe
theconventional
conventional macrocell
macrocell counterpart
counterpart [11].[11].
ThisThis is because
is because the incident
the incident waves waves from
from these
these scatterers
scatterers are notareuniformly
not uniformly distributed,
distributed, but occur
but occur in clusters,
in clusters, withwith
strongstrong contributions
contributions fromfroma fewa
few directions.
directions. In addition,
In addition, the radiation
the radiation pattern
pattern of a vehicular
of a vehicular antenna antenna
changes changes
greatlygreatly
due to theduemutual
to the
mutual electromagnetic
electromagnetic couplingcoupling
betweenbetween the antenna
the antenna and the and the dynamic
dynamic motion of motion of especially
the car, the car, especially
when it
when it turns
turns right right
or left at or
an left at an intersection.
intersection. This that
This indicates indicates that the
the mutual mutualof
influence influence of the
the incident incident
waves and
waves
the car’s and the car’scould
antennas antennas
resultcould result in a complicated
in a complicated MIMO channel MIMO channel response.
response.
Based on the above-mentioned phenomena, the characteristics of a vehicular MIMO antenna are
summarized as follows:
(1) Anticipated
(1) Anticipated changes
changes inin received
received power
power when
when the
the position
position of
of the
the car
car changes
changes relative
relative to
to the
the
incident
incident waves.
waves.
(2) Increase
(2) Increase in
in spatial
spatial fading
fading correlation
correlation between
between the
the array
array branches
branches due
due to
to the
the narrow
narrow APS.
APS.
(3) Changes in both the received power and correlation, which may be encountered at the same
(3) Changes in both the received power and correlation, which may be encountered at the same time,
time, as the car moves in different directions relative to the incident waves.
as the car moves in different directions relative to the incident waves.
(4) A possible increase or decrease in the MIMO channel capacity caused by (1), (2), or (3).
(4) A possible increase or decrease in the MIMO channel capacity caused by (1), (2), or (3).
These phenomena may occur simultaneously in an actual connected car system. Hence, all the
These phenomena may occur simultaneously in an actual connected car system. Hence, all the
solutions with regard to the above issues must be integrated into one unit in the developed MIMO
solutions with regard to the above issues must be integrated into one unit in the developed MIMO
antenna. Considering this technical background, we have developed new technologies in an ongoing
antenna. Considering this technical background, we have developed new technologies in an ongoing
project with the following objectives:
project with the following objectives:
(A) Development of higher order MIMO arrays, such as 8 × 8 [12], 16 × 16 [13], and 32 × 32 [14,15]
MIMO antenna systems, suitable for implementation in an automobile.
Micromachines 2020, 11, 871 3 of 25

(A) Development of higher order MIMO arrays, such as 8 × 8 [12], 16 × 16 [13], and 32 × 32 [14,15]
MIMO antenna systems, suitable for implementation in an automobile.
(B) Development of radiation pattern steering capability to achieve a large signal-to-noise power
ratio (SNR) that can direct the peak radiation toward the communication target.
(C) Realization of low correlation between the MIMO channels established by the orthogonal
relationship between the array alignment and the incident waves over the full azimuth.
(D) Development of an angle of arrival (AOA) estimation antenna that obtains bearing information
from radio waves incident on the MIMO antenna, using an RF-based interferometric monopulse
technique with reduced hardware complexity.
To realize these objectives simultaneously, we are currently developing a 32 × 32 MIMO antenna
system that utilizes circular array beam steering technology. Great emphasis is placed on a way of
achieving a large-scale vehicle borne MIMO antenna that provides full-azimuth coverage. The new array
configuration proposed in this paper yields the full availability of the received signals, which results
in a channel capacity of several tens of gigabits owing to the effective operation of the MIMO
multiplexing transmission.
This paper, which is part of the extensive R&D activities we have performed so far in this field,
is devoted to a comprehensive description of the development work undertaken for a 32 × 32 MIMO
antenna system, including Monte Carlo simulations and over-the-air (OTA) testing to investigate how
the developed 32 × 32 MIMO array performs in a propagation environment where it is anticipated the
forthcoming 5G and 6G connected car systems will be used.
In order to tackle the four challenges—(A), (B), (C), and (D), listed above—we start by describing
the methodology and the basic characteristics of MIMO antennas. We first present a brief explanation
of the configuration of the whole antenna system. Studies on the AOA antenna are presented in
separate papers [16,17].

2. New Concept of a Large-Scale MIMO Antenna

In Figure 2 an illustrated overview of our ongoing project, conducting research and development
work toward a full-azimuth beam steering MIMO array, is shown [18]. We have taken on the
big challenge of achieving a 100 Gbps channel capacity on a moving vehicle for the forthcoming
sixth-generation (6G) mobile communications. To this end, we have devised effective means of
constructing a large-scale MIMO array antenna, with distinct features that cannot be achieved
using previous technologies commonly aimed at enhancing the channel capacity at a base station;
the objective of our project is accomplished by the emergence of a new technology named “Daisy Chain
MIMO Antenna”.
Figure 2A shows a family of daisy chain MIMO antennas, which comprises a plurality of
4 × 4 MIMO antennas, as a basic component module, arranged in a variety of two-dimensional
geometrical configurations. The term “4 × 4 MIMO” means that the number of elements at the base
station (former number) is four and that at the mobile station (latter number) is four. Therefore,
we have four subchannels for MIMO transmission in this case. This new approach enables us to
construct a high-order MIMO antenna with great ease since the same fabrication process used to
fabricate 4 × 4 MIMO antennas is used for each configuration. For example, as shown in Figure 2A,
the 32 × 32 MIMO antenna has a circular array of eight 4 × 4 MIMO antennas, resembling a daisy chain
with eight pink flowers, forming a 32 × 32 MIMO array as a whole. Hence, we have named this type of
antenna “Daisy Chain MIMO Antenna”.
Figure 2B shows the history of the development of the daisy chain MIMO antenna. The figure
illustrates the channel capacity achieved using the different size arrays, together with the abbreviated
name of the international conference where the corresponding paper was presented. As shown in the
figure, we have accomplished a twofold increase in channel capacity as the size of the array doubles.
Consequently, we have achieved a channel capacity greater than 100 Gbps for a bandwidth of 100 MHz
by means of a 128 × 128 MIMO array, as indicated in Figure 2B. Among the various configurations
 In Figure 2 an illustrated overview of our ongoing project, conducting research and development
work toward a full-azimuth beam steering MIMO array, is shown [18]. We have taken on the big
challenge of achieving a 100 Gbps channel capacity on a moving vehicle for the forthcoming sixth-
generation (6G) mobile communications. To this end, we have devised effective means of
constructing a large-scale
Micromachines 2020, 11, 871 MIMO array antenna, with distinct features that cannot be achieved using 4 of 25
previous technologies commonly aimed at enhancing the channel capacity at a base station; the
objective of our project is accomplished by the emergence of a new technology named “Daisy Chain
categorized2020,
Micromachines in the
11, xfamily in Figure 2A, this paper focuses on a detailed description of the design 4 ofand
25
MIMO Antenna”.
performance of a daisy chain 32 × 32 MIMO antenna.
Figure 2a shows a family of daisy chain MIMO antennas, which comprises a plurality of 4 × 4
MIMO antennas, as a basic component module, arranged in a variety of two-dimensional geometrical
configurations. The term “4 × 4 MIMO” means that the 1200
number of elements at the base station (former
number) is four and that at the mobile station (latter number) is four. Therefore, we have four
PIERS2019
1000

Channel Capacity [bits/s/Hz]

subchannels for MIMO transmission in this case. This new approach enables us to construct a high-
100Gbps @100MHz
order MIMO antenna with great ease since the same fabrication process used to fabricate 4 × 4 MIMO
800
antennas is used for each configuration. For example, as shown in Figure 2a, the 32 × 32 MIMO
antenna
(a) 4 × 4 has
MIMOa circular array of (c)
(b) 8 × 8 MIMO eight 4 MIMO
16 × 16 × 4 MIMO antennas, resembling a daisy chain with eight pink
EMTS2019
600
flowers, forming a 32 × 32 MIMO array as a whole. Hence, we have named this type of antenna “Daisy
Chain MIMO Antenna”. This Paper
400
Figure 2b shows the history of the development of the daisy chain MIMO EMTS2019 antenna. The figure
illustrates the channel capacity achieved using the different size APMC2018
arrays, together with the abbreviated
200
PIERS2018
name of the international conference where the corresponding paper was presented. As shown
APMC2017 in the
SNR=30dB
figure, we have accomplished a twofold increase in channel 0 capacity as the size of the array doubles.
Consequently,
(d) 32 × 32 MIMO we have (e) 64achieved
× 64 MIMO a(f)channel capacity greater 4×4
128 × 128 MIMO
8×8 16×16 32×32 64×64 128×128
than 100 Gbps for a bandwidth of 100
MHz by means of a 128 × 128 MIMO array, as indicated in Figure 2b. Among the various
configurations categorized (A)in the family in Figure 2a, this paper focuses on(B) a detailed description of
the design
Figure and
Figure 2.
performance
2. AAbig
bigchallenge
oftoward
challengetoward
a daisy100 chain
100 Gbps
32channel
× 32 MIMO
Gbps channel
antenna.
capacity.
capacity. (A)
(A)AAfamily
familyofof daisy
daisy chain
chain MIMO
MIMO
antennas.
antennas.(B)
(B)History
Historyof
ofthe
thedevelopment.
development.
3. Daisy Chain 32 ×32 Multiple-Input Multiple-Output (MIMO) Antenna
3. Daisy Chain 32 × 32 Multiple-Input Multiple-Output (MIMO) Antenna
Figure 3 shows the configuration of a beam steering MIMO antenna arranged in a daisy chain
array Figure 3 shows
structure [14]. the configuration
Figure 3a shows of thea beam
wholesteering MIMO
structure of theantenna
MIMOarranged
antennainsystem.
a daisy Figure
chain array
3b
structure
shows a 4 ×[14]. Figurecircular
4 MIMO 3a shows the[19],
array whole structure
which is usedof as
thea MIMO
componentantenna system.
module Figure
to form the3b
32shows
× 32
a 4 × 4array.
MIMO MIMO Ascircular
shown in array [19],3a,
Figure which is used as MIMO
the developed a component
antennamodule to form
comprises the 32 ×arranged
64 elements 32 MIMO
inarray. As chain
a daisy shownarrayin Figure 3a, theforming
structure, developed MIMO antenna
a dual-ring comprises
configuration; 64 ring
one elements arranged
is formed in a4daisy
by the ×4
chain array
MIMO array,structure, forming
and the other ringa is
dual-ring configuration;
constructed from eightonesetsring
of 4is×formed
4 MIMO the 4 × arranged
byarrays, 4 MIMO array,
in a
and the
circle withother ringr at
radius is constructed
equal angular from eight sets
intervals of 45°, × 4 MIMO arrays,
of 4constructing the 32arranged
× 32 MIMO in aarray
circleas
with radius
a whole.
r at equal angular intervals of 45◦ , constructing the 32 × 32 MIMO array as a whole.

Subarray 9-12
4 Subarrays
Creation of
Subarray 13-16 Subarray 5-8 #3 #2 #2 + #3 Beam
E0 E0 e  jkd1 Beam1
MIMO Array d1
Subarray1 (1) Large SNR
#4  jkd 2
#1 #1 + #4
AOA Antenna E0 d2 E0 e Cluster
Subarray 17-20 Subarray 1-4 Beam2 Incident Wave
Subarray2 y
Subarray2 
Subarray18
#9 z x 2
#5 + #8
 jkd 2
r E0 E0 e
Beam3 Orthogonal to
Subarray 21-24 Subarray 29-32 Subarray3 d2 a
MIMO Array
#5 #8
Subarray4 #6 + #7
Subarray 25-28 d1
E0 E0 e  jkd1 Beam4
(2) Low
4×4 MIMO #6 #7 Correlation
System

(a) (b)
Figure
Figure3.3.Beam
Beamsteering
steeringMIMO
MIMOantenna
antennaarranged
arrangedinina adaisy
daisychain
chainarray
arraystructure.
structure.(a)
(a)The
Thewhole
whole
structure
structureofofa a3232××3232MIMO
MIMOsystem.
system.(b) 4 4× ×
(b) 4 MIMO
4 MIMO system.
system.

32 subarrays are created by pairing elements in each 4 × 4 MIMO array; the antenna yields 32
independent subchannels for MIMO transmission, and covers all communication targets regardless
of their position relative to the array, as mentioned below. As illustrated in Figure 3b, four pairs of
the eight elements form four subarrays (Elements #2–3, #1–4, #5–8, and #6–7). All the elements are
half-wavelength dipole antennas. These subarrays create four independent radiation patterns
Micromachines 2020, 11, 871 5 of 25

32 subarrays are created by pairing elements in each 4 × 4 MIMO array; the antenna yields
32 independent subchannels for MIMO transmission, and covers all communication targets regardless
Micromachines 2020, 11,relative
of their position x to the array, as mentioned below. As illustrated in Figure 3b, four 5pairs of 25
of the eight elements form four subarrays (Elements #2–3, #1–4, #5–8, and #6–7). All the elements
(Beam1–Beam4),
are half-wavelength eachdipole
of which acts as a These
antennas. subchannel
subarraysfor the MIMO
create fourarray by enabling
independent excitation
radiation of the
patterns
four subarrays which achieves an in-phase state toward the target
(Beam1–Beam4), each of which acts as a subchannel for the MIMO array by enabling excitation of direction of communication. Asthea
result,
four subarrays which achieves an in-phase state toward the target direction of communication. Asaa
the peak gain of the beam is larger than that of an ordinary dipole antenna, which results in
large SNR.
result, the peak gain of the beam is larger than that of an ordinary dipole antenna, which results in a
In Figure 3b, k represents the wave number, d1 and d2 denote the distance between Elements #2
large SNR.
and #3, and Elements #1 and #4,
3b, k represents therespectively. E0 dsignifies the amplitude of the electric field, and j
In Figure wave number, 1 and d2 denote the distance between Elements #2
indicates
and #3, and the Elements
complex unit. #1 and Using these parameters,
#4, respectively. the excitation
E0 signifies conditions
the amplitude for the
of the fourfield,
electric subarrays
and j
to establish an in-phase state for the formation of a beam toward
indicates the complex unit. Using these parameters, the excitation conditions for the four subarrays the communication target are
to
described
establish an in in-phase
Section 5.state for the formation of a beam toward the communication target are described
Another
in Section 5. unique feature of the 4 × 4 MIMO antenna is the realization of low correlation. When
an incident
Another waveuniquewithfeature
a narrow APS
of the 4 ×arrives
4 MIMO from the right
antenna is theside of the figure,
realization of low the four subarrays
correlation. Whenare an
arranged perpendicular to the incident wave. In general, in a cluster
incident wave with a narrow APS arrives from the right side of the figure, the four subarrays are propagation environment, a
MIMO array parallel to an incident wave results in high correlation
arranged perpendicular to the incident wave. In general, in a cluster propagation environment, between the branches [20]. In
contrast,
a MIMOaarray MIMO array orthogonal
parallel to an incident to anwave
incident wave
results inyields low correlation.
high correlation betweenHence, thethe orthogonal
branches [20].
arrangement between the subarray and the cluster incident wave, as shown
In contrast, a MIMO array orthogonal to an incident wave yields low correlation. Hence, the orthogonal in Figure 3b, results in a
smaller correlation coefficient, which is beneficial for the enhancement
arrangement between the subarray and the cluster incident wave, as shown in Figure 3b, results in a of the MIMO channel capacity,
together with the creation
smaller correlation of a large
coefficient, whichSNR due to the
is beneficial forfour subarrays mentioned
the enhancement of the MIMOabove.channel capacity,
together with the creation of a large SNR due to the four subarrays mentioned above. the subarrays
In Figure 3b, as the eight elements form a circular arrangement with 45° intervals,
can be Inrotated
Figure 3b, every 45°eight
as the in theelements
azimuthform plane. Figure arrangement
a circular 4 shows the four with combinations
45◦ intervals, the of subarrays
subarrays
according to the angle of
◦ the incident wave [19]. The yellow zone
can be rotated every 45 in the azimuth plane. Figure 4 shows the four combinations of subarrays shows the angular range of the
incident wave applicable to MIMO communications. Therefore,
according to the angle of the incident wave [19]. The yellow zone shows the angular range of thewhen an automobile turns right or
left at an intersection, which results in a situation where an incident
incident wave applicable to MIMO communications. Therefore, when an automobile turns right wave arrives from other azimuth
angles,
or left atweancan select other
intersection, appropriate
which results in combinations
a situation where from among the possible
an incident wave arrivescombinations
from other of
subarrays. This function can also be applied effectively to the 32
azimuth angles, we can select other appropriate combinations from among the possible combinations × 32 MIMO beam steering array
antenna, by performing
of subarrays. This function synchronized
can also beswitching of all the to
applied effectively 32the
subarray
32 × 32beams.
MIMOThis beamunique
steering feature
array
allows the SNR of the 32 subarray beams to be large, and simultaneously,
antenna, by performing synchronized switching of all the 32 subarray beams. This unique feature the correlation to be small,
which
allows keeps
the SNR theofchannel capacitybeams
the 32 subarray in a high-bit
to be large, rateand condition, even with
simultaneously, the considerable
correlation todynamic
be small,
motion of the car.
which keeps the channel capacity in a high-bit rate condition, even with considerable dynamic motion
of the car.

y y 67.5deg y 112.5deg y
112.5deg 67.5deg
#3 #3 #3 #3
#2 #2 #2 #2
157.5deg 22.5deg #4
#4 #4 #4
22.5deg 157.5deg
#1 #1 #1 #1
x x x x
#5 #5 #5 #5
202.5deg 337.5deg
#8 337.5deg
#8 #8 #8
202.5deg
#6 #7 #6 #6 #7 #6 #7
#7
247.5deg 247.5deg 292.5deg 292.5deg

(a) (b) (c) (d)

Figure
Figure4.4.Combinations
Combinationsofofthe subarrays.
the (a)(a)
subarrays. Combination1. (b) (b)
Combination1. Combination2. (c) Combination3.
Combination2. (d)
(c) Combination3.
Combination4.
(d) Combination4.

Figure55shows
Figure shows a schematic
a schematic diagram
diagram of aof a beam
beam forming
forming network
network for full-azimuth
for full-azimuth steering,steering,
which
which comprises
comprises switches,
switches, phase phase shifters,
shifters, and and
powerpower dividers
dividers [15].[15]. Using
Using thisnetwork,
this network,the the four
four
combinations of subarrays, as shown in Figure 4, can be chosen according to the angle of
combinations of subarrays, as shown in Figure 4, can be chosen according to the angle of the incident the incident
wave to
wave to deliver
deliver signals
signals to
to Elements
Elements #1–8
#1–8 for
for achieving
achieving an
an in-phase
in-phase state
state between a pair of elements.
 Micromachines
Micromachines 11,11,
2020,
2020, x 871 6 of6 25
of 25
Micromachines 2020, 11, x 6 of 25
#1 #2 #3 #8
#1 #2 #3 #8
Switch
Switch

1 2 3 4 Phase shifter
1 2 3 Power divider
4 Phase shifter
ch1 ch2 ch3 ch4 Power divider
ch1 ch2 ch3 ch4
Figure 5. Beam forming network for full-azimuth steering.
Figure 5.
Figure Beamforming
5. Beam formingnetwork
networkfor
forfull-azimuth
full-azimuth steering.
steering.

4. 4.
Formulation
Formulation of of
Monte
Monte Carlo
CarloSimulation
Simulation
4. Formulation of Monte Carlo Simulation
This
Thissection
sectionisisdevoted
devoted toto the
theformulation
formulationofof a Monte
a Monte Carlo Carlo simulation
simulation usedused to analyze
to analyze the
the channel
channel This section of
responses is the
devoted
daisy to theMIMO
chain formulation of As
antenna. a Monte
shown Carlo
in simulation
Figure 6a, the used
conceptto of
analyze the
clusters
responses of the daisy chain MIMO antenna. As shown in Figure 6a, the concept of clusters [21]
channel
[21] responses
is employed of the daisy chain MIMO antenna. As shown in Figure 6a, the concept ofstation
clusters
is employed to to simulate
simulate a set
a set of discrete
of discrete waves
waves with
with narrow
narrow APSAPS arriving
arriving at aatmobile
a mobile
station (MS).
[21]
(MS). is employed
In Figure to simulate
6, bell-shaped
the bell-shaped a set of discrete waves with narrow APS arriving at a mobile station
In Figure 6, the blueblue curves
curves represent
represent a cluster
a cluster incident
incident wave.wave. We have
We have developed
developed Monte
(MS). Carlo
Monte In Figure 6, the bell-shaped
simulation software blue curves
using MATLAB,represent a cluster incident wave.model
We have developed
Carlo simulation software using MATLAB, in whichina which
cluster achannel
cluster channel
model suitable forsuitable for a
simulating
Monte
simulating Carlo
a simulation
large-scale software
MIMO antennausing
at MATLAB,
the mobile in
side which
is a cluster
formulated. channel
The modelsoftware
developed suitable is
for
large-scale MIMO antenna at the mobile side is formulated. The developed software is an improved
simulating
anversion
improved a large-scale
version MIMO
of the antenna
channelused at
modelingthe mobile side
used for handsetis formulated.
simulating The
handset developed
adaptive software
and MIMO is
of the channel modeling for simulating adaptive and MIMO arrays with a
an improved
arrays versionazimuthal
withazimuthal
a uniform of the channel modeling used for simulating handset adaptive and MIMO
uniform APS [22,23].APS [22,23].
arrays with a uniform azimuthal APS [22,23].
m=1 M q z
BS m=1 M q z
BS Cluster q=1
Cluster q=1
Mobile q = p /2
MS
Mobile
Station q = p /2
MS
n=1 N 
Station
v
y
n=1 N k=1 v  q
y
k , mq
k=1 k , m k-th scatterer
Qc fork-th
q-thscatterer
cluster
Qc Km Incident wave for q-th cluster
k Km Moving directionIncident wave
k x
scatterer Moving direction x Moving direction
scatterer Moving direction
(a) (b)
(a) (b)
Figure 6. Channel
Figure model
6. Channel used
model forfor
used performing
performing thethe
Monte
MonteCarlo simulation.
Carlo (a)(a)
simulation. Cluster channel
Cluster model
channel model
ofFigure
MM
of × N× 6.
NChannel
MIMO.
MIMO. model
(b)(b) used for
Coordinates performing
of the
Coordinates k-th
of the the Monte Carlo simulation. (a) Cluster channel model
scatterer.
k-th scatterer.
of M × N MIMO. (b) Coordinates of the k-th scatterer.
InIn ordertotosimulate
order simulateaa small cell cell scenario,
scenario,eacheachcluster
cluster is is
assumed
assumed to have a large
to have number
a large numberof arrival
of
In order
paths,paths,
arrival ensuring to simulate
the full-rank
ensuring a small cell scenario,
status ofstatus
the full-rank a channel each
of amatrix
channelcluster
created is
matrix assumed
by created to have
a large-scale
by aMIMO a large number
antenna.
large-scale MIMO of
Here,
arrival
the term
antenna. paths,
‘path’
Here, ensuring
is term
the the
used ‘path’ full-rank
to represent
is used status
independent of a channel
to representsubchannels
independent matrix created
forsubchannels by
MIMO transmission a
for MIMOlarge-scale
which MIMO
are created
transmission
antenna.
by Here,
reflecting or the term
diffraction‘path’
pointsis used
from to represent
surrounding independent
objects in
which are created by reflecting or diffraction points from surrounding objects in a cell, sucha subchannels
cell, such as for MIMO
buildings, transmission
trees, and ascars.
which
buildings, are
Hence, intrees, created
an actual by reflecting
cellular
and cars. Hence, or
system, diffraction
in anthe number
actual points from
of paths
cellular system, surrounding
may thechange
number objects
greatly, in a cell,
depending
of paths such
may change as
on the
buildings,
propagation
greatly, trees,
depending and
environment cars. Hence, in an
arising fromenvironment
on the propagation actual cellular
congested urban arising system,
or non-crowdedthe
from congested number
rural of paths
areas.
urban may
orConsidering
non-crowded change
this
greatly,
fact, the depending
prime on
objective the
of propagation
the channel environment
model used inarising
this from
paper is
rural areas. Considering this fact, the prime objective of the channel model used in this paper is tocongested
to assess urban
the or non-crowded
theoretical limitation
rural
assess areas.
of thethechannel Considering
capacity
theoretical this fact,
achieved
limitation the
by
of the theprime
daisyobjective
channel chain MIMO
capacity of the
achieved channel
antenna model
when
by the used
chainin
a sufficiently
daisy this
MIMO paper
large is to
number
antenna
assess
of paths
when thearetheoretical
available
a sufficiently largelimitation
innumber
a cell whereof paths
of thea channel
relevant capacity
antenna
are available in achieved
operates.
a cell where by the daisy chain
a relevant MIMO
antenna antenna
operates.
whenMM a sufficiently
base
base station
station large
(BS)
(BS) number
antennas
antennas of paths
create
create are
a set
a set ofofavailable
QcQclusters, in
c clusters,
a cell
each
each where
of of a
which
which relevant antenna
comprises
comprises MM operates.
uncorrelated
uncorrelated
waves,
waves, Mforming
base
formingstation
KmKm (BS) antennas
scatterers
scatterers create a set
surrounding
surrounding NN of
MS Qcantennas.
MS clusters,
antennas. each of the
Thus,
Thus, whichMM
the comprises M uncorrelated
uncorrelated
uncorrelated waves
waves areare
waves,
subject forming
to an K m scatterers
independent and surrounding
identically N MS
distributed antennas.
(i.i.d) Thus,
complex
subject to an independent and identically distributed (i.i.d) complex Gaussian process. Furthermore, the M
Gaussianuncorrelated
process. waves
Furthermore, are
subject
thethe to
correlation an
correlation independent and
characteristicsofofthe
characteristics identically
theBSBSand
andMSdistributed
MSsides
sidesare (i.i.d)
aretaken complex
taken to to be Gaussian process.
be independent of each Furthermore,
each other,
other,based
basedon
onthe
thecorrelation
Kronecker characteristics
assumption [24]. of theMS BS and MS sides
antennas are are taken to
assumed tobebeindependent
surroundedofbyeach Km other, based
scatterers,
on the Kronecker assumption [24]. MS antennas are assumed to be surrounded by Km scatterers,
Micromachines
Micromachines2020,
2020,11,
11,x871 77of
of25
25

created by Qc clusters. The Km scatterers belonging to the q-th cluster are Gaussian distributed in
the Kronecker
azimuth assumption [24].
in two-dimensional MS antennas
coordinates. are assumed
Figure to bethe
6b depicts surrounded
coordinatesby K scatterers,
ofmthe created in
k-th scatterer, by
Qc clusters.
which the MS is Kmoving
The m scatterers belonging
toward the to
azimuththe q-th cluster
direction, ϕ vare
. Gaussian
Using the distributed
above-mentionedin azimuth
channelin
modeling, we carried
two-dimensional out a Monte
coordinates. Carlo
Figure simulation
6b depicts according to
the coordinates ofthe k-th scatterer,
thefollowing procedure:
in which the MS is
moving toward the azimuth direction, φv . Using the above-mentioned channel modeling, we carried
4.1.
out Step 1: Generation
a Monte of Km-Pathaccording to the following procedure:
Carlo simulation
The multipath property is modeled by the angular power spectrum. The spectrum of the
4.1. Step 1: Generation of Km -Path
Gaussian distribution for the q-th cluster is shown in Figure 7a, and given below.
The multipath property is modeled by the angular power spectrum. 2The spectrum of the Gaussian
 p  
 given
distribution for the q-th cluster is shown in Figure 7a, and  
below.
   q 
p XPR    2  o 
PHq ( ,  ) = AHq exp  2 (1)
 π2
q 

22q
n
2π 1  XPR
XPR φ − − µ 
  

PHθ ( , φ) = AHθ exp

  (1)
2 1 + XPR   2σq 2


p 11 pπ
PPHφ ( π ,, 
H ( φ)) == q (( ,
PPHHθ , φ) ) (2)(2)
2 XPR
XPR 22
where µ
where µqq isis the
the azimuth
azimuthangleangleof of the
the mean
mean of of the
the incident wave, σσqq isis the
incident wave, the standard
standard deviation
deviationof of the
the
powerspectrum,
power spectrum,and andXPRXPRisisthe
thecross-polarization
cross-polarizationpower powerratio,
ratio,given
givenas asthe
thepropagation
propagationparameter,
parameter,
and isis assumed
and assumed to to be
be the
the average
average value
value [25].
[25]. TheThe channel
channelresponses
responsesof of the
the signals
signals for
for aa particular
particular
snapshot of
snapshot of fading
fading areare generated
generated usingusing random
random numbers,
numbers, which
which ensures
ensures orthogonality
orthogonality between
between
subchannels for
subchannels for the
the purpose
purpose of of ideal
ideal MIMO
MIMO transmission
transmission withwith aa full-rank
full-rank channel
channel matrix.

hVq k ,nm
Average Power
hHq k ,nm

n-th antenna Eq n
2q E n

output
0 q Azimuth Angle 180deg
hkq,nm
(a) (b)
Figure
Figure7.7.Incident
Incidentwave
wavemodel
modelfor
forthe
theMonte
MonteCarlo
Carlosimulation.
simulation.(a)
(a)Gaussian
Gaussianincident
incidentwave
waveininazimuth.
azimuth.
(b) Two polarization components.
(b) Two polarization components.

4.2.Step
4.2. Step2:2:Generation
GenerationofofPolarization
Polarizationand
andSummation
SummationofofAll
AllClusters
Clusters

ItIt isis assumed

assumed thatthat the k-th path
the k-th path for
for the q-th cluster
the q-th cluster has
has aa transfer
transfer function
function with
with vertical
vertical and
and
horizontal components, as shown in Figure 7b. The vertical and horizontal components
horizontal components, as shown in Figure 7b. The vertical and horizontal components of the transfer of the transfer
functionof
function of the k-th path
the k-th path for
for the q-th cluster
the q-th cluster at
at the n-th antenna
the n-th antenna are
are given
given by
by

p
XPR
 
r
q q
= hkh,mk,mEqEn θ( n (,π ,qkφ,m)exp
XPR q q

q
q q 
hVk , nm=
h Vk,nm ) j
exp jϕ (3)(3)
11+XPR
Vk , mVk,m
XPR 2 2 k,m
11 qhq E p( π ,qφ q ) exp qjϕ q 
r

 
q q
h Hk,nm
h = = hk ,mk,mE nφ( n ,2 k ,m)exp
k,m j Hk ,mHk,m (4)(4)
11+XPR
XPR
Hk , nm
2
where Eθn (π/2, φqq k,m ) and Eφn (π/2, φq ) are the complex electric field directivities of the n-th antenna
where Eθn (π/2, ϕ k,m) and Eϕn (π/2, ϕqk,mk,m
) are the complex electric field directivities of the n-th antenna
element in the xy-plane for the θ and φ components, respectively, which are defined at the origin
element in the xy-plane for the θ and ϕ components, respectively, which are defined at the origin of
of the coordinates. hq k,m represents the equivalent amplitude of the incident waves and can be set
the coordinates. hqk,m represents the equivalent amplitude of the incident waves and can be set to an
to an arbitrary value; thus, it is assumed to have unity amplitude. XPR is the cross-polarization
arbitrary value; thus, it is assumed to have unity amplitude. XPR is the cross-polarization power
Micromachines 2020, 11, 871 8 of 25

power ratio. The phases of the vertical and horizontal polarization components, ϕq Vk,m and ϕq Hk,m ,
are independent of each other, and are uniformly distributed from 0 to 2π. For each path, the two
polarization components are combined with reference to the schematic diagram shown in Figure 7b,
as the complex sum of the vertical and horizontal components, and we have
q q q
h k,nm = h Vk,nm + h Hk,nm (5)

where the transfer function of the k-th path is evaluated by summing all the cluster responses, and is
given by
XQc
q
hk,nm = hk,nm (6)
q=1

4.3. Step 3: Generation of the Resultant Channel Response

Using the transfer function represented by Equation (6), the resultant channel response at the n-th
antenna is calculated using the equation

Km ( )
X 2πd
hnm = hk,nm exp j cos(φ k,m − φ v ) (7)
λ
k =1

where λ is the wavelength in free space, and d is the distance traveled by a mobile station moving
toward the azimuth direction φv , as shown in Figure 6b. This scheme is applied repeatedly to generate
the following channel response matrix at the s-th snapshot
 
 h11 h12 ··· h1m 
 h21 h22 ··· h2m
 

Hs = [hnm ] =  . (8)
 
 .. .. .. .. 
 . . . 

hn1 hn2 ··· hNM
 

The distance traveled by the mobile station between two successive points, ∆d, in the simulation
is changed according to the angular spread of the incident waves, because a narrow spread results in a
long fading period. For each value of d, fading is assumed to be quasi-static, i.e., the channel response
is kept constant.
The complex correlation coefficient between two channel responses is defined as

hh1 ∗ h2 i
ρc = √ √ (9)
hh1 ∗ h1 i hh2 ∗ h2 i

where h1 and h2 represent the two distinct channel responses in the matrix of Equation (8), and <X>
denotes the ensemble average of X where Y* represents the complex conjugate of Y. Using this
definition, the absolute value of the complex correlation coefficient can be obtained from
 
ρa = ρc  (10)

Here, the absolute value of the complex correlation coefficient ρa in Equation (10) is used as a
general description of the correlation behavior of the MIMO antenna performance throughout this
paper, as mentioned in Section 5.

4.4. Step 4: Evaluation of the Channel Capacity

The eigenvalues and eigenvectors are obtained using the following SVD (singular value
decomposition) operation
Hs = Us Ds VH
s (11)
Micromachines 2020, 11, 871 9 of 25

Us = [er1 , · · · , erL ] (12)

Vs = [et1 , · · · , etL ] (13)

hp p i
Ds = diag λ1 , · · · , λL (14)

where L = min (N, M). Us and Vs are the singular vectors of the receiving (mobile station) and
transmitting (base station) antennas, respectively. Ds represents the singular values, where λi
denotes the eigenvalue of the i-th subchannel for spatial multiplexing transmission in the MIMO
system. Using these eigenvalues, the instantaneous channel capacity and its average value are finally
evaluated by
L
γ λi
X !
Cs = log2 1 + (15)
M
i=1
S
1X
C= Cs (s = 1 · · · S) (16)
S
s=1

where γ is the input SNR (signal-to-noise power ratio), defined as the SNR for each incident wave
when an isotropic antenna is used for receiving the incident wave, permitting the performance of the
antenna elements used in the MIMO array to be included in the simulation results.

5. Theoretical Investigation

5.1. Design of 4 × 4 MIMO Antenna

As described in Section 3, the 4 × 4 MIMO antenna depicted in Figure 3b is used as a basic
component module or a fundamental functional unit cell to construct the 32 × 32 MIMO antenna array.
Hence, in the first step of our investigation, we considered the design of the 4 × 4 MIMO antenna [19].
Two of the eight antenna elements arranged in the circle form a subarray that receives a subchannel
of the MIMO communication, where the number of subarrays in the 4 × 4 MIMO antenna is four.
The radius of the circle, a, was set to 4.9 cm so that the distance between adjacent elements (d1 shown
in Figure 3b) was a quarter-wavelength at 2 GHz. In this case, the directivity of Subarray1 is expected
to have a cardioid radiation pattern in the absence of mutual electromagnetic coupling between
antenna elements.
Figure 8 shows the fundamental principle of the cardioid radiation pattern created by two isotropic
point sources corresponding to Elements #2 and #3 in Figure 3b, in which there is no electromagnetic
mutual coupling between the two sources. As shown in Figure 8a, the distance between the two
sources is set to d1 = λ/4. Furthermore, the received signal from Element #2 is delayed by φp = π/2
using a phase shifter. The total received signal, Et , is given by summing the received signals from
Elements #2 and #3 using a signal combiner, and calculated from the equation

π
  
Et = Ea 1 + exp j (cos φ − 1) (17)
2

where φ represents the azimuthal angle defined in Figure 8. Ea denotes the amplitude of excitation of
each element. Using this circuit topology shown in Figure 8a, an actual signal processing unit was
fabricated, as described in Section 6 (see Figure 18a).
Figure 8b illustrates the radiation pattern normalized to its peak value, calculated by the absolute
value of Equation (17). As can be seen from Figure 8b, the radiation pattern directs its maximum
towards φ = 0◦ and exhibits a deep null towards φ = 180◦ . Therefore, the array works as a phased array
antenna that yields a strong radiation intensity in the direction of the line between the two sources.
Micromachines
Micromachines 2020,
2020, 11,
11, x871 1010of
of 25
25

1
#3 φ #2
0.8

Ea λ Ea e − jπ /2 0.6
d1 =
4 0.4
π
Phase shifter φp = 0.2
φ
2 φ = 180° φ = 0°
0

-0.2

-0.4
+ Signal combiner
-0.6

-0.8

Et -1
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

(a) (b)
Figure 8. Principle
Figure 8. Principleofof
thethe cardioid
cardioid radiation
radiation patternpattern created
created by by twopoint
two isotropic isotropic point
sources. sources. (a)
(a) Configuration
Configuration
of the array. (b)ofCardioid
the array. (b) Cardioid
radiation pattern.radiation pattern.

When thesubarrays
When the subarraysdescribed
described in Figure
in Figure 3binvestigated,
3b are are investigated, the electromagnetic
the electromagnetic mutual
mutual coupling
coupling
between the dipole antenna elements must be considered. The signals received by two elementstwo
between the dipole antenna elements must be considered. The signals received by are
elements
multipliedarebymultiplied
the weightbyfunctions
the weightandfunctions and phase
phase control between control between
the two the two
elements. elements. In
In Subarray1 in
Subarray1 in Figure
Figure 3b, the weight3b, the weight
functions functions
of each elementof(weach
2 and element
w 3 ) and (w
the2 and
phasew ) and
shift
3 the
value phase
(τ 2 ) areshift value
optimized
(τ
by2)Equation
are optimized by Equation
(18) using the two (18) using
electric thedirectivities
field two electric(Efield
2 anddirectivities (E2 and
E3 ) of the angle of E 3) of
the the angle
incident of
wave
the
(φ) incident
when thewave (ϕ) when
antenna the are
elements antenna elements
excited are excited individually.
individually.
ww2 EE2 ((φφ))ee−jτ
− jτ 2
+ww3EE3(φ(φ) )
EEs1s1((φφ)) ==
2 +
2 2 3 3
(18)
(18)
60 ( + PP3 )
60(PP22 + )
p

where P22 andand PP33are arethe

theinput
inputpower
powerat at the
the feed
feed point
point ofof Elements
Elements #2 #2 and
and #3,
#3, respectively.
respectively. The The phase
phase
of
of one
one element
element shiftsshifts to
to an
an in-phase
in-phase state
state with
with respect
respect to to the
the other
other element
element at at the
the angle
angle of of the
the incident
incident
wave in consideration of the mutual electromagnetic coupling. The The weight
weight functions
functions are are determined
determined
in
in such
such aa way
way thatthat the
the gain
gain in the communication
communication direction direction of the combined radiation radiation pattern
pattern yields
yields
the
the maximum
maximum value. value. In InSubarray2,
Subarray2,the thesame
sameprocedure
procedureisisapplied
appliedto tooptimize
optimizethe theradiation
radiationpattern.
pattern.
In
In Section
Section 6, 6, we
we will
will introduce
introduce aa microwave
microwave circuitcircuit for
for realizing
realizing weight
weight functions
functions in in an
an experimental
experimental
way (see Figure 18a).
Since
Since thethe maximum
maximum radiation
radiation gain
gain of of each
each element
element is is approximately
approximately 00 dBd, dBd, inin order
order toto further
further
enhance
enhance the the radiation
radiation gain,
gain, we
we examined
examined placing
placing aa parasitic
parasitic oror unexcited
unexcited element,
element, Element
Element #9 #9 shown
shown
in
in Figure
Figure 3b 3b at the center of the circle. Figure Figure 99 shows
shows the the radiation
radiation gaingain at 0◦ as
at 0° as aa function
function of of the
the
length
length of the parasitic element [19]. The parasitic Element #9 works as a reflector
parasitic element [19]. The parasitic Element #9 works as a reflector or director, similar to or director, similar
to a Yagi–Uda
a Yagi–Uda array array antenna,
antenna, when
when its length
its length is long
is long or short,
or short, respectively.
respectively. The blue
The blue and red andsolid
red solid
lines
lines indicate
indicate the radiation
the radiation 0◦ ofatElements
gain atgain 0° of Elements
#2 and #1, #2 respectively,
and #1, respectively,
which arewhichthe most areinfluential
the most
influential
elements on elements on the directivities
the directivities of Subarray1 of Subarray1 and Subarray2,
and Subarray2, when thewhen the incident
incident wave comes wavefromcomes0◦ .
from 0°. The
The black lineblack
shows line
theshows thevalue
average average value
of the of the gain
radiation radiation gain of#1
of Elements Elements ◦
and #2 at#10 and. The#2star
at 0°. The
marks
star
plotted on the right axis represent the radiation gains of each element (#1, #2, #3, and #4) when #4)
marks plotted on the right axis represent the radiation gains of each element (#1, #2, #3, and the
when theelement
parasitic parasitic#9element
was not#9inwas notFigure
place. in place. Figurethat
9 shows 9 shows
whenthat when the
the length length
of the of theelement
parasitic parasiticis
element is shorter
shorter than a half than
wavelength (0.5 λ = 7.5 cm),
a half wavelength (0.5 λthe= radiation
7.5 cm), the radiation
gains gains#1
of Elements of and
Elements
#2 are #1 and #2
enhanced.
are
Whenenhanced.
the length When the
of the length of
parasitic the parasitic
element is set toelement is setλ),
6.2 cm (0.41 tothe
6.2 average
cm (0.41radiation
λ), the average
gain of radiation
Elements
gain
#1 andof #2
Elements ◦
at 0 is #1 andmaximum
at its #2 at 0° isvalue.
at its maximum value.
Micromachines 2020, 11, 871 11 of 25
Micromachines 2020, 11, x 11 of 25

5
#1 (main element of Subarray2)
2.5
0
#2 (main element of Subarray1)
Gain g0 [dBd]
Gain g0 [dBd]
-2.5 Ave. (#1, #2)
#3
-5
-7.5
-10 Max
#4
-12.5
without parasitic element
-15
5.5 6 6.5 7 7.5 8 8.5 9
Length of Parasitic Element [cm]
Figure9.9.Radiation
Figure Radiationgain
gainofofeach
eachelement.
element.

The
Theexcitation
excitationconditions
conditionsfor forSubarrays
Subarrays1–41–4are
aresummarized
summarizedin inTable
Table1,1,which
whichisisobtained
obtainedby by
searching
searchingfor forthe
theoptimum
optimumamplitude
amplitudeand andphase
phasefrom
fromamong
amongthe thepossible
possibleweight
weightfunctions
functionsofofeach each
element
elementusingusingEquation
Equation (18).
(18).Furthermore,
Furthermore, Element
Element #9 #9
is used
is usedas as
a parasitic
a parasiticor non-exciting
or non-exciting element to
element
enhance
to enhancethe directivity.
the directivity.FromFrom
the analytical resultsresults
the analytical given in Figure
given in 9, the length
Figure 9, theoflength
the parasitic
of the element
parasitic
iselement
set to 6.2iscm to 6.2λ).
set (0.41 cmAs shown
(0.41 in shown
λ). As Figure 4,
inthe subarray
Figure 4, thecombinations are the same
subarray combinations arewhen the angle
the same when
ofthe
theangle
incident wave ◦
is 0 and ◦
180is. 0°
However, theHowever,
optimizedthe weight function and phase shift and
values of
of the incident wave and 180°. optimized weight function phase
each
shiftelement
values of φ = 0element
at each ◦ are different from those at φ = 180◦ . In Combination1, since the subarray has a
at ϕ = 0° are different from those at ϕ = 180°. In Combination1, since the
symmetrical
subarray has configuration
a symmetrical with configuration y-axis,
respect to thewith the optimum
respect valuethe
to the y-axis, of element
optimum at φ =of
#2value 0◦element
is that
of#2#3atatϕ φ = 180
= 0° ◦
is that, whereas
of #3 at ϕthe optimum
= 180°, value
whereas #3 at φ =value
theofoptimum ◦
0 is that
of #3ofat#2
ϕ at φ is
= 0° = that ◦
180 of. Thus,
#2 at ϕ we can
= 180°.
select
Thus,other
we appropriate
can select other combinations fromcombinations
appropriate among the possible combinations
from among of subarrays.
the possible combinations of
subarrays.
Table 1. Excitation conditions of each element.
Table 1. Excitation conditions of each element
Frontward for Incident Wave Backward for Incident Wave
Subarray1 #2 1 V, −160◦
Frontward for Backward for0◦
#3 0.72 V,
Subarray2 #1 Incident
1 V, −330◦ Wave Incident V, 0◦
Wave
#4 0.38
Subarray3 Subarray1 #8 1 ◦V, −160°
1 V,#2−330 #3 #50.72 V,V,0°0◦
0.38
Subarray4 Subarray2 #7 1 V,#1−160 ◦ ◦
1 V, −330° #4 #60.38
0.72
V,V,0°0
Subarray3 #8 1 V, −330° #5 0.38 V, 0°
Figure 10 shows the θ-polarized radiation patterns in the xy-plane of Subarray2 as a function
Subarray4 #7 1 V, −160° #6 0.72 V, 0°
of the angle of the incident wave [19]. The angle of the incident wave is changed from 0◦ to 135◦ in
45◦ intervals. The black arrows in the figure indicate the angle of the incident wave. The radius of
Figure 10 shows the θ-polarized radiation patterns in the xy-plane of Subarray2 as a function of
the circle, a, is set to 4.9 cm. The frequency used in the analysis is 2 GHz. As shown in Figure 10,
the angle of the incident wave [19]. The angle of the incident wave is changed from 0° to 135° in 45°
the radiation patterns vary depending on the angle of the incident wave, which is in accordance with
intervals. The black arrows in the figure indicate the angle of the incident wave. The radius of the
the subarray combination, as illustrated in Figure 4. Furthermore, a radiation gain of 4.1 dBd at the
circle, a, is set to 4.9 cm. The frequency used in the analysis is 2 GHz. As shown in Figure 10, the
angle of the incident wave is achieved and is larger than an ordinary dipole antenna, which results
radiation patterns vary depending on the angle of the incident wave, which is in accordance with the
in a large SNR. The directivities of the other three subarrays also control the direction of the beam.
subarray combination, as illustrated in Figure 4. Furthermore, a radiation gain of 4.1 dBd at the angle
Therefore, high signal levels can be maintained even though the direction of the incoming wave varies
of the incident wave is achieved and is larger than an ordinary dipole antenna, which results in a
due to the dynamic motion of the automobile in connected car applications.
large SNR. The directivities of the other three subarrays also control the direction of the beam.
Therefore, high signal levels can be maintained even though the direction of the incoming wave
varies due to the dynamic motion of the automobile in connected car applications.
 Micromachines2020,
Micromachines 11,x871
2020,11, 1212ofof2525

XY Plane XY Plane XY Plane XY Plane

10 10 10 10
y y y y
0 0 0 0

-10 -10 -10 -10

-20 -20 -20 -20

x x x x
-10 -10 -10 -10

0 0 0 0
Eθ Eθ Eθ Eθ
10 10 10 10
10 0 -10 -20 -10 0 10 10 0 -10 -20 -10 0 10 10 0 -10 -20 -10 0 10 10 0 -10 -20 -10 0 10
[dBd] [dBd] [dBd] [dBd]

(a) (b) (c) (d)

Figure
Figure10.
10. Radiation
Radiationpattern
patternofofSubarray2
Subarray2when
whenthe
the44××44MIMOMIMOantenna
antennaisis used. (a)ϕφ==00deg
used. (a) deg
(Combination1).
(Combination1).(b)(b)ϕφ==4545deg (Combination2).(c)(c)ϕφ==9090deg
deg(Combination2). deg(Combination3).
(Combination3).(d) (d)ϕφ==135
135deg
deg
(Combination4).
(Combination4).

5.2.Performance
5.2. PerformanceEvaluation
Evaluationofof32
32××3232MIMO
MIMOAntenna
Antenna
Table 22 lists
Table liststhe
theanalytical conditions
analytical used
conditions for performing
used for performingthe Monte Carlo simulation.
the Monte The number
Carlo simulation. The
of clusters
number and number
of clusters of scatterers
and number were set
of scatterers to set
were 1 and
to 130,
andrespectively. The number
30, respectively. The numberof samples was
of samples
was assumed to be 5000. In the Monte Carlo simulation, the mobile station was moved by a distanceof
assumed to be 5000. In the Monte Carlo simulation, the mobile station was moved by a distance
100λ
of 100λtoto
1000λ,
1000λ, which
which was changed
was changeddepending
depending on on
thethe
angular spread,
angular with
spread, the the
with distance between
distance two
between
successive
two points
successive in the
points simulation
in the being
simulation ∆d =Δd0.02λ
being to 0.2λ.
= 0.02λ to 0.2λ.

Analyticalconditions.
Table2.2.Analytical
Table conditions.

Frequency
Frequency 2000 MHz
2000 MHz
Number of elements (M, N) M = N = 32
Number of elements (M, N) M = N = 32
Number of clusters (Qc ) 1
Number of clusters (Qc) 1
Number of scatterers (Km ) 30
Number of scatterers (Km) 30
Initial phase for scatterers Random
Initial phase for scatterers Random
Traveling distance (d) 1000λ for σ = 10◦
Traveling distanceon(d)
(dependent angular spread) 1000λ
100λ for σfor σ =◦10°
= 100
(dependent on of
Number angular
samplesspread)
(S) 100λ5000
for σ = 100°
Number of samples
Sampling (S) (∆d = d/S)
interval 0.2λ for 5000
σ = 10◦
(dependent
Sampling intervalon angular
(Δd = d/S)spread) 0.02λ σ =σ100
forfor
0.2λ
◦
= 10°
Moving direction (φV ) ◦
(dependent on angular spread) 0.02λ5for σ = 100°
XPR 50 dB (Vertical Pol.)
Moving direction (ϕV) 5°
Antenna element Half-wavelength dipole
XPR 50 dB (Vertical Pol.)
Method of EM analysis Method of moments
Antenna element Half-wavelength dipole
Method of EM analysis Method of moments
One of the most important considerations in the development of the 32 × 32 MIMO antenna is
that the desired directivities of the 32 subarrays are varied via electromagnetic (EM) coupling due to
One of the most important considerations in the development of the 32 × 32 MIMO antenna is
the proximity of the antenna elements. In particular, the radiation patterns of the 16 elements that are
that the desired directivities of the 32 subarrays are varied via electromagnetic (EM) coupling due to
closest to the AOA antenna located at the center of the circle, exhibit a significant reduction. This fact
the proximity of the antenna elements. In particular, the radiation patterns of the 16 elements that are
allows the channel capacity to be reduced because the directivity of the subarray, which is mainly
closest to the AOA antenna located at the center of the circle, exhibit a significant reduction. This fact
composed of these elements, cannot lead the beam in the direction of the incoming wave.
allows the channel capacity to be reduced because the directivity of the subarray, which is mainly
Figure 11 shows the radiation patterns of two subarrays, Subarray2 and Subarray18, as functions
composed of these elements, cannot lead the beam in the direction of the incoming wave.
of the radius r of the array when the angle of the incident wave φ is 0◦ [14]. These two subarrays are
Figure 11 shows the radiation patterns of two subarrays, Subarray2 and Subarray18, as functions
depicted in Figure 3a by the two red broken rectangles. The excitation conditions listed in Table 1 are
of the radius r of the array when the angle of the incident wave ϕ is 0° [14]. These two subarrays are
used for all the subarray groups in the 32 × 32 MIMO antenna.
depicted in Figure 3a by the two red broken rectangles. The excitation conditions listed in Table 1 are
used for all the subarray groups in the 32 × 32 MIMO antenna.
Micromachines 2020, 11, x 13 of 25
Micromachines 2020, 11, 871 13 of 25

10 10
y y
0 0

-10 -10

-20 -20
x x
-10 -10

0 r = 30cm 0 r = 30cm
r = 15cm r = 15cm
4 4MIMO 4 4MIMO
10 10
10 0 -10 -20 -10 0 10 10 0 -10 -20 -10 0 10
[dBd] [dBd]

(a) (b)
Figure11.
Figure 11. Radiation
Radiation patterns
patterns of
of Subarray2
Subarray2 and
and Subarray18
Subarray18 when
when the
the 32
32 ×× 32
32 MIMO
MIMO antenna
antenna is
is used.
used.
(a) Subarray2. (b) Subarray18.
(a) Subarray2. (b) Subarray18.

InFigure
In Figure11, 11,the theblack
blackandand redred solid
solid lines
lines indicate
indicate thethe radiation
radiation patterns
patterns at r at= 30
r =cm30 andcm andr = 15 r =cm,15
cm, respectively.
respectively. The blue
The blue broken broken line indicates
line indicates the radiation
the radiation pattern pattern × 44MIMO
of theof4 the × 4 MIMO arrayarraywithoutwithoutEM
EM coupling
coupling caused caused
by other 4 × 4 4MIMO
by other × 4 MIMO arraysarrays
[19]. The[19].black
The black
arrows arrows
in the in the figure
figure show the show theof
angle angle
the
of the incident
incident wave. As wave.shown As in
shown
Figurein11, Figure 11, the radiation
the radiation gains at an gains at an incident-wave
incident-wave ◦
angle of 0 angle of 0° for
for Subarray2
Subarray2
is is constantof
constant regardless regardless
the radiusofr,the whileradius r, while the
the radiation gain at φ = 0◦gain
radiation for theat ϕ32= × 0°32for the 32array
MIMO × 32
MIMO array agrees well with that for the 4 × 4 MIMO array. This is because the radiation beam from
agrees well with that for the 4 × 4 MIMO array. This is because the radiation beam from Subarray2 is
Subarray2
directed to theis directed
outside of tothe
thewhole
outside of thesystem
antenna wholeand antenna
is lesssystem
affectedand by theis lessEM affected
couplingby the EM
caused by
coupling
the other 4caused
× 4 MIMO by the other 4 × 4 MIMO arrays.
arrays.
In contrast,
In contrast, the the radiation
radiation gainsgains at at φϕ== 00° ◦ for
for Subarray18
Subarray18 are are significantly
significantly degradeddegraded by by the
the EMEM
coupling when
coupling when r == 15 15cm.cm.ThisThisisisdue
duetotothe the fact
factthatthatSubarray18
Subarray18 directs
directsits its
radiation
radiation beam beamtoward
toward the
AOA
the AOAantenna
antenna andand is significantly
is significantly affected
affectedby by EMEM coupling
coupling withwiththe the
AOA AOA antenna.
antenna. However,
However, the
radiation
the radiation gain at rat=r30
gain = 30cmcm is is
improved
improved considerably
considerably and
andisisclose
closetotothat
thatof ofthethe 44 ××44 MIMO
MIMO array.
Theseresults
These resultsshowshowthat thatthe
theradiation
radiationgain gainat atthe
the angle
angle of of the
the incident
incident wavewave increases
increases with with increasing
increasing
radius due to smaller
radius due to smaller EM coupling. EM coupling.
The final
The final target
target of ofour
ourstudy
studyisisto toobtain
obtaintransmission
transmission rates rates greater
greater than than aa gigabit
gigabit to to ensure
ensure the the
successof
success ofthese
thesearrays
arraysin inupcoming
upcomingconnectedconnectedcar carsystems.
systems. Hence,
Hence, to to investigate
investigate the the optimum
optimum radius radius
for this
for this purpose,
purpose, the the channel
channel capacity
capacity isis calculated
calculated through
through Monte Monte Carlo
Carlo simulations,
simulations, with with aa single
single
clusterGaussian
cluster Gaussianspectrumspectrumcoming comingfrom fromazimuth
azimuthangle angleφ, ϕ,asasdepicted
depictedin inFigure
Figure3b. 3b.
Figure12
Figure 12shows
showsthe the average
average channel
channel capacity,
capacity,calculated
calculatedfrom fromEquation
Equation(16), (16),asasaafunction
functionof ofthe
the
radius rr at
radius at 22 GHz
GHz [14].[14]. The
The black
black andand redred curves
curves correspond
correspond to to incident-wave
incident-wave angles angles φ ϕ of ◦
of 00° and 45 45°.◦.

Half-wavelengthdipole
Half-wavelength dipoleantennas
antennasare areused
usedfor forthe
the64 64elements.
elements.The Theradius
radiusofofthe the44×× 44 MIMO
MIMO antenna antenna
a is set to 4.9 cm. The cross-polarization power ratio, XPR, is set
a is set to 4.9 cm. The cross-polarization power ratio, XPR, is set to 50 dB, which is equivalent to to 50 dB, which is equivalent to aa
verticallypolarized
vertically polarizedpropagation
propagationenvironment.
environment. The Thestandard
standarddeviation
deviationof ofthetheincident
incidentwave waveσσ(σ (σ11 for
for
the cluster1
the cluster1in inFigure
Figure7a) 7a)isisset
settoto30 ◦ . The
30°. The SNRSNR of of the
the incident
incident wave wave isissetsetto to30 30dB.
dB.
Figure12
Figure 12 shows
shows that that aa channel
channel capacity
capacity of of over
over 220220 bits/s/Hz
bits/s/Hzisisachieved
achievedwhen whenrr is is greater
greaterthan than
30 cm,
30 cm, which
which isisequivalent
equivalent to to2222Gbps
Gbpsfor foraabandwidth
bandwidthof of100100MHz.
MHz.The Thechannel
channelcapacity
capacityincreases
increases
by 52
by 52bits/s/Hz
bits/s/Hzwhen whenrr== 30 30cm,
cm,compared
comparedwith withthe thecase
casewherewhererr==15 15cm.
cm. Furthermore,
Furthermore, the thechannel
channel
capacityremains
capacity remainsconstantconstantafterafterrr== 30 30 cm,
cm, meaning
meaning that, that, considering
consideringthe thesize
sizeof ofthe array,rr== 30
thearray, 30 cmcm is is
oneofofthe
one thebest best possible
possible solutions
solutions for obtaining
for obtaining the maximum
the maximum channel channel
capacity capacity
with anwith arrayanofarraylimited of
limited
size. Notesize. Note
that thethatblacktheand
blackredand red curves
curves at angles of 0◦ and
at angles ◦ closely
of 0°45and 45° closely
overlap. overlap.
This This is due is due
to the to
the that,
fact fact that, as described
as described in Figurein Figure
4, the 324,×the 32 × 32array
32 MIMO MIMO usesarray usescombinations
different different combinations of subarrays, of
subarrays, Combination1
Combination1 and Combination2, at φ = 0◦ and 45
and Combination2, at ◦ϕ, which
= 0° and 45°, which
enables enables
the radiation theinradiation
gain gain in
these directions
these
to directionswith
be enhanced to beequal
enhanced withConsequently,
intensity. equal intensity. weConsequently,
have the samewe have
level the samecapacity
of channel level of at channel
these
 Micromachines 2020, 11, x 14 of 25
Micromachines 2020,
Micromachines 11,11,
2020, 871x 1414
of of
25 25
capacity at these angles, which confirms that the beam steering function operates correctly, regardless
capacity
of the sizeatofthese angles, which confirms that the beam steering function operates correctly, regardless
the array.
angles,
of thewhich
size ofconfirms
the array.that the beam steering function operates correctly, regardless of the size of
the array.
240
240
230
[bits/s/Hz]
[bits/s/Hz]
230
220
220
210 52 bits/s/Hz
210 52 bits/s/Hz  = 30deg
Capacity

200
aσ==4.9cm
30deg
Capacity

200
190 f a==2GHz
4.9cm
190 f = 2GHz
XPR = 50dB
180
Channel

180 XPR
SNR ==30dB
50dB
Channel

170 SNR = 30dB

170  = 0deg
160 φ==45deg
0deg
160 φ = 45deg
150
10
150 15 20 25 30 35 40 45 50
10 15 20Radius
25 of Array
30 r35[cm] 40 45 50
Radius of Array r [cm]
Figure 12. Channel capacity as a function of the radius of the array.
Figure Channel
12.12.
Figure capacity
Channel as as
capacity a function of of
a function thethe
radius of of
radius thethe
array.
array.
To understand the mechanism for the radius-channel capacity relationship more specifically, we
ToTounderstand
performed understand
an analysisthe
themechanism
mechanism
of the channel for the
the radius-channel
forgain. radius-channel
Figure 13 shows capacity
capacity relationship
relationship
the channel asmore
gain more specifically,
specifically,
a function we
of the
we performed
performed anan analysis
analysis of
of the
the channel
channel gain.
gain. Figure
Figure 13
13 shows
shows the
thechannel
channel
number of the subarray with the radius as a parameter when a single Gaussian wave with a standard gain
gainas
asa function
a function of the
of the
number
number
deviationof of
the σsubarray
ofthe =subarray
30° comeswith
with thethe
from radius
radius
an as as
anglea parameter
a parameter
of ϕ = 0°. The when
when a single Gaussian
a single
channel gain Gaussian wave
is defined aswith
wave with
the aratio
standard
a standard
of the
deviation
deviation of σ
of =σ 30
=
◦ comes from an angle of φ = 0◦ . The channel gain is defined as the ratio of the
30° comes from an angle of ϕ = 0°. The channel gain
received power of the signal at each subarray to that of a single isotropic antenna. Thus, a channelis defined as the ratio of the
received
receivedpower
power of the
of signal
the at
signal
gain exceeding 2.15 dB means that the each
at subarray
each to
subarray that
to of a
that single
of a isotropic
single antenna.
isotropic Thus,
antenna. a channel
Thus,
has more gain than an ordinary dipole antenna. Note a gain
channel
exceeding
gainwe
that 2.15 dB2.15
exceeding
obtained means
the dB that
means
channel the subarray
that
gain inthe
such has
subarray
a waymore hasgain
that the than
more gain
averageanthan
ordinary dipole
an ordinary
operation antenna.
dipole
is performed Note
antenna.
with that
Note
respect
we obtained
that we the
obtained channel
the
to all 32 incident waves. gain
channel in such
gain a
in way
such that
a way the average
that the operation
average is performed
operation is with
performed respect
with to all
respect
32to
incident waves. waves.
all 32 incident

10
10
10
10 Subarrays 1-4
[dB]

Subarrays 1-4
[dB]

55 Subarrays 17-20
[dB]

55 Subarrays 17-20
Gain
[dB]
Gain
Gain

00
Gain
of Channel

00
Channel
Channel
Channel

-5
-5
-5
-5 9-12
ofofof

13-16 9-12 5-8

-10
-10  = 0deg
Average

13-16 5-8
Average

-10
-10 1-4 2 φ==30deg
0deg
Average

17-20
Average

17-20 1-4 2σ rσ==30cm

30deg
-15
-15 21-24 29-32 Incident
-15
-15 rr==20cm
30cm
21-2425-28 29-32 Wave
Incident
25-28 rr==15cm
20cm
-20 Wave r = 15cm
-20
1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132
-20
-20
1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132
Number of Subarray
Number of Subarray
Figure 13.Channel
Figure13. Channelgain
gainasasa afunction
functionofofthe
thenumber
numberofofeach
eachsubarray.
subarray.
Figure 13. Channel gain as a function of the number of each subarray.
In Figure 13, we calculated three cases; r = 30 cm, r = 20 cm, and r = 15 cm. The inset in the figure
In Figure 13, we calculated three cases; r = 30 cm, r = 20 cm, and r = 15 cm. The inset in the figure
shows the geometrical
13, werelationship between the
r =array structure and ther incident
15 cm. wave. Theinblue and
showsIntheFigure
geometrical calculated three
relationship cases;
between the 30 cm,
array r = 20 cm,
structure and
and the= incidentThe inset
wave. The the
bluefigure
and
red subarrays
shows the indicate that
geometrical these subarrays,
relationship between Subarrays
the array 1–4 and Subarrays
structure and the 17–20, are
incident located
wave. The in the
blue and
red subarrays indicate that these subarrays, Subarrays 1–4 and Subarrays 17–20, are located in the
red subarrays indicate that these subarrays, Subarrays 1–4 and Subarrays 17–20, are located in the
 Micromachines 2020, 11, x 15 of 25
Micromachines 2020, 11, 871 15 of 25
forward and backward directions with respect to the incident wave, as shown in the inset. The blue
and red arrows included in the upper part of Figure 13 indicate the locations of Subarrays 1–4 and
forward and backward directions with respect to the incident wave, as shown in the inset. The blue
Subarrays 17–20 on the lateral axis.
and red arrows included in the upper part of Figure 13 indicate the locations of Subarrays 1–4 and
Figure 13 shows that, regardless of the radius, Subarrays 1–4, which face the incident wave,
Subarrays 17–20 on the lateral axis.
provide a large channel gain of more than 2.15 dB, meaning that these subarrays operate more
Figure 13 shows that, regardless of the radius, Subarrays 1–4, which face the incident wave,
effectively than a conventional dipole antenna. On the other hand, for Subarrays 17–20, the channel
provide a large channel gain of more than 2.15 dB, meaning that these subarrays operate more effectively
gains exhibit a significant reduction as the radius decreases. In particular, when r = 15 cm the channel
than a conventional dipole antenna. On the other hand, for Subarrays 17–20, the channel gains exhibit a
gain is reduced to −10 dB, implying that these subarrays do not yield a useful subchannel for MIMO
significant reduction as the radius decreases. In particular, when r = 15 cm the channel gain is reduced
transmission. This phenomenon for the reduction in channel gain can also be understood from the
to −10 dB, implying that these subarrays do not yield a useful subchannel for MIMO transmission.
radiation patterns, as described in Figure 11. It is concluded from these results that the reduction in
This phenomenon for the reduction in channel gain can also be understood from the radiation patterns,
channel capacity with decreasing radius r, as described in Figure 12, is caused by a decrease in the
as described in Figure 11. It is concluded from these results that the reduction in channel capacity
channel gains of the subarrays, particularly that situated at the back with respect to the incident wave,
with decreasing radius r, as described in Figure 12, is caused by a decrease in the channel gains of the
as illustrated in Figure 13.
subarrays, particularly that situated at the back with respect to the incident wave, as illustrated in
Figure 14 shows the channel capacity of the 32 × 32 MIMO array as a function of the angle ϕ
Figure 13.
when r = 30 cm. The angular spread σ and SNR were set to 30° and 30 dB, respectively. For
Figure 14 shows the channel capacity of the 32 × 32 MIMO array as a function of the angle φ when
comparison, the channel capacity of a 32 × 32 MIMO dipole array is included in the graph. The dipole
r = 30 cm. The angular spread σ and SNR were set to 30◦ and 30 dB, respectively. For comparison,
array comprises eight groups of four-element dipole linear arrays with a spacing of 3 cm arranged at
the channel capacity of a 32 × 32 MIMO dipole array is included in the graph. The dipole array
equal 45-degree angles aligned along the y-axis, situated at the same locations as Subarrays 1–4 to
comprises eight groups of four-element dipole linear arrays with a spacing of 3 cm arranged at equal
29–32, as indicated by the inset in the figure. As shown in Figure 14, when the proposed antenna is
45-degree angles aligned along the y-axis, situated at the same locations as Subarrays 1–4 to 29–32,
used, the channel capacity remains constant regardless of the angle of the incident wave. In contrast,
as indicated by the inset in the figure. As shown in Figure 14, when the proposed antenna is used,
in the case of the dipole array antenna, more than 32-bits/s/Hz degradation in the channel capacity is
the channel capacity remains constant regardless of the angle of the incident wave. In contrast, in the
observed when the angle of the incident wave is 90° or 270°, in which the dipole array is parallel to
case of the dipole array antenna, more than 32-bits/s/Hz degradation in the channel capacity is observed
the incident wave. The dipole array gives a channel capacity of less than 205 bits/s/Hz because of low
when the angle of the incident wave is 90◦ or 270◦ , in which the dipole array is parallel to the incident
SNR or high correlation. In contrast, the 32 × 32 MIMO array achieves a constant channel capacity of
wave. The dipole array gives a channel capacity of less than 205 bits/s/Hz because of low SNR or high
220 bits/s/Hz regardless of the direction of the incoming wave owing to the full-azimuth beam
correlation. In contrast, the 32 × 32 MIMO array achieves a constant channel capacity of 220 bits/s/Hz
steering function.
regardless of the direction of the incoming wave owing to the full-azimuth beam steering function.

240

230
Channel Capacity [bits/s/Hz]

220

210 32bits/s/Hz
200

190

180 Dipole Array r = 30cm

r σ = 30deg
170 f = 2GHz Proposed
SNR = 30dB Dipole
160
0 45 90 135 180 225 270 315 360
Angle of Incident Wave φ [deg]
Figure Channel
14.14.
Figure capacity
Channel asas
capacity a function ofof
a function the angle
the ofof
angle the incident
the wave.
incident wave.
5.3. Antenna–Propagation Mutual Interactions
5.3. Antenna–Propagation Mutual Interactions
As mentioned in the introduction, the MIMO antenna for a connected car system, implemented in
As mentioned in the introduction, the MIMO antenna for a connected car system, implemented
the framework for the fifth and sixth generation (5G and 6G) mobile communications, is anticipated
in the framework for the fifth and sixth generation (5G and 6G) mobile communications, is
to be used in a system where the distance between the base station and the mobile station is small,
anticipated to be used in a system where the distance between the base station and the mobile station
compared with the conventional macrocell counterpart. In a small cell system, there are two significant
is small, compared with the conventional macrocell counterpart. In a small cell system, there are two
Micromachines
Micromachines2020, 11,11,
2020, 871x 16
16ofof2525

significant factors to be considered that affect the channel capacity of a MIMO system; the signal-to-
factors
noise to be considered
power ratio (SNR) that
and affect the channel
the spatial capacity
angular powerof a MIMO
spread system;
of the the signal-to-noise
incoming power
waves. The interaction
ratio (SNR) and the spatial angular power spread of the incoming waves. The interaction
between these two physical quantities is complex and there might be a tradeoff relationship because, between these
two
for physical
example,quantities
when a mobileis complex
stationand there might
is situated be aproximity
in close tradeoff relationship because,
to a base station for example,
the received power
when a mobile station is situated in close proximity to a base station the
of the mobile station increases due to the large SNR but the angular power spread of the received power of the received
mobile
station
signalincreases
decreasesdue to the
owing to large SNR but the
the line-of-sight angular
(LOS) powerofspread
property the radioof the received signal
propagation decreases
mechanism.
owingOn to the
theline-of-sight
basis of the (LOS) property of the
above-mentioned radio propagation
background, we carried mechanism.
out a Monte Carlo simulation to
investigate how the developed 32 × 32 MIMO array performs in aout
On the basis of the above-mentioned background, we carried a Monte Carlo
propagation simulation
environment to
where
investigate
forthcoming how 5Gtheanddeveloped
6G connected 32 ×car
32 systems
MIMO array performs in
are anticipated a propagation
to be used. Figureenvironment
15 shows the where
channel
forthcoming 5G and 6G connected car systems are anticipated to be
capacity as a function of the angular power spread of an incident wave with the SNRused. Figure 15 showsas the channel
a parameter.
capacity
In Figure as 15,
a function
the solidoflines
the angular power
denote the spread
cases where ofthe
an incident
radius ofwave withrthe
the array SNR
is 30 cmas
anda parameter.
the dashed
Inlines
Figure 15, the solid lines denote the cases where the radius of the array r is
represent the cases of r = 20 cm. We assume a single incident wave coming from ϕ = 0° along 30 cm and the dashed the
lines represent
x-axis. the cases of r = 20 cm. We assume a single incident wave coming from φ = 0 ◦ along

the x-axis.

300
r = 30cm
r = 20cm
Channel Capacity [bits/s/Hz]

250
φ = 0deg
SNR = 30dB
200

150
SNR = 20dB
100

50
SNR = 10dB

0
0 20 40 60 80 100
Angular Spread σ [deg]
Figure 15. Channel capacity vs. the angular power spread.
Figure 15. Channel capacity vs. the angular power spread.

It can be seen from Figure 15 that the 32 × 32 MIMO array achieves a maximum channel capacity
It can be seen from Figure 15 that the 32 × 32 MIMO array achieves a maximum channel capacity
of 277 bits/s/Hz when SNR = 30 dB, r = 30 cm, and σ = 100◦ . Moreover, Figure 15 shows that the
of 277 bits/s/Hz when SNR = 30 dB, r = 30 cm, and σ = 100°. Moreover, Figure 15 shows that the
channel capacity decreases significantly as the angular power spread of the incoming wave decreases.
channel capacity decreases significantly as the angular power spread of the incoming wave decreases.
Specifically, when SNR = 30 dB and r = 30 cm the channel capacity is 268 bits/s/Hz in the case of
Specifically, when SNR = 30 dB and r = 30 cm the channel capacity is 268 bits/s/Hz in the case of σ =
σ = 60◦ . In contrast, the channel capacity reduces to 135 bits/s/Hz in the case of σ = 10◦ , which is
60°. In contrast, the channel capacity reduces to 135 bits/s/Hz in the case of σ = 10°, which is equivalent
equivalent to a 50% reduction. In [10], the channel model parameters for various cellular systems
to a 50% reduction. In [10], the channel model parameters for various cellular systems are
are summarized, and the angular spread at the mobile station is found to vary between 30◦ and 80◦
summarized, and the angular spread at the mobile station is found to vary between 30° and 80° (see
(see Table 2 in [10]). Judging from this fact, Figure 15 indicates that the developed antenna has good
Table 2 in [10]). Judging from this fact, Figure 15 indicates that the developed antenna has good
performance in situations encountered in typical application scenarios.
performance in situations encountered in typical application scenarios.
To clarify the mechanism responsible for the observations described in Figure 15, the channel gain
To clarify the mechanism responsible for the observations described in Figure 15, the channel
and correlation characteristics were investigated.
gain and correlation characteristics were investigated.
Figure 16 shows the channel gain and correlation coefficient of each subarray of the 32 × 32 MIMO
Figure 16 shows the channel gain and correlation coefficient of each subarray of the 32 × 32
array as a function of the number of the subarray. The black lines with round symbols denote the
MIMO array as a ◦function of the number of the subarray. The black lines with round symbols denote
case where σ = 10 , whereas the red lines with square symbols represent the case where σ = 60◦ .
the case where σ = 10°, whereas the red lines with square symbols represent the case where σ = 60°.
The correlation coefficient is defined as the absolute value of the complex correlation coefficient between
The correlation coefficient is defined as the absolute value of the complex correlation coefficient
Subarray1 and the other subarrays, defined as ρa in Equation (10). Note that we obtained the correlation
between Subarray1 and the other subarrays, defined as ρa in Equation (10). Note that we obtained the
coefficient in such a way that the average operation is performed with respect to all 32 incident waves.
correlation coefficient in such a way that the average operation is performed with respect to all 32
As described in Figure 6, and Equations (3) and (4) in Section 4, the incident waves from the base
incident waves. As described in Figure 6, and Equations (3) and (4) in Section 4, the incident waves
from the base station are assumed to be plane waves, i.e., far-field assumption, and thus the channel
dB and is larger than that for σ = 60°. This is due to the fact that when σ = 10° the peak gain at ϕ = 0°
results in a large channel gain due to in-phase excitation of the array, as depicted in Figure 11. In
contrast, the channel gain for the subarrays numbered 13–24 is small in magnitude because of the
shadowing effects caused by the subarrays located in front of them. Figure 16b shows that the
Micromachinescoefficients
correlation 2020, 11, 871 for σ = 10° are larger than those for σ = 60°. Especially, the correlations between
17 of 25

Subarrays 1 and 2, 1 and 3, and 1 and 4, which belong to the same 4 × 4 MIMO array group, are more
than 0.6. In contrast, when σ = 60° the correlation is less than 0.5 over the entire array.
station are assumed to be plane waves, i.e., far-field assumption, and thus the channel model does not
include the effects of the distance between the base and mobile stations on the correlation.

9 1
 = 10deg  = 10deg

Absolute of Complex Correlation

8 0.9
Average of Channel Gain [dB]

 = 60deg  = 60deg
7 0.8
r = 30cm r = 30cm
6 0.7
SNR = 30dB SNR = 30dB
5  = 0deg 0.6  = 0deg
4 0.5
3 0.4
2 0.3
1 0.2
0 0.1
-1 1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132
01 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132
Number of Subarray Number of Subarray

(a) (b)
Figure 16. Channel
Figure 16. Channelgain
gain and
and correlation
correlation characteristics
characteristics of eachofsubarray.
each subarray. (a)gain.
(a) Channel Channel gain. (b)
(b) Correlation.
Correlation.
It can be seen from Figure 16a that a channel gain of more than 2 dB is obtained for the subarrays
The changes
numbered 1–8 and in29–32
the channel gain
for both σ =and
10◦correlation
and 60◦ . Particularly,
coefficientsthe mentioned
channel above σ = 10
result
gain for in ◦aexceeds
variety
4 dB and is larger than that for σ = 60 . This is due to the fact that when σ = 10 the peak gain
of distributions of eigenvalues. Figure ◦
17a,b show the cumulative distribution ◦
functions (CDF) ofat
theφ
= 0 results when
◦
eigenvalues σ = 60°
in a large and σgain
channel = 10°.
due The eigenvalues
to in-phase are obtained
excitation of the from
array,Equation (14)in
as depicted through
Figure an 11.
SVD operation.
In contrast, the channel gain for the subarrays numbered 13–24 is small in magnitude because of
the shadowing17a,
In Figure the presence
effects caused byofthe 32 subarrays
densely distributed
located ineigenvalues
front of them. in the graph
Figure means
16b shows that there
that the
are 32 effective
correlation coefficients for σ =for
subchannels ◦
10 MIMO transmission.
are larger for σindicates
than those This = 60 . Especially,
◦ that the the
channel matrix between
correlations defined
by Equation
Subarrays (8) has
1 and 2, 1 aand
full-rank
3, and 1property corresponding
and 4, which belong to theto the
samenumber of subarrays
4 × 4 MIMO because
array group, areallmore
the
subarrays
than 0.6. In can receivewhen
contrast, signalsσ=from ◦
60 the communication
correlation is lesstarget
than owing
0.5 over tothe
theentire
beamarray.
steering function of
the proposed
The changes antenna.
in theThis feature
channel gainis quite different from
and correlation a conventional
coefficients mentioned circular
above array antenna
result with
in a variety
aofnumber of patch antennas arranged in a cylindrical manner [7,8], as described
distributions of eigenvalues. Figure 17a,b show the cumulative distribution functions (CDF) of the in the introduction.
Micromachines
eigenvalues2020,
Consequently, the
when xσ = 60◦ and
11, developed σ = 10◦operates
antenna normally are
. The eigenvalues andobtained
effectively as aEquation
from 32-element MIMO
(14) 18
through of an
array25
antenna.
SVD operation.
In contrast, the sparsely distributed 18 eigenvalues, shown in Figure 17b, indicate that the
developed 100 array works as a MIMO antenna with a small 100 number of elements. More specifically, a
90 r = 30cm
in the case of σ = 10° provides large 90
Cumulative Percentage [%]
Cumulative Percentage [%]

large channel gain

SNR = 30dB first and second eigenvalues, as indicated by
80 80
 = 0deg
the symbol “A” on the right side of Figure 17b. Moreover, a high correlation coefficient for σ = 10°
70 70
creates small
60 eigenvalues distributed on the left side of60Figure 17b, as B indicated by the symbol A “B”.
Eventually, 50 when σ = 10° these two observations result50in a small channel capacity, as described in
Figure 15. 40 40
30 30
20 20
10 10
0 0
10-4 10-2 100 102 104 10-4 10-2 100 102 104
Eigenvalue  Eigenvalue 

(a) (b)
Figure 17.
Figure 17. CDF
CDF characteristics
characteristics of eigenvalues. (a) σ = 60°.
of eigenvalues. (b)σσ==10°.
60◦ . (b) 10◦ .

In Figure
From the 17a, the presence
extensive of 32investigations
analytical densely distributed eigenvalues
performed in thisinpaper,
the graph
we means
concludethatthat
there
a
are 32 effective
remarkably highsubchannels for MIMO
channel capacity transmission.
of more This indicates
than 200 bits/s/Hz that
achieved bythe
thechannel matrix defined
beam steering 32 × 32
by Equation
MIMO (8) has
array—as a full-rank
depicted property
in Figures corresponding
12, 14, and 15—can tobethe numbertooftwo
attributed subarrays because
significant all the
factors; the
subarrays can
improved receive signals
directivity due to from the communication
in-phase target
excitation of the owing
array andtothe
the low
beamcorrelation
steering function
between of
subarrays caused by the orthogonal alignment of the array with respect to the incident waves.

6. Experimental Verification
To confirm the validity of the proposed antenna, this section presents experiments carried out
 100 100
90 r = 30cm 90

Cumulative Percentage [%]

Cumulative Percentage [%]
SNR = 30dB
80 80
 = 0deg
Micromachines 2020, 11, 871 18 of 25
70 70
60 60 B A
50 50
the proposed
40
antenna. This feature is quite different from
40
a conventional circular array antenna with
a number30of patch antennas arranged in a cylindrical manner 30 [7,8], as described in the introduction.
Consequently,
20 the developed antenna operates normally 20 and effectively as a 32-element MIMO
10
array antenna. 10
0 0
In contrast,
10-4 the
10-2 sparsely
100 distributed
102 1018
4 eigenvalues,
10-4 shown
10-2 in Figure
100 17b,
102 indicate
104 that the
a MIMO antenna with a small number of elements.
developed array works asEigenvalue  specifically, a large
More
Eigenvalue
channel gain in the case of σ = 10◦ provides large first and second eigenvalues, as indicated by the
symbol “A” on the right(a) side of Figure 17b. Moreover, a high correlation coefficient (b) for σ = 10◦ creates
small eigenvalues distributed on the
Figure 17. CDF left side of Figure
characteristics 17b, as indicated
of eigenvalues. (a) σ = 60°.by
(b)the
σ =symbol
10°. “B”. Eventually,
when σ = 10◦ these two observations result in a small channel capacity, as described in Figure 15.
From the
From theextensive
extensive analytical
analytical investigations
investigations performed
performed in this in thiswe
paper, paper, we that
conclude conclude that a
a remarkably
remarkably
high channelhigh channel
capacity capacity
of more of more
than 200 thanachieved
bits/s/Hz 200 bits/s/Hz
by the achieved by the
beam steering 32beam steering
× 32 MIMO 32 × 32
array—as
MIMO array—as
depicted in Figure depicted
12, Figurein
14,Figures 12, 14,
and Figure and 15—can
15—can be attributed
be attributed to two significant
to two significant factors; thefactors; the
improved
improved due
directivity directivity dueexcitation
to in-phase to in-phase excitation
of the array andofthethe
lowarray and the
correlation low correlation
between between
subarrays caused by
subarrays
the caused
orthogonal by the orthogonal
alignment of the arrayalignment of the
with respect array
to the with respect
incident waves.to the incident waves.

6. Experimental Verification
6. Experimental Verification
To
To confirm the validity
confirm the validityofofthe
theproposed
proposedantenna,
antenna, this
this section
section presents
presents experiments
experiments carried
carried out
out to evaluate the 4 × 4 and 32 × 32 MIMO arrays. We performed two types of measurements;
to evaluate the 4 × 4 and 32 × 32 MIMO arrays. We performed two types of measurements; radiation
radiation pattern measurements
pattern measurements in an anechoic
in an anechoic chamberchamber and over-the-air
and over-the-air (OTA) (OTA)
testingtesting
using using
a two-a
two-dimensional fading emulator in a cluster propagation environment
dimensional fading emulator in a cluster propagation environment [26]. [26].

6.1. Impedance and

6.1. Impedance and Radiation
Radiation Measurements
Measurements
Figure
Figure 18 18 shows
shows thethe setup
setup for
for measuring
measuring the the radiation
radiation pattern
pattern ofof aa 44 ×
× 44 MIMO
MIMO array
array in
in an
an
anechoic chamber [15]. The signal processing unit was designed using the circuit
anechoic chamber [15]. The signal processing unit was designed using the circuit topology depicted topology depicted in
Figure 8a, including a phase shifter and a signal combiner. Figure 18a shows
in Figure 8a, including a phase shifter and a signal combiner. Figure 18a shows a fabricated a fabricated microwave
circuit for the
microwave weight
circuit forfunctions
the weightof Subarry2
functions (w and w4 ) using
of1Subarry2 a double-sided
(w1 and printed circuitprinted
w4) using a double-sided board
with
circuit board with a thickness of 1 mm and a relative permittivity of 4.2 (FR4). The phase shift signal
a thickness of 1 mm and a relative permittivity of 4.2 (FR4). The phase shift of the received of the
for each element was realized using different lengths of microstrip transmission lines. The 30◦ phase
received signal for each element was realized using different lengths of microstrip transmission lines.
difference
The 30° phasebetween #1 andbetween
difference #4 is equivalent
#1 and #4 to is
a difference
equivalentintolength of 7.6 mm.
a difference The phase
in length of 7.6difference
mm. The
of the produced network was 34◦ , indicating that the phase difference was achieved. Furthermore,
phase difference of the produced network was 34°, indicating that the phase difference was achieved.
to get the correct
Furthermore, weight
to get function,
the correct the difference
weight function,inthe
power loss between
difference in powertheloss
microstrip
between lines,
the which was
microstrip
0.21 dB, was considered.
lines, which was 0.21 dB, was considered.

/2-dipole Coaxial cable

Element #4
Input port

Phase shifter Subarray2

Power divider Element #1

Microwave circuit Attenuators

(a) (b)
Figure 18. Radiation
Radiationpattern
patternmeasurement
measurementsetupsetupofofa 4a ×4 4×MIMO
4 MIMO array in an
array in anechoic chamber.
an anechoic (a)
chamber.
Fabricated
(a) microwave
Fabricated circuit.
microwave (b) (b)
circuit. Prototype of aof4 a×44 ×MIMO
Prototype 4 MIMOarray.
array.

An attenuator was used for realizing the weight function. Since w1 = 1.0 and w4 = 0.38 for
Subarray2, as listed in Table 1, the received signal of #4 had to be synthesized by attenuating 8.40 dB
compared to that of #1. Since the power loss for #4 due to the microstrip line was 0.21 dB larger than
 Micromachines 2020, 11, x 19 of 25

An attenuator was used for realizing the weight function. Since w1 = 1.0 and w4 = 0.38 for
Subarray2,2020,
Micromachines as listed
11, 871in Table
1, the received signal of #4 had to be synthesized by attenuating 8.40
19 of dB
25
compared to that of #1. Since the power loss for #4 due to the microstrip line was 0.21 dB larger than
that for #1, a power loss of 8.19 dB had to be realized by the attenuator. 0 dB and 8 dB attenuators
that for #1, a power loss of 8.19 dB had to be realized by the attenuator. 0 dB and 8 dB attenuators were
were connected to the terminals of #1 and #4 of the fabricated microwave circuit, respectively. The 0
connected to the terminals of #1 and #4 of the fabricated microwave circuit, respectively. The 0 dB
dB attenuator was used for realizing the retention of the phase of the received signals of #1 and #4.
attenuator was used for realizing the retention of the phase of the received signals of #1 and #4. Hence,
Hence, the difference in the received power between #1 and #4 was 8.50 dB, indicating that the weight
the difference in the received power between #1 and #4 was 8.50 dB, indicating that the weight function
function had been achieved.
had been achieved.
The radiation pattern of a 4 × 4 MIMO array antenna was measured in an anechoic chamber.
The radiation pattern of a 4 × 4 MIMO array antenna was measured in an anechoic chamber.
Figure 18b shows the measurement setup of a 4 × 4 MIMO antenna, which comprises eight half-
Figure 18b shows the measurement setup of a 4 × 4 MIMO antenna, which comprises eight
wavelength dipole antennas designed at 2 GHz. Two attenuators and the fabricated microwave
half-wavelength dipole antennas designed at 2 GHz. Two attenuators and the fabricated microwave
circuit were located under the antenna and connected to the antenna using coaxial cables. As shown
circuit were located under the antenna and connected to the antenna using coaxial cables. As shown in
in Figure 18b, the prototype of a 4 × 4 MIMO antenna was mounted on a turntable for sampling
Figure 18b, the prototype of a 4 × 4 MIMO antenna was mounted on a turntable for sampling signals
signals of the radiation patterns at azimuth angles in the xy-plane using a vector network analyzer.
of the radiation patterns at azimuth angles in the xy-plane using a vector network analyzer.
Figure 19 exhibits the impedance characteristics of Subarray2 drawn on a Smith chart measured
Figure 19 exhibits the impedance characteristics of Subarray2 drawn on a Smith chart measured
at the input port of the microwave circuit, as indicated in Figure 18a, using a network analyzer [15].
at the input port of the microwave circuit, as indicated in Figure 18a, using a network analyzer [15].
In Figure 19, the black broken circle represents the impedance locus corresponding to a voltage
In Figure 19, the black broken circle represents the impedance locus corresponding to a voltage standing
standing wave ratio (VSWR) of 2. It can be seen from Figure 19 that a good matching condition with
wave ratio (VSWR) of 2. It can be seen from Figure 19 that a good matching condition with a VSWR of
a VSWR of less than 2 was achieved in the frequencies between 1.8 GHz and 2.2 GHz, where existing
less than 2 was achieved in the frequencies between 1.8 GHz and 2.2 GHz, where existing MIMO-based
MIMO-based cellular systems are being operated.
cellular systems are being operated.

0.5

2 GHz
1.8 GHz
0

2.2 GHz
VSWR = 2
-0.5

-1
-1 -0.5 0 0.5 1
Figure 19. Measured impedance of Subarray2.
Figure 19. Measured impedance of Subarray2.
Figure 20 presents the measured and calculated θ-polarized radiation patterns of Subarray1
Figure 20 presents
and Subarray2 the measured
in the horizontal and calculated
(xy) plane when theθ-polarized
angle of theradiation
incidentpatterns ◦ [27]. In the
wave isof0Subarray1 and
Subarray2 in we
measurement, the performed
horizontal a(xy) plane when
calibration the anglefor
to compensate ofthe
thepower
incident
loss.wave is 0°
The red [27].
line In the
indicates
measurement,
the we performed
measured results, while the a calibration to compensate
blue line represents for the power
the calculated loss.The
results. Thefrequency
red line indicates
for the
the measured was
measurements results, while the blue line represents the calculated results. The frequency for the
2 GHz.
measurements
As shown in was 2 GHz.
Figure 20, there are some appreciable discrepancies between the measured and
As shown
calculated results,inparticularly
Figure 20, inthere
the are someand
sidelobe appreciable discrepancies
backlobe angular regions.between the measured
A possible and
cause of these
calculated results, particularly in the sidelobe and backlobe angular regions. A possible
discrepancies is the electromagnetic distortion arising from coaxial feed cables connected between the cause of these
discrepancies
dipole elementsisandthe the
electromagnetic distortion
microwave circuit, arisinginfrom
as shown coaxial
Figure feed cables connected
18b. Nevertheless, between
good agreement
the dipole
between elements results
the measured and the andmicrowave
the analyticalcircuit, as shown
outcome in Figure
can be observed 18b.direction
in the Nevertheless, good
of the x-axis
agreement
indicated bybetween
the blackthe measured
arrows, resultsthat
confirming andthetheweight
analytical outcome
functions andcanthebe observed
phase in the direction
shift values listed in
of the1 x-axis
Table indicated
work well by the
according toblack arrows,
the angle confirming
of the that the weight functions and the phase shift
incident wave.
values listed in Table 1 work well according to the angle of the incident wave.
Micromachines 2020,
Micromachines 11, 11,
2020, x x 2020
of 25
of 25
Micromachines 2020, 11, 871 20 of 25

1010 1010
y y y y
0 0 0 0

-10-10 -10-10

-20-20 -20-20
x x x x
-10-10 -10-10

0 0 Measured
0 0 Measured
Measured Measured
Calculated
Calculated Calculated
Calculated
1010 1010
1010 0 0 -10-10-20-20-10-10 0 0 1010 1010 0 0 -10-10-20-20-10-10 0 0 1010
[dBd]
[dBd] [dBd]
[dBd]
(a)(a) (b)(b)
Figure 20.20.
Figure Measured andand
Measured calculated results
calculated of radiation
results patterns. (a) (a)
Subarray1. (b)(b)
Subarray2.
Figure 20. Measured and calculated results ofofradiation
radiation patterns.
patterns. Subarray1.
(a) Subarray1. Subarray2.
(b) Subarray2.

6.2. Over-the-Air
6.2.
6.2. Testing
Over-the-Air
Over-the-Air Testing
Testing
The
Thechannel
The capacity
channel
channel was
capacity
capacity measured
was
was using
measured
measured a two-dimensional
using
using fading
aatwo-dimensional
two-dimensional emulator
fading
fading [26],
emulator
emulator asas
[26],
[26], illustrated
as illustrated
illustrated
inin
Figure
inFigure21.
Figure21.
21.

#5 #4
#6 #3 Scatterer
14 Scatterers
(Sleeve Dipole)
#7 #2
32×32 MIMO

1.2m
#8 2 #1
0.3m
Gaussian
#9 Wave #14

#10 #13
#11 #12

(a)(a) (b)(b)
Figure
Figure 21.21.
Figure OTA
21. OTA
OTAtesting of of
testing
testing aof32 ×
3232
aa32 daisy
×× 32
32 chain
daisy
daisy MIMO
chain
chain antenna.
MIMO
MIMO (a) (a)
antenna.
antenna. Two-dimensional
(a) fading
Two-dimensional
Two-dimensional emulator.
fading
fading emulator.
emulator.
(b)(b)
External
(b)External
Externalview.
view.
view.

InIn
In [28],
[28],
[28], the
the the appropriate
appropriate
appropriate number
number
number ofof probe
probe
of probe antennas
antennas
antennas forfora afading
for a fading
fading emulator
emulator
emulator was
waswas investigated
investigated
investigated
theoretically.
theoretically.
theoretically. TheThe
The results
results
results show
showshow that
that 11–15
11–15
that 11–15 scatterers
scatterers
scatterers need
needneedtotobebe
to implemented
implemented
be implemented forfor achieving
achieving
for achieving a agood
agood
good
multi-cluster
multi-cluster
multi-cluster propagation
propagation
propagation environment.
environment.
environment. On
OnOn the
thethe basis
basis of
of of
basis this
this knowledge,
knowledge,
this knowledge, we
wewe have
havehave developed
developed
developed a fading
a fading
a fading
emulator
emulator
emulator with
with
with1414 scatterers.
scatterers.
14 scatterers.
Fourteen
Fourteen scatterers
scatterers
Fourteen scatterers with
with withequal
equal
equal angular
angular
angular intervals
intervals
intervals were
were
were arranged
arranged
arranged onona circle
a circlewith
with aa radius
with radius of
a radius of1.2
1.2m.
of 1.2
The
m.m.Thescatterers
scattererscomprise
comprise vertically polarized
vertically polarizedhalf-wavelength
half-wavelength sleeve dipole
sleeve
The scatterers comprise vertically polarized half-wavelength sleeve dipole antennas. The beam antennas.
dipole The
antennas. beam
The steering
beam
MIMO
steering
steering array
MIMO
MIMO antenna
array
array is located
antenna
antenna isatlocated
is locatedthe center
at at
thethe ofcenter
the of
center emulator,
thethe
of as shown
emulator,
emulator, as as inshown
shown Figure
in in21.
Figure This
Figure fading
21.21.This
This
emulator
fading has
emulator thehasfunction
the of an
function operating
of an algorithm
operating on
algorithmthe basis
on theof the
basis
fading emulator has the function of an operating algorithm on the basis of the same formulation same
of formulation
the same described
formulation
in Section
described
described in4.
in To
Section change
Section 4. the
4. To To anglethe
change
change ofthe
the
angleincident
angleof of
thethewave
incidentfrom
incident wave0◦ to
wave 360
from ◦ , the beam steering array was
from0° 0°
to to
360°,
360°,thethe
beam beamsteering
steering
mounted on a turntable to rotate at 45 ◦ steps in azimuth while maintaining a power spectrum
array
array was mounted on a turntable to rotate at 45° steps in azimuth while maintaining power
was mounted on a turntable to rotate at 45° steps in azimuth while maintaining a a with
powera
Micromachines
Micromachines2020, 11,11,
2020, 871x 21
21ofof2525

spectrum with a fixed Gaussian distribution generated on the scatterers, as depicted by the blue curve
fixed Gaussian distribution generated on the scatterers, as depicted by the blue curve in Figure 21a.
in Figure 21a. The experiments were performed in a cluster propagation environment, which had an
The experiments were performed in a cluster propagation environment, which had an angular power
angular power spectrum with a Gaussian distribution with an incident wave angle of µ = 0° and a
spectrum with a Gaussian distribution with an incident wave angle of µ = 0◦ and a standard deviation
standard deviation of σ = 30°. The frequency was 2 GHz. The SNR was set to 30 dB and 40 dB.
of σ = 30◦ . The frequency was 2 GHz. The SNR was set to 30 dB and 40 dB.
Figure 22 shows the measured channel capacity of the 4 × 4 MIMO antenna, as illustrated in
Figure 22 shows the measured channel capacity of the 4 × 4 MIMO antenna, as illustrated in
Figure 18b, as a function of the angle of the incident wave. Half-wavelength dipole antennas were
Figure 18b, as a function of the angle of the incident wave. Half-wavelength dipole antennas were used
used for the array elements. In the experiments, using the microwave circuit in Figure 18a, we
for the array elements. In the experiments, using the microwave circuit in Figure 18a, we manually
manually switched the RF ports to set the amplitude and phase appropriate for each subarray, in
switched the RF ports to set the amplitude and phase appropriate for each subarray, in accordance
accordance with Table 1. The development of a switching network, as depicted in Figure 5, for
with Table 1. The development of a switching network, as depicted in Figure 5, for conducting
conducting time-independent measurements is a subject for our future studies.
time-independent measurements is a subject for our future studies.

40
Channel Capacity [bits/s/Hz]

35

30
Dipole
Array
25
Ver. Pol. OTA (4×4 MIMO)
OTA (Dipole)
SNR = 30dB Cal. (4×4 MIMO)
3cm
f = 2GHz Cal. (Dipole)
20
0 45 90 135 180 225 270 315 360
Angle of Incident Wave φ [deg]
Figure22.
Figure 22.Channel
Channelcapacity
capacityofofa a4 4××44MIMO
MIMOarray
arrayantenna
antennameasured
measuredby
byOTA
OTAtesting.
testing.

InInFigure
Figure22, 22,The
Theblack
blackround
roundsymbols
symbolsdenotedenotethe themeasured
measuredresults
resultsofofthethe4 4××44MIMOMIMOantenna,
antenna,
whereas
whereasthe theblue
bluecurve
curverepresents
representsthe theanalytical
analyticalresults
resultsusing
usingMonte
MonteCarloCarlosimulations.
simulations.The Theredredcurve
curve
depicts
depictsthe themeasured
measuredresultsresultsofofa afour-element
four-elementlinear lineardipole
dipolearray
arraycomprising
comprisingvertically
verticallypolarized
polarized
half-wavelength
half-wavelengthdipole dipole antennas
antennas with
with aaspacing
spacing ofof3 3cm,
cm,asasdenoted
denotedbybythe theinset
insetininFigure
Figure22, 22,ininwhich
which
the
theentire
entirearrayarraysize
sizeisisthethesame
sameasasthe the4 4× ×4 4MIMO
MIMOantenna.
antenna.Figure
Figure22 22shows
showsthat thatthere
thereisisgood
good
agreement
agreementbetween betweenthe themeasured
measuredand andcalculated
calculatedresults
resultsfor the4 4× ×4 4MIMO
forthe MIMOantenna.
antenna.Furthermore,
Furthermore,
the
theproposed
proposedantenna antennamaintains
maintainsaaconstant
constanttransmission
transmissionrate rateover
over360 360degrees
degreesofofazimuthal
azimuthalangle,angle,
whereas
whereasthe thedipole
dipole array
array yields
yields heavy
heavy fluctuations
fluctuations with
withdeep
deepnulls at angles
nulls at anglesof 90of◦ 90°
andand 270◦270°,
, which is
which
similar to the
is similar result
to the for the
result 32 ×3232× MIMO
for the 32 MIMO array antenna
array antenna described
describedin Figure 14. 14.
in Figure
Figure
Figure2323showsshowsthe thechannel
channelcapacity
capacityofofa aprototype
prototypeofofthe the3232××3232daisy
daisychain
chainMIMOMIMOantennaantennaasas
a afunction
functionofofthe theangle
angleofofthetheincident
incident wave
wave [15].
[15].The
Theradius
radiusofofthe
thearray
array was
wassetset
totor =r =3030cm.
cm.TheTheSNR
SNR
was set to 30 dB and 40 dB. The red round and square symbols denote
was set to 30 dB and 40 dB. The red round and square symbols denote the measured results. The the measured results. The black
curves
black represent the analytical
curves represent results from
the analytical Monte
results fromCarlo
Monte simulations. Figure 23Figure
Carlo simulations. shows23 that therethat
shows is
good
thereagreement between the
is good agreement measured
between theand calculated
measured andresults. Furthermore,
calculated the proposed the
results. Furthermore, antenna can
proposed
maintain
antenna acan constant
maintain transmission rate over 360 rate
a constant transmission degrees
overin azimuth.
360 degreesInterestingly, the red round
in azimuth. Interestingly, and
the red
square ◦ and 337.5◦ are situated at the borders between two
roundsymbols
and square corresponding to angles of 22.5
symbols corresponding to angles of 22.5° and 337.5° are situated at the borders
adjacent
betweensectors, as illustrated
two adjacent sectors, in Figure 4. These
as illustrated measured
in Figure 4. Thesedata do notdata
measured show doappreciable changes,
not show appreciable
compared ◦ ◦
changes, with 0 andwith
compared 360 0° , indicating
and 360°, that the proposed
indicating that theantenna
proposed works well,works
antenna even on the even
well, borders on of
the
the sectors,
borders of owing to theowing
the sectors, good beamto thesteering
good beam ability of theability
steering array. of the array.
Micromachines 2020, 11, x 22 of 25

Micromachines
Micromachines 2020, 11, x871 2222of
of 25
25
400
400
350

[bits/s/Hz]
350
300

[bits/s/Hz]
300
250
Capacity
Capacity 250
200
200
150
Channel

150
100 Cal. (SNR = 40dB)
Channel

r = 30cm OTA (SNR = 40dB)

100 Cal. (SNR = 40dB)
50  = 30deg Cal. (SNR = 30dB)
f r==2GHz
30cm OTA (SNR = 40dB)
OTA (SNR = 30dB)
50
0 σ = 30deg Cal. (SNR = 30dB)
0 45 90 f135 180
= 2GHz 225
OTA270(SNR 315 360
= 30dB)
0 Angle of Incident Wave  [deg]
0 45 90 135 180 225 270 315 360
Angle
Figure 23. Channel capacity of a 32 × of Incident
32 MIMO Wave
array φ [deg]
antenna measured by OTA testing.

Figure Figure 23. Channel capacity of a 32 × 32 MIMO array antenna measured by OTA testing.
In Section 5.3,23.the
Channel capacity
channel gain of anda 32 × 32 MIMOwere
correlation arrayobtained
antenna measured by OTA testing.
from a theoretical investigation
using In
Monte Carlo
Section 5.3, simulation,
5.3, thethe channel as
channel gainshown
gain and in Figure
and correlation 16.
correlation were In this section,
were obtained
obtained from to investigate
from aa theoreticalthe mechanism
In Section theoretical investigation
investigation
for achieving
using Monte a largesimulation,
channel capacity in an empirical way, we tooktoaninvestigate
experimental approach by
using Monte Carlo
Carlo simulation,as asshown
shownininFigure
Figure16.16.InInthis section,
this section, to investigatethethe
mechanism
mechanism for
means of
achieving OTA
a testing
large channelto evaluate
capacity the
in channel
an gain
empirical and
way, correlation.
we took an experimental approach by means of
for achieving a large channel capacity in an empirical way, we took an experimental approach by
OTA Figure
testing24toshows
evaluate thethe
channel
channel gain and
gain andcorrelation
correlation. coefficient of each subarray of the 32 × 32
means of OTA testing to evaluate the channel gain and correlation.
MIMOFigure
array as shows
a function channel
of the number of correlation
the subarray [15]. Theofblack curves indicate
of the the analytical
Figure 24 24 showsthe the channel gain and
gain and correlation coefficient
coefficient each
of subarray
each subarray of ×
32 the3232
MIMO
× 32
results
array using
as a Monte
function Carlo
of the simulation,
number of while
the the
subarray red
[15].curves
The indicate
black curvesthe measured
indicate the results
analyticalthrough
results
MIMO array as a function of the number of the subarray [15]. The black curves indicate the analytical
OTA
usingtesting,
Monte depicted in Figurewhile
21. The theblue and redindicate
subarrays, indicated results
by the through
inset in Figure 24a,
results usingCarlo
Monte simulation,
Carlo simulation, red curves
while the red curves theindicate
measured the measured OTAthrough
results testing,
represent
depicted that these21.
in Figure subarrays.
The blueSubarrays
and 1–4 and Subarrays
red subarrays, indicated 17–20
by the are located
inset inby in 24a,
Figure the forward
represent and
that
OTA testing, depicted in Figure 21. The blue and red subarrays, indicated the inset in Figure 24a,
backward
these directions
subarrays. with
Subarrays respect
1–4 andto the incident
Subarrays wave.
17–20 The
are blue
located and
in thered arrows
forward included
and backwardin the upper
directions
represent that these subarrays. Subarrays 1–4 and Subarrays 17–20 are located in the forward and
part of Figure to
with respect
24atheindicate
incident
thewave.
locations of Subarrays 1–4 and included
Subarraysin17–20 on thepart
lateralFigure
axis. 24a
backward directions with respect toThe
the blue andwave.
incident red arrows
The blue and red the upper
arrows includedof in the upper
indicate
part the locations
of Figure 24a indicateof Subarrays 1–4 and
the locations Subarrays1–4
of Subarrays 17–20andonSubarrays
the lateral17–20
axis. on the lateral axis.

10 1
Subarrays 1-4 r = 30cm
Absolute of Complex Correlation

9 r = 30cm
Average of Channel Gain [dB]

0.9
8 SNR = 30dB SNR = 30dB
1
710 Subarrays 1-4  r==0deg
30cm 0.8  r==0deg
30cm
Absolute of Complex Correlation

69 SNR 0.9
Average of Channel Gain [dB]

8 = 30deg = 30dB 0.7 SNR

= 30deg
= 30dB
5
47 Subarrays φ17-20 = 0deg 0.8 φ = 0deg
0.6
36 σ = 30deg 0.7 σ = 30deg
25 0.5
14 9-12
Subarrays 17-20 0.6
0 3 13-16 5-8 0.4
-1 2 0.5
9-12 1-4 2
0.3
-2 117-20 0.4
-3 0 13-16 5-8 0.2
-4-1 17-20 25-28
21-24 29-32
1-4 2σ
Incident OTA Testing 0.3 OTA Testing
0.1
-5-2
-6-3 21-24
Wave Monte Carlo 0.2 Monte Carlo
29-32 0
-41 2 3 4 525-28 Incident
12 13 14 15 16 17 18 19 20 21 22 23 24 25OTA
26 27 28Testing OTA28293031
Testing32
6 7 8 9 10 11 29 30 31 32
0.11 2 3 4 5 6 7 8 9 1011121314 15161718 1920212223 24252627
-5 Wave of Subarray Monte Carlo Number of Subarray Monte Carlo
-6 Number
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 1 2 3 4 5 6 7 8 9 1011121314 15161718 1920212223 24252627 28293031 32

Number
(a) of Subarray Number
(b) of Subarray
Figure
Figure24.
24.Channel
Channel gain andcorrelation
(a) and correlationofof a 32
a 32 × 32× MIMO
32 MIMO
arrayarray measured
measured bytesting.
by (b)
OTA OTA testing. (a)
(a) Channel
Channel
gain. (b)gain. (b) Correlation.
Correlation.
Figure 24. Channel gain and correlation of a 32 × 32 MIMO array measured by OTA testing. (a)
Channel
Figure24a
Figure gain.
24a (b) Correlation.
shows
shows thatSubarrays
that Subarrays1–4,
1–4,which
whichfacefacethe
theincident
incidentwave,
wave,provide
providea alarge
largechannel
channelgain
gain
ofofmore
morethan
than 2.15 dB,
dB,meaning
meaningthat
thatthese subarrays
these operate
subarrays more
operate effectively
more thanthan
effectively a conventional dipole
a conventional
Figure
antenna. 24a shows
On theOn
other that Subarrays 1–4, which face the incident wave, provide a large channel
to less gain
dipole antenna. the hand, for Subarrays
other hand, 17–20, 17–20,
for Subarrays the channel gain is significantly
the channel reducedreduced
gain is significantly than
to
of more than 2.15 dB, meaning that these subarrays operate more effectively than a conventional
dipole antenna. On the other hand, for Subarrays 17–20, the channel gain is significantly reduced to
Micromachines 2020, 11, 871 23 of 25

0 dB. This implies that these subarrays do not contribute to the formation of subchannels available for
MIMO transmission. It is worth noting that, in Figure 24a, there is an appreciable discrepancy between
the OTA measurements and the analytical outcomes. This is due to the fact that Subarrays 1–4, 5–8,
and 29–32, located in the forward direction with respect to the incident wave, receive strong signals
from the scatterers because of the small distance between these subarrays and the scatterers emitting
a cluster wave, as shown in Figure 21a. In contrast, Subarrays 13–16, 17–20, and 21–24, located at
the back with respect to the direction of the incident wave, receive weak signals because of the large
distance between these subarrays and the scatterers.
Figure 24b shows that, in the OTA measurements, the correlation coefficient increases in a periodic
manner, which is not observed in the analytical outcome from the Monte Carlo simulations. A possible
cause of this phenomenon is the insufficient orthogonality between the incident waves coming from the
scatterers, because, in the current experiment, the number of scatterers in the emulator is 14, which is
considerably smaller than the number of subarrays, i.e., 32. Nevertheless, it can be seen from Figure 24b
that except for the correlation between those particular subarray numbers showing a periodic increase,
the correlation coefficients are less than 0.5 over the entire array because of the orthogonal alignment
between the MIMO array and the incident waves.

7. Conclusions
This paper presents a 32 × 32 MIMO antenna system with the ability to perform full-azimuth
beam steering. Analyses revealed that the proposed antenna system can provide a maximum channel
capacity of 277 bits/s/Hz at an SNR of 30 dB when the radius of the array is 30 cm, which is equivalent
to 27.7 Gbps with a bandwidth of 100 MHz. It was clarified that this remarkably high channel capacity
is due to two significant factors; the improved directivity due to the optimum in-phase excitation
and the low correlation between the subarrays due to the orthogonal alignment of the array with
respect to the incident waves. We further performed an experimental validation of the channel capacity
using over-the-air (OTA) testing. The results show that the proposed antenna can maintain a constant
channel capacity over 360 degrees in azimuth.
The developed technology can be applied to the achievement of a 64 × 64 MIMO array [29] and
a 128 × 128 MIMO array [18] in pursuit of much larger transmission rates in the 5 GHz frequency
band. Thereby, the daisy chain array structure proposed in this paper provides a strong solution for
realizing a much higher order MIMO array antenna at higher frequencies, such as in the 5.9 GHz band,
where the emerging vehicle-to-everything (V2X) and cellular-V2X (C-V2X) technologies are anticipated
to be operated for future connected cars [30–32].

Author Contributions: Conceptualization, K.O.; Funding acquisition, K.H.; Investigation, K.H.; Methodology,
K.H.; Validation, T.F.; Writing—original draft, K.H. All authors provided critical feedback and helped shape the
research, analysis, and manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This research and development work was supported by the MIC/SCOPE #205005002.
Acknowledgments: The authors acknowledge the funding from the MIC/SCOPE #205005002.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Rusek, F.; Persson, D.; Lau, B.K.; Larsson, E.G.; Marzetta, T.L.; Edfors, O.; Tufvesson, F. Scaling Up MIMO:
Opportunities and challenges with very large MIMO. IEEE Signal. Process. Mag. 2012, 30, 40–60. [CrossRef]
2. Larsson, E.G.; Edfors, O.; Tufvesson, F.; Marzetta, T.L. Massive MIMO for Next Generation Wireless Systems.
IEEE Commun. Mag. 2014, 52, 186–195. [CrossRef]
3. Kibaroglu, K.; Sayginer, M.; Phelps, T.; Rebeiz, G.M. A 64-Element 28-GHz Phased-Array Transceiver
With 52-dBm EIRP and 8–12-Gb/s 5G Link at 300 Meters without Any Calibration. IEEE Trans. Microw.
Theory Technol. 2018, 66, 5796–5811. [CrossRef]
Micromachines 2020, 11, 871 24 of 25

4. Yamaguchi, S.; Nakamizo, H.; Shinjo, S.; Tsutsumi, K.; Fukasawa, T.; Miyashita, H. Development of active
phased array antenna for high SHF wideband massive MIMO in 5G. In Proceedings of the 2017 IEEE
International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting,
San Diego, CA, USA, 9–14 July 2017; pp. 1463–1464.
5. Mao, C.-X.; Gao, S.; Wang, Y. Broadband High-Gain Beam-Scanning Antenna Array for Millimeter-Wave
Applications. IEEE Trans. Antennas Propag. 2017, 65, 4864–4868. [CrossRef]
6. Al-Tarifi, M.A.; Faouri, Y.S.; Sharawi, M.S. A printed 16 ports massive MIMO antenna system with directive
port beams. In Proceedings of the 2016 IEEE 5th Asia-Pacific Conference on Antennas and Propagation
(APCAP), Kaohsiung, Taiwan, 26–29 July 2016; pp. 125–126.
7. Kataoka, R.; Nishimori, K.; Tran, N.; Imai, T.; Makino, H. Interference Reduction Characteristics by Circular
Array Based Massive MIMO in a Real Microcell Environment. IEICE Trans. Commun. 2015, 1447–1455.
[CrossRef]
8. Schneider, C.; Thoma, R.S. Empirical Study of Higher Order MIMO Capacity at 2.53 GHz in Urban Macro Cell.
In Proceedings of the 2013 7th European Conference on Antennas and Propagation (EuCAP), Gothenburg,
Sweden, 8–12 April 2013; pp. 477–481.
9. Mumtaz, S.; Huq, K.; Ashraf, M.; Rodriguez, J.; Monterio, V.; Politisand, C. Cognitive Vehicular
Communication for 5G. IEEE Commun. Mag. 2015, 53, 109–117. [CrossRef]
10. Chong, C.; Watanabe, F.; Kitao, K.; Imai, T.; Inamura, H. Evolution Trends of Wireless MIMO Channel
Modeling towards IMT-Advanced. IEICE Trans. Commun. 2009, 2773–2788. [CrossRef]
11. Kuchar, A.; Rossi, J.; Bonek, E. Directional macro-cell channel characterization from urban measurements.
IEEE Trans. Antennas Propag. 2000, 48, 137–146. [CrossRef]
12. Honda, K.; Fukushima, T.; Ogawa, K. Dual-Circular 8 × 8 MIMO Array with Synchronized Beam Scan for
Over-Gbps Inter-Vehicle Communications. In Proceedings of the Progress in Electromagnetics Research
Symposium 2018 (PIERS2018), Toyama, Japan, 1–4 August 2018; pp. 1989–1994.
13. Fukushima, T.; Honda, K.; Ogawa, K. 140 bps/Hz 16 × 16 MIMO Whole Azimuth Beam Steering Array
for Connected Car Applications. In Proceedings of the 2018 Asia-Pacific Microwave Conference (APMC),
Kyoto, Japan, 6–9 November 2018; pp. 503–505.
14. Narukawa, N.; Fukushima, T.; Honda, K.; Ogawa, K. Daisy Chain MIMO Antenna for Whole Azimuth
Tens-of-Gigabit Connected Cars. In Proceedings of the 2019 URSI International Symposium on Electromagnetic
Theory (EMTS), San Diego, CA, USA, 27–31 May 2019; pp. 1–4.
15. Fukushima, T.; Narukawa, N.; Honda, K.; Ogawa, K. Over-the-Air Testing of a 32 × 32 Daisy Chain MIMO
Antenna. In Proceedings of the Progress in Electromagnetics Research Symposium 2019 (PIERS2019),
Roma, Italy, 17–20 June 2019; pp. 824–830.
16. Honda, K.; Iwamoto, D.; Ogawa, K. Angle of Arrival Estimation Embedded in a Circular Phased Array
4 × 4 MIMO Antenna. In Proceedings of the 2017 IEEE Asia Pacific Microwave Conference (APMC),
Kuala Lumpur, Malaysia, 13–16 November 2017; pp. 93–96.
17. Iwamoto, D.; Narukawa, N.; Honda, K.; Ogawa, K. Experiments on Interferometric Angle of Arrival
Estimation Using a Simple Weight Network. In Proceedings of the 2018 International Symposium on
Antennas and Propagation (ISAP), Busan, Korea, 23–26 October 2018; pp. 351–352.
18. Narukawa, N.; Fukushima, T.; Honda, K.; Ogawa, K. Daisy Chain MIMO Antenna: A Big Challenge to
Full-Azimuth 100 Gbps Capacity. In Proceedings of the Progress in Electromagnetics Research Symposium
2019 (PIERS2019), Roma, Italy, 17–20 June 2019; pp. 817–823.
19. Honda, K.; Fukushima, T.; Ogawa, K. Development of a Circular Phased Array 4 × 4 MIMO Antenna for
Ad-hoc Connected Car System. In Proceedings of the 2017 IEEE Asia Pacific Microwave Conference (APMC),
Kuala Lumpur, Malaysia, 13–16 November 2017; pp. 795–798.
20. Buehrer, R.M. The impact of angular energy distribution on spatial correlation. In Proceedings of the IEEE
56th Vehicular Technology Conference, Vancouver, BC, Canada, 24–28 September 2002; pp. 1173–1177.
21. 3GPP-3GPP2 Spatial Channel Model Ad-hoc Group. Spatial Channel, Model for MIMO Systems. 3GPP and
3GPP2 Technical Report. 2003. Available online: http://www.3gpp.org (accessed on 1 August 2020).
22. Ogawa, K.; Yamamoto, A.; Takada, J. Multipath Performance of Handset Adaptive Array Antennas in the
Vicinity of a Human Operator. IEEE Trans. Antennas Propag. 2005, 53, 2422–2436. [CrossRef]
Micromachines 2020, 11, 871 25 of 25

23. Ogawa, K.; Iwai, H.; Yamamoto, A.; Takada, J. Channel Capacity of a Handset MIMO Antenna Influenced by
the Effects of 3D Angular Spectrum, Polarization, and Operator. In Proceedings of the 2006 IEEE Antennas
and Propagation Society International Symposium, Albuquerque, NM, USA, 9–14 July 2006; pp. 153–156.
24. Chuah, C.-N.; Kahn, J.M.; Tse, D.N.C. Capacity Scaling in MIMO Wireless Systems under Correlated Fading.
IEEE Trans. Inf. Theory 2002, 48, 637–650. [CrossRef]
25. Taga, T. Analysis for mean effective gain of mobile antennas in land mobile radio environments. IEEE Trans.
Veh. Technol. 1990, 39, 117–131. [CrossRef]
26. Sakata, T.; Yamamoto, A.; Ogawa, K.; Iwai, H.; Takada, J.; Sakaguchi, K. A Spatial Fading Emulator for
Evaluation of MIMO Antennas in a Cluster Environment. IEICE Trans. Commun. 2014, 2127–2135. [CrossRef]
27. Fukushima, T.; Honda, K.; Ogawa, K. Directivity Measurement of Circular Phased Array 4 × 4 MIMO
Antenna. In Proceedings of the Progress in Electromagnetics Research Symposium 2018 (PIERS2018),
Toyama, Japan, 1–4 August 2018; pp. 1984–1988.
28. Imai, T.; Okano, Y.; Koshino, K.; Saito, K.; Miura, S. Theoretical Analysis of Adequate Number of
Probe Antennas in Spatial Channel Emulator for MIMO Performance Evaluation of Mobile Terminals.
In Proceedings of the Fourth European Conference on Antennas and Propagation (EuCAP), Barcelona, Spain,
12–16 April 2010; pp. 1–5.
29. Narukawa, N.; Fukushima, T.; Honda, K.; Ogawa, K. 64 × 64 MIMO Antenna Arranged in a Daisy Chain Array
Structure at 50 Gbps Capacity. In Proceedings of the 2019 URSI International Symposium on Electromagnetic
Theory (EMTS), San Diego, CA, USA, 27–31 May 2019; pp. 1–4.
30. 3G Partnership Project. Study on LTE-Based V2X Services; TR 36.885 v14.0.0; 3GPP: Sophia Antipolis,
France, 2016.
31. Chen, S.; Hu, J.; Shi, Y.; Zhao, L. LTE-V: A TD-LTE-Based V2X Solution for Future Vehicular Network.
IEEE Internet Things J. 2016, 3, 997–1005. [CrossRef]
32. MacHardy, Z.; Khan, A.; Obana, K.; Iwashina, S. V2X Access Technologies: Regulation, Research,
and Remaining Challenges. IEEE Commun. Surv. Tutor. 2018, 20, 1858–1877. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).