ACES JOURNAL, Vol. 36, No.

6, June 2021 762

Compact 2x2 and 4x4 MIMO Antenna Systems for 5G Automotive

Applications

Mohamed O. Khalifa, Ahmad M. Yacoub, and Daniel N. Aloi

Department of Electrical Engineering
Oakland University, Rochester, Michigan 48309, USA
MKhalifa@Oakland.edu, Ahmadyacoub@Oakland.edu, Aloi@Oakland.edu

Abstract ─ In this paper, three Vehicular multiple-input independent MIMO antennas radiation patterns are, for
multiple-output (MIMO) 5G antenna systems have been example, for a 2X2 MIMO system, if one antenna is
constructed from using a newly developed 5G cellular vertically polarized and the other one is horizontally
branched Monopole element are presented. The MIMO polarized then the ECC value will be 0. There are 2 ways
systems operates in the 5G frequency bands (617MHz- to calculate the ECC in a MIMO system [4], the first
5GHz) with a compact structure that allows for up to four method used S parameters to find ECC and it assumes
elements to be integrated in the same Sharkfin. The 3 lossless/60% or more efficient antennas which is
configurations of MIMO systems have been simulated unrealistic. Whereas the second method (which is more
using HFSS, measured on a 1-meter ground plane accurate) used throughout this paper utilizes radiation
(GND), then measured on a vehicle roof and the patterns of individual antennas to calculate the ECC
individual antenna parameters in terms of reflection of MIMO system. Modern vehicles are equipped with
coefficient and efficiency have captured. The MIMO multiple wireless services such as Global Navigation
antenna systems performance in terms of passive Satellite Systems (GNSS), Remote Keyless Entry
isolation, combined radiation pattern, envelope (RKE), Satellite Digital Audio radio service (SDARS),
correlation coefficient (ECC), and diversity gain (DG) etc. [5]-[7]. Each of these services requires dedicated
have been reported and discussed. antennas and it is unfeasible to distribute them all over
the vehicle and consequently they all integrated in a
Index Terms ─ Automotive antennas, correlation single package (Sharkfin). Since the available space
coefficient, 5G, 2x2 and 4x4 MIMO systems. with the Sharkfin for each antenna is much smaller than
the wavelengths at the which the antenna expected to
I. INTRODUCTION operate, issues such as antenna size and bandwidth
With the expansion of the cellular systems being limitations as well as passive port isolation between
integrated in cars to support the connected vehicle effort, different elements within the Sharkfin will take place.
a need for multiple antennas support wireless service has Furthermore, the Sharkfin size limitation will also
emerged. The Multiple Input Multiple Output (MIMO) impact the port isolation and correlation of MIMO
system consists of two antennas or more that receive or antennas and impose some challenges to design an
transmit multiple layers of orthogonal data streams from antenna system that satisfies the diversity requirements
cellular base stations which allow for increased channel [8],[9].
capacity, data rate and the total throughput of the system In [10], [11], 2X2 MIMO systems based on
without increasing the operating frequency band or Monopole and PIFA elements respectively have been
the transmit power [1]. To this date, 2X2 MIMO introduced, however the bandwidth of operation is
configuration with de-correlated antennas that receive very small (700-925MHz). The 2X2 MIMO systems
two data streams is being used to realize downlink bandwidth have been increased to cover from 790MHz
reception in modern vehicles [2]. The performance of the to 3GHz with reasonable antenna dimensions and an
MIMO system is highly dependable on the efficient ECC of less than 0.3 and 0.05 in [12] and [13]
design of the MIMO antennas that should low correlation respectively but the bandwidth of operation doesn’t
between them and a high total antenna efficiency [3]. cover 5G frequencies (617MHz-5GHz). In [14]-[18], the
An important characteristic of the communication authors have developed 2X2 MIMO structures that are
systems is the ECC between received signals of the constructed from either Monopoles or PIFA elements
antennas that construct the MIMO system. Small ECC with a less than 0.5 ECC however, these MIMO systems
values are crucial to increase transmission capacity as are only covering LTE (698MHz-3GHz) frequency
well as to improve the multipath fading. ECC tells how bands and not 5G frequencies. A much broader

Submitted On: May 24, 2021 https://doi.org/10.47037/2020.ACES.J.360619

Accepted On: July 6, 2021 1054-4887 © ACES
763 ACES JOURNAL, Vol. 36, No. 6, June 2021

bandwidth (700MHz-6GHz) with less than 0.16 of where 𝐸𝜃𝑖 and 𝐸𝜃𝑗 are the values of electric field in the
ECC (using first method in [4]) is achieved by authors in theta axis while 𝐸∅𝑖 and 𝐸∅𝑗 are the values of electric
[19] but it doesn’t include the B71 band (617MHz- field in phi axis. XPR is the cross-polarization
698MHz) as well as it comes at an increased volume discrimination factor tells the difference between
(70x70x29𝑚𝑚3 ) which makes it impossible to fit in incident electromagnetic wave vertical and horizontal
production Sharkfins. Finally, a 4X4 MIMO system is polarization. 𝑃𝜃 and 𝑃∅ are the theta and phi power
presented in [20], However, B71 frequency band is not densities. Equation (1) can be simplified by setting XPR
included (which will introduce a big challenge trying to = 1 assuming uniform power densities.
fit the antenna in a sharkfin), there are no information The other MIMO system performance metric is the
about system volume, and also the work has not been DG which is defined as the quantified improvement in
supported by either ground plane or vehicle measurements signal-to-noise ratio (SNR) by the receiving signals from
data. Several studies outside the automotive industry the MIMO antennas and usually calculated in dB. The
have targeted decoupling antennas in a MIMO system DG can be calculated as in [28]:
packaged in a small volume with various techniques like 𝐷𝐺 = 𝐷𝐺0 . 𝐷𝐹. 𝐾, (2)
the use of electromagnetic band gap (EBG) structures
where 𝐷𝐺0 is the ideal case diversity gain which is
[21], Ceramic Superstrate-Based Decoupling Method
[22], and metamaterials [23]-[27]. The automotive 10dB. 𝐷𝐹 is the degradation factor which shows how
MIMO systems presented in this work do not require a much the ECC impacts DG and is calculated as:
special decoupling mechanism. In fact, decoupling of √(1 − 𝜌). 𝐾 represents the ratio of the mean effective
antennas is obtained by mainly using spatial and pattern gain (MEG) between the MIMO antenna elements (𝐾 =
diversity at the low and high band respectively which 𝑀𝐸𝐺𝑖 ⁄𝑀𝐸𝐺𝑗 ). 𝑀𝐸𝐺 is the effective gain ratio at the
puts more burden on compactness and performance of antenna element, in other words 𝑀𝐸𝐺 is the received to
building block antenna element. incident power ratio at the element. (K∼=1) condition
This work introduces three novel MIMO structures; should be satisfied for received signals by MIMO
configuration I of a 2x2 MIMO antenna system, systems assuming good channel characteristics.
configuration II of a 2x2 MIMO antenna system, and a
4x4 MIMO antenna system. The three structures are III. COMPARATIVE STUDY OF THE
mainly developed for automotive industry to cover 5G PROPOSED MIMO SYSTEMS
cellular frequencies (617MHz-5GHz), fit in a car roof In this section, three different MIMO antennas
sharkfin, and have an inherited GNSS frequencies configurations are being studied. The building block for
rejection which makes it easy to integrate with other each configuration is the branched Monopole element in
antennas within a sharkfin. In this work, HFSS has been Fig. 1. The MIMO configurations are then simulated and
used to simulate the 3 MIMO systems configurations, measured on a 1-meter ground plane and on a vehicle’s
then the systems have been measured inside and roof inside an anechoic chamber. The obtained data from
anechoic chamber on ground plane and on a vehicle roof. simulations and chamber measurements are directly
The antennas important parameters such as of reflection reported whereas ECC and DG results are generated
coefficient, isolation, total efficiency, radiation pattern using Octave software for each MIMO configuration in
have been captured. Then an Octave code is used to this section. Table 1 shows general design guidelines:
calculate the ECC and DG from the captured data.
The work in this paper is organized in two sections: Table 1: Design guidelines
Section II explains the theory about correlation coefficient Parameter Value
and diversity gain calculation equations and it also shows Polarization Vertical Linear
the design goals. Section III presents the proposed 2x2 Polarization (VLP)
MIMO configurations, the novel 4x4 MIMO system, and Reflection Coefficient -5.4dB (3.3VSWR) at
finally a comparison between proposed systems and 5G bands
available work in literature.
Avg. Total Efficiency 60%
Isolation Minimum 10dB across
II. MIMO ANTENNA SYSTEMS 5G bands
PERFORMANCE METRICS ECC Less than 0.5
All the most important metrics in MIMO antenna
systems are ECC and DG. The ECC can be related to the
A. Configuration I of a 2x2 monopole-based MIMO
electric field radiation pattern through the equation in [4]
system
as below:
2𝜋 𝜋
2 Figure 1 shows the Monopole element that will be
∗ ∗
∫0 ∫0 (𝑋𝑃𝑅 . 𝐸𝜃𝑖 . 𝐸𝜃𝑗 . 𝑃𝜃 + 𝑋𝑃𝑅 . 𝐸𝜙𝑖 . 𝐸𝜙𝑗 . 𝑃𝜙 ) sin(𝜃) 𝑑𝜃𝑑𝜙
𝜌𝑒,𝑖𝑗 = || || , (1) used construct high order MIMO structures. The antenna
2𝜋 𝜋 ∗
√∏𝑘=𝑖,𝑗 ∫0 ∫0 (𝑋𝑃𝑅 . 𝐸𝜃𝑘 . 𝐸𝜃𝑘 ∗
. 𝑃𝜃 + 𝑋𝑃𝑅 . 𝐸𝜑𝑘 . 𝐸𝜑𝑘 . 𝑃𝜙 ) sin(𝜃)𝑑𝜃𝑑𝜙 functions in the 5G frequency bands (617MHz-5GHz).
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 764

Multiple current paths have been realized by adding 2 The two Monopole elements were cut from a metal
arms to the antenna covering 5G frequencies of interest. sheet and placed in such a way that will result in an
Band (617 MHz – 960 MHz) is mainly covered by Arm omnidirectional combined radiation pattern as well as
1 currents whereas band (1.71-5 GHz) frequencies are a minimal ECC value. The system is simulated then
radiated by Arm 2 as in Fig. 1. To obtain omnidirectional measured on 1-meter GND and on a car’s roof with a
radiation pattern, high efficiencies, and good matching port-to-port distance of 135mm (which is higher than
performance, Arm1 and Arm2 of the antenna are loaded 𝝀𝑳𝒐𝒘𝒆𝒔𝒕_𝒇𝒓𝒆𝒒
𝝀 𝝀
= 114mm on FR4 PCB with εr = 4.4) between
𝟐∗√𝜺𝒓
at 𝑳𝑩 and 𝑯𝑩 distances from ground plane respectively
𝟒 𝟒 Monopole elements. Figure 2 shows Configuration I of a
where 𝝀𝑳𝑩 and 𝝀𝑯𝑩 are the mid-low and mid-high bands 2x2 MIMO systems on a car roof.
wavelengths respectively. The geometrical values of the The performance of Antenna1 (Ant1) and Antenna2
building block antenna are shown in Table 2. (Ant2) of this MIMO system has been reported in terms
of reflection coefficient and isolation as in Fig. 3. Both
L antennas show an agreement between simulation and
GND measurement. Good reflection coefficient values
Ls
Ha1
have been observed of a worse of -5.6dB and -6.4dB at
617MHz of Ant1 and Ant2 respectively with reasonable
Lg1 Lg2 GNSS bands (1560MHz-1190MHz) rejection. Figure 3
also shows a good isolation between Ant1 and Ant2 in
H

La2 this configuration of a worse case 12dB (expressed as

Ha2 -12 dB in S21 format).
Next, the total efficiencies of Ant1 and Ant2 have
been captured after a successful placement of the MIMO
system on 1-meter GND and then on a car roof and
the results are shown in Fig. 4. Both antennas exhibit
Fig. 1. Building block antenna with side and front views, higher efficiencies when placed on GND compared to
arms structure, and antenna dimensions. placement on a car roof. It can be noticed that both
antennas measured on GND have an average total
Table 2: Values of the geometrical parameters of MIMO efficiency of 79.8% across all 5G frequency bands while
systems building block antenna the average total efficiencies decrease to 74% and
Parameter Value Parameter Value 71% when measured on a car roof for Ant1 and Ant2
(mm) (mm) respectively.
H 60 Lg1 9 Simulation, GND measurement, and vehicle
L 38.5 Lg2 2 measurement of a combined MIMO system radiation
W 14.9 Ls 2.6 pattern sample have been presented in Fig. 5. The sample
Ha1 30 La2 39.8 represents a Gain theta horizontal cut at 80 degrees of
Ha2 29.3 theta and four frequencies namely 617MHz, 1900MHz,
3900MHz, and 5000MHz to provide a good idea about
Ant1
Ant1 the system performance. The combined MIMO system
Ant2
Ant2 pattern is obtained by measuring each antenna when the
Ant1 other antenna is loaded by a 50Ohm terminator and
Ant2
then combine the resultant individual antenna patterns
selecting the maximum values of Gain theta between
Ant1 and Ant2 measurements for a specific phi-frequency
pair at 80 degree of theta. Finally, the average gain of the
MIMO system combined radiation patterns measured on
car roof is found to be -1.14dBi, 3.15dBi, 1.81dBi, and
Fig. 2. Configuration I of a 2x2 MIMO system placement 2.84dBi at frequencies 617MHz, 1900MHz, 3900MHz,
on vehicle roof and simulation setup. and 5000MHz, respectively.
765 ACES JOURNAL, Vol. 36, No. 6, June 2021

0 1
Simulated_Ant1
Measured_Ant1
Reflection Coefficient (dB) -10 0.8

Efficiency (%)
-20 0.6

-30 0.4

-40 0.2 GND_Ant1

Vehicle_Ant1
GND_Ant2
-50 Vehicle_Ant2
0 1000 2000 3000 4000 5000 6000 0
600 650 700 750 800 850 900 950
Frequency (MHz) Frequency (MHz)
(a) (a)
0 1
Simulated_Ant2
Measured_Ant2
-10 0.8
Reflection Coefficient (dB)

Efficiency (%)
-20 0.6

-30 0.4

-40 0.2 GND_Ant1

Vehicle_Ant1
GND_Ant2
Vehicle_Ant2
-50 0
0 1000 2000 3000 4000 5000 6000 1800 2000 2200 2400 2600
Frequency (MHz) Frequency (MHz)

(b) (b)
0 1
Simulated_S21
Measured_S21
-10 0.8
Efficiency (%)

-20 0.6
S21 (dB)

-30 0.4

-40 0.2 GND_Ant1

Vehicle_Ant1
GND_Ant2
Vehicle_Ant2
-50 0
0 1000 2000 3000 4000 5000 6000 3400 3600 3800 4000 4200 4400 4600 4800 5000
Frequency (MHz) Frequency (MHz)

(c) (c)

Fig. 3. Configuration I of a 2x2 MIMO system simulated Fig. 4. Antenna efficiency for Configuration I MIMO
and measured: (a) Antenna1 S11 in dB, (b) Antenna2 systems Monopoles measured on GND and on vehicle
S11 in dB, and (c) S21 between Antenna1 and Antenna2 roof for frequency ranges: (a) 617-960MHz, (b) 1710-
in dB. 2690MHz, and (c) 3400-5000MHz.
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 766

90 90
10 120 10 60
120 60
5 5
0 0
-5 -5
150 30 150 30
-10 -10
-15 -15
-20 -20
-25 -25
180 -30 0 180 -30 0

210 330 210 330

240 300 240 300

270 270

(a) (b)
90 90
10 10
120 60 120 60
5 5
0 0
-5 -5
150 30 150 30
-10 -10
-15 -15
-20 -20
-25 -25
180 -30 0 180 -30 0

210 330 210 330

240 300 240 300

270 270

(c) (d)

Fig. 5. Combined radiation pattern of simulation, GND measurement, and vehicle measurement in (dBi) at theta = 80
deg. for frequencies: (a) 617 MHz, (b) 1900 MHz, (c) 3900 MHz, and (d) 5000 MHz.

ECC and DG on GND and on car roof for 0.5

Measured_GND
configuration I of a 2x2 MIMO systems are depicted in Measured_Vehicle
Fig. 6 (a) and Fig. 6 (b) respectively. The two figures 0.4
suggests that higher values of ECC and consequently
lower values of DG occurs at low frequencies (i.e., 0.3
617MHz) because the wavelength is big which leads to
ECC

more correlation between the antennas. In this MIMO 0.2

configuration an ECC of better than 0.13 and a DG of
better than 9.92dB have been realized.
0.1

0
0 1000 2000 3000 4000 5000
Frequency (MHz)
(a)
767 ACES JOURNAL, Vol. 36, No. 6, June 2021

10 0
Simulated_Ant1
Measured_Ant1
-10

Reflection Coefficient (dB)

9.95

-20
DG (dB)

9.9

-30

9.85
-40
Measured_GND
Measured_Vehicle
9.8 -50
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000 6000
Frequency (MHz)
Frequency (MHz)
(b) (a)
0
Fig. 6. Configuration I of a 2x2 MIMO system: (a) Simulated_Ant2
envelope correlation coefficient, and (b) diversity gain. Measured_Ant2
-10

Reflectoin Coefficient (dB)

B. Configuration II of a 2x2 monopole-based MIMO
system
-20
Similar to Configuration I in subsection A, two
Monopole elements have been placed on a Printed
Circuit Board (PCB) with a distance of 125mm (from -30
port to port) between them. The monopoles are place on
the back of the roof of a car as in Fig. 7 to allow for an
omnidirectional combined radiation pattern. -40
The performance of Ant1 and Ant2 of this MIMO
system has been reported in terms of reflection coefficient -50
and isolation as in Fig. 8. Both antennas show an 0 1000 2000 3000 4000 5000 6000
agreement between simulation and GND measurement. Frequency (MHz)
Good reflection coefficient values have been observed (b)
from GND measurements of a worse of -7.4dB and 0
-6.4dB at 617MHz of Ant1 and Ant2 respectively Simulated_S21
Measured_S21
with reasonable GNSS bands rejection. Figure 8 also -10
shows a good isolation between Ant1 and Ant2 in this
configuration of a worse case 15dB (expressed as -15 dB
-20
in S21 format).
S21 (dB)

Ant1 -30
Ant2
Ant1

Ant2 -40

-50

-60
0 1000 2000 3000 4000 5000 6000
Frequency (MHz)

Fig. 7. Configuration II of a 2x2 MIMO system placement (c)

on vehicle roof and simulation setup.
Fig. 8. Configuration II of a 2x2 MIMO system
A comparison of on GND and on vehicle simulated and measured reflection coefficient for: (a)
measurements of Ant1 and Ant2 total antenna efficiencies Antenna1, (b) Antenna2 in dB, and (c) S21 between
in this configuration is illustrated in Fig. 9. Antenna1 and Antenna2 in dB.
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 768

1 whereas the vehicle measurement is 5% less namely an

average total efficiency of 71.7%.
ECC and DG on GND and on car roof for
0.8
configuration II of a 2x2 MIMO systems is depicted
in Fig. 10 where an ECC of better than 0.02 and an
Efficiency (%)

0.6 approximately 10dB of DG has been achieved using this

configuration.
0.4
0.1
Measured_GND
0.2 GND_Ant1 Measured_Vehicle
Vehicle_Ant1
GND_Ant2 0.08
Vehicle_Ant2
0
600 650 700 750 800 850 900 950
Frequency (MHz) 0.06

ECC
(a)
1
0.04

0.8
0.02
Efficiency (%)

0.6
0
0 1000 2000 3000 4000 5000
Frequency (MHz)
0.4
(a)
10
0.2 GND_Ant1
Vehicle_Ant1
GND_Ant2
Vehicle_Ant2
0 9.995
1800 2000 2200 2400 2600
Frequency (MHz)
(b)
DG (dB)

 9.99


9.985
Efficiency (%)


Measured_GND
Measured_Vehicle
9.98
0 1000 2000 3000 4000 5000

Frequency (MHz)
(b)
 GND_Ant1
Vehicle_Ant1
GND_Ant2
Vehicle_Ant2
Fig. 10. Configuration II of a 2x2 MIMO system (a) ECC
 and (b) DG.
3400 3600 3800 4000 4200 4400 4600 4800 5000
Frequency (MHz)
(c) Figures 11 (a)-(d) presents combined radiation
patterns of horizontal cuts at elevation 80 degrees
Fig. 9. Total antenna efficiency for Configuration II for frequencies 617MHz, 1900MHz, 3900MHz and
MIMO systems monopoles measured on GND and on 5000MHz utilizing the same method described in
vehicle roof for frequency ranges: (a) 617-960MHz, (b) subsection A.
1710-2690MHz, and (c) 3400-5000MHz. The average gain values recorded from the
combined vehicle radiation patterns are -0.71dBi,
The GND measurement has an average efficiency 3.16dBi, 0.57dBi, and 1.51dBi at frequencies 617MHz,
higher than 76% for both elements across all 5G bands 1900MHz, 3900MHz, and 5000MHz, respectively.
769 ACES JOURNAL, Vol. 36, No. 6, June 2021

90 90
120 10 60 120 10 60
5 5
0 0
-5 -5
150 30 150 30
-10 -10
-15 -15
-20 -20
-25 -25
180 -30 0 180 -30 0

210 330 210 330

240 300 240 300

270 270
(a) (b)
90 90
10 10
120 60 120 60
5 5
0 0
-5 -5
150 30 150 30
-10 -10
-15 -15
-20 -20
-25 -25
180 -30 0 180 -30 0

210 330 210 330

240 300 240 300

270 270

(c) (d)

Fig. 11. Combined radiation pattern of simulation, GND measurement, and car measurement in (dBi) at theta = 80deg.
for frequencies: (a) 617 MHz, (b) 1900 MHz, (c) 3900 MHz, and (d) 5000 MHz.

C. A 4x4 monopole-based MIMO system Ant1

Ant2
In this subsection, four Monopole elements are Ant2
Ant4
Ant1
integrated in the same sharkfin to operate as a 4x4
MIMO system. The building block antenna element
for this configuration is the same antenna used in Ant3
subsections A and B of this section. The four elements Ant3
Ant4
are placed in such a way that, the combined radiation
pattern is omnidirectional with good isolation and
correlation figures between antennas. The system
placement on the car roof and simulation setup are
shown in Fig. 12. The distances between Monopole pairs Fig. 12. A 4x4 MIMO system placement on vehicle roof
in this MIMO systems are listed in Table 3. and simulation setup.
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 770

Table 3. Values of the geometrical parameters of MIMO 0

Simulated_Ant2
systems building block antenna Measured_Ant2
Elements Pairs Distance (mm) -10

Reflection Coefficient (dB)

Ant1-Ant2 77
Ant1-Ant3 141
Ant1-Ant4 180 -20

Ant2-Ant3 100
Ant2-Ant4 118 -30
Ant3-Ant4 70
-40
The simulated reflection coefficient (in dB) for
each element in this configuration is shown in Fig. 13.
The four elements from Ant1 to Ant4 shows good -50
0 1000 2000 3000 4000 5000 6000
matching characteristics across the whole 5G bands
Frequency (MHz)
with a reflection coefficient of less than -5.2dB and a
(b)
reasonable GNSS bands rejection. 0
The isolation in terms of S21 between each pair of Simulated_Ant3
Measured_Ant3
antennas within this 4x4 MIMO system is shown in -10
Fig. 14. In general, the shorter the distance between

Reflection Coefficient (dB)

the antennas, the worse the isolation is. However, the
-20
antenna placement and orientation also contribute to the
overall isolation performance. A worse value of 10dB
-30
of isolation between Ant1-Ant2 and Ant2-Ant4 can
be observed from the GND measurements on the 617-
960MHz band. -40
Using similar approach for combining individual
antennas radiation patterns as in subsection A, the -50
combined radiation patterns at 80 degree of theta for the
four elements are shown is Fig. 15. Higher average gain -60
0 1000 2000 3000 4000 5000 6000
values are observed from the combined vehicle radiation
Frequency (MHz)
patterns measurements compared to the other 2x2
configurations. The average gain is found to be: -0.88dBi, (c)
4.64dBi, 3.14dBi, and 3.74dBi at frequencies 617MHz, 0
Simulated_Ant4
1900MHz, 3600MHz, and 5000MHz, respectively. Measured_Ant4
-10
Reflection Coefficient (dB)

0
Simulated_Ant1
Measured_Ant1
-20
-10
Reflection Coefficient (dB)

-30
-20

-40
-30

-40
-50
0 1000 2000 3000 4000 5000 6000
Frequency (MHz)
-50 (d)
0 1000 2000 3000 4000 5000 6000
Frequency (MHz) Fig. 13. 4x4 MIMO system simulated and measured
(a) reflection coefficient in dB for: (a) Antenna1, (b)
Antenna2, (c) Antenna3, and (4) Antenna4.
771 ACES JOURNAL, Vol. 36, No. 6, June 2021

0 0
Simulated_S21_Ant1&Ant2 Simulated_S21_Ant2&Ant3
Measured_S21_Ant1&Ant2 Measured_S21_Ant2&Ant3
-10
-10

-20
-20
S21 (dB)

S21 (dB)
-30

-30
-40

-50 -40

-60 -50
0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000
Frequency (MHz) Frequency (MHz)
(a) (d)
0 0
Simulated_S21_Ant1&Ant3 Simulated_S21_Ant2&Ant4
Measured_S21_Ant1&Ant3 Measured_S21_Ant2&Ant4

-10 -10

-20 -20
S21 (dB)
S21 (dB)

-30 -30

-40 -40

-50 -50
0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000
Frequency (MHz) Frequency (MHz)
(b) (e)
0 0
Simulated_S21_Ant1&Ant4 Simulated_S21_Ant3&Ant4
Measured_S21_Ant1&Ant4 Measured_S21_Ant3&Ant4
-10
-10

-20
-20
S21 (dB)

-30
S21 (dB)

-30
-40
-40
-50

-60 -50

-70 -60
0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000
Frequency (MHz) Frequency (MHz)
(c) (f)
Fig. 14. 4x4 MIMO system simulated and measured Isolation in dB between: (a) Ant1 and Ant2: (b) Ant1 and Ant3;
(c) Ant1 and Ant4; (d) Ant2 and Ant3; (e) Ant2 and Ant4; (f) Ant3 and Ant4.
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 772

90 90
10 120 10 60
120 60
5 5
0 0
-5 -5
150 30 150 30
-10 -10
-15 -15
-20 -20
-25 -25
180 -30 0 180 -30 0

210 330 210 330

240 300
240 300
270
270
(a) (b)
90 90
120 10 60 120 10 60
5 5
0 0
-5 -5
150 30 150 30
-10 -10
-15 -15
-20 -20
-25 -25
180 -30 0 180 -30 0

210 330 210 330

240 300 240 300

270 270

(c) (d)

Fig. 15. Combined radiation pattern of simulation, GND measurement, and vehicle measurement in (dBi) at theta =
80 deg. for frequencies: (a) 617 MHz, (b) 1900 MHz, (c) 3900 MHz, and (d) 5000 MHz.

The on GND and on car roof measured efficiencies below 0.5 in all the 6 correlation cases with
are reported in Figs. 16 (a), (b), and (c) for the four corresponding DG values of higher than 8.9dB. The
antenna elements in this MIMO system. It can be noticed worst value of ECC (0.45) and DG exists at low
that the GND measurement has an average total frequency bands and it is the same case where worst
efficiency higher than 72.5% for all the four elements passive isolation occurs namely between Ant3 and Ant4.
across the whole 5G frequency bands whereas the Table 4 lists a literature review summary of MIMO
vehicle measurement has a reduced average total antennas systems used in automotive industry. The table
efficiency of slightly higher than 65% for all elements. also compares works in terms type of type of antennas,
The ECC and DG were also calculated in this MIMO bandwidth of operations, systems volume, and method
configuration with the help of equations (1) and (2) using of ECC calculation in use.
Octave script. As Fig. 17 suggests, the ECC is well kept
773 ACES JOURNAL, Vol. 36, No. 6, June 2021

1 1
GND_Ant1 GND_Ant2
Vehicle_Ant1 Vehicle_Ant2
0.8 0.8
Efficiency (%)

Efficiency (%)
0.6 0.6

0.4 0.4

0.2 0.2

0 0
600 650 700 750 800 850 900 950 600 650 700 750 800 850 900 950
Frequency (MHz) Frequency (MHz)
(a) Ant1: 617-960MHz (d) Ant2: 617-960MHz
1 1

0.8 0.8
Efficiency (%)

Efficiency (%)
0.6 0.6

0.4 0.4

0.2 0.2

GND_Ant1 GND_Ant2
Vehicle_Ant1 Vehicle_Ant2
0 0
1800 2000 2200 2400 2600 1800 2000 2200 2400 2600
Frequency (MHz) Frequency (MHz)
(b) Ant1: 1710-2690MHz (e) Ant2: 1710-2690MHz
1 1

0.8 0.8
Efficiency (%)

Efficiency (%)

0.6 0.6

0.4 0.4

0.2 0.2

GND_Ant1 GND_Ant2
Vehicle_Ant1 Vehicle_Ant2
0 0
3400 3600 3800 4000 4200 4400 4600 4800 5000 3400 3600 3800 4000 4200 4400 4600 4800 5000
Frequency (MHz) Frequency (MHz)
(c) Ant1: 3400-5000MHz (f) Ant2: 3400-5000MHz
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 774

1 1
GND_Ant3 GND_Ant4
Vehicle_Ant3 Vehicle_Ant4
0.8 0.8
Efficiency (%)

Efficiency (%)
0.6 0.6

0.4 0.4

0.2 0.2

0 0
600 650 700 750 800 850 900 950 600 650 700 750 800 850 900 950
Frequency (MHz) Frequency (MHz)
(g) Ant3: 617-960MHz (j) Ant4: 617-960MHz
1 1

0.8 0.8
Efficiency (%)

Efficiency (%)
0.6 0.6

0.4 0.4

0.2 0.2

GND_Ant3 GND_Ant4
Vehicle_Ant3 Vehicle_Ant4
0 0
1800 2000 2200 2400 2600 1800 2000 2200 2400 2600
Frequency (MHz) Frequency (MHz)
(h) Ant3: 1710-2690MHz (k) Ant4: 1710-2690MHz
1 1

0.8 0.8
Efficiency (%)
Efficiency (%)

0.6 0.6

0.4 0.4

0.2 0.2

GND_Ant3 GND_Ant4
Vehicle_Ant3 Vehicle_Ant4
0 0
3400 3600 3800 4000 4200 4400 4600 4800 5000 3400 3600 3800 4000 4200 4400 4600 4800 5000
Frequency (MHz) Frequency (MHz)
(i) Ant3: 3400-5000MHz (l) Ant4: 3400-5000MHz

Fig. 16. Antenna efficiency for a 4x4 MIMO system.

775 ACES JOURNAL, Vol. 36, No. 6, June 2021

Measured_GND
Measured_GND Measured_GND
Measured_GND
Measured_Vehicle
Measured_Vehicle Measured_Vehicle
Measured_Vehicle
0.5 10 0.5 10

0.4 9.8 0.4 9.8

0.3 9.6 0.3 9.6

DG (dB)

DG (dB)
ECC

ECC
0.2 9.4 0.2 9.4

0.1 9.2 0.1 9.2

0 9 0 9
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Frequency (MHz) Frequency (MHz)

(a) Ant1-Ant2 (d) Ant2-Ant3

Measured_GND
Measured_GND Measured_GND
Measured_GND

Measured_Vehicle
Measured_Vehicle Measured_Vehicle
Measured_Vehicle
0.2 10 0.5 10

0.4 9.8
0.15 9.99

0.3 9.6
DG (dB)

DG (dB)
ECC

ECC

0.1 9.98
0.2 9.4

0.05 9.97
0.1 9.2

0 9.96 0 9
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Frequency (MHz) Frequency (MHz)
(b) Ant1-Ant3 (e) Ant2-Ant4

Measured_GND
Measured_GND Measured_GND
Measured_GND
Measured_Vehicle
Measured_Vehicle Measured_Vehicle
Measured_Vehicle
0.5 10 0.5 10

0.4 9.8 0.4 9.6

0.3 9.6 0.3 9.2

DG (dB)

DG (dB)
ECC

ECC

0.2 9.4 0.2 8.8

0.1 9.2 0.1 8.4

0 9 0 8
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Frequency (MHz) Frequency (MHz)
(c) Ant1-Ant4 (f) Ant3-Ant4

Fig. 17. A 4x4 MIMO system ECC and DG.

KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 776

Table 4: Literature review summary

Ref. Type BW Antenna Dimension ECC Method / Value
(LxWxH) (mm3)
10 2x2 monopole 700MHz-900MHz Not reported S-param only / lower than 0.02
11 2x2 PIFA 775MHz-925MHz 59.5x12.4x21 S-param only / lower than 0.5
12 2x2 PIFA 790MHz-2.69GHz 50x50x28 E field components / lower than 0.3
13 2x2 printed monopole 790MHz-3GHz 30x0.8x80 S-param only / lower than 0.05
14 2x2 printed planar 698MHz-2700MHz 52x1.6x65 S-param only/ lower than 0.5
monopoles
15 2x2 PIFA 690MHz-2700MHzPIFA length is S-param only /lower than 0.5
75.9mm / height
25.5mm
16 2x2 monopole 698MHz-2.69GHz Monopole heights are S-param only / lower than 0.5
55 and 45mm
17 2x2 printed monopoles 698MHz-2690MHz 25x2x55 Not reported / lower than 0.3
18 2x2 printed monopole 698MHz-3GHz PIFA 65x62x20 / Not reported
PIFA Monopole height is 53
19 2x2 Nefer Antenna 700MHz-6GHz 70x70x29 S-param only / lower than 0.16
20 4x4 sleeve monopoles 790MHz-5GHz Not reported Not reported / lower than 0.12
29 2x2 printed Yagi 2GHz-4.5GHz 60x1.6x55 Not reported/ lower than 0.5
30 2x2 loaded monopoles 2.4GHz-11GHz 24x2.2x29 E-field components / lower than 0.02

VI. CONCLUSION ACKNOWLEDGMENT

Three MIMO systems based on a novel branched The authors would like to thank Oakland University
Monopole structure that operates in cellular 5G bands for supporting this research with measurements tools and
(617MHz-5GHz) and can easily be in integrated inside a simulation software.
sharkfin package on a car roof have been presented in
this paper. The ECC and DG derived from the radiation REFERENCES
patterns of each antenna element were calculated for [1] J. Malik, D. Nagpal, and M. V. Kartikeyan,
each of the three MIMO configurations. The configuration “MIMO antenna with omnidirectional pattern
I MIMO system in Fig. 1 represents a 2x2 MIMO diversity,” Electronics Letters, vol. 52, no. 2, pp.
systems with antennas separated by a 135mm distance 102-104, Jan. 2016.
integrated on a sharkfin module on a car roof and it [2] L. Lanctot and O. Jonah, “Cellular antenna
allows for a passive isolation better than 12dB, total performance impact on MIMO in vehicle,” 2018
average efficiencies higher than 71% on vehicle across IEEE International Symposium on Antennas and
all bands, ECC lower than 0.13, and DG higher than Propagation & USNC/URSI National Radio
9.9dB. The configuration II MIMO system in Fig. 7 Science Meeting, Boston, MA, pp. 353-354, July
represents a 2x2 MIMO systems with antennas separated 2018.
by a 125mm distance integrated on a sharkfin module on [3] A. M. Elshirkasi, A. Abdullah Al-Hadi, M. F.
a car roof and it provides a passive isolation better than Mansor, R. Khan, and P. J. Soh, “Envelope
15dB, total average efficiencies higher than 71.7% on correlation coefficient of a two-port MIMO
vehicle across all bands, ECC lower than 0.02, and DG terminal antenna under Uniform and Gaussian
of 10dB. Finally, in order to increase the system capacity angular power spectrum with user's hand effect,”
and data rate by allowing four data streams to be sent and Progress in Electromagnetics Research C, vol. 92,
received simultaneously, a novel 4x4 MIMO antenna 123-136, Apr. 2019.
system constructed of four Monopole elements as in Fig. [4] S. Zhekov, A. Tatomirescu, E. Foroozanfard, and
12 has been designed. It achieves passive isolation better G. F. Pedersen, “Experimental investigation on the
than 10dB, total average efficiencies higher than 65% on effect of user’s hand proximity on a compact
vehicle across all bands, ECC lower than 0.46, and DG ultrawideband MIMO antenna array,” IET
higher than 8.9dB. In general, each MIMO configuration Microwaves, Antennas Propag., vol. 10, no. 13, pp.
has a satisfactory performance and can be used easily 1402-1410, Oct. 2016.
in the vehicular application depending on the desired [5] L. L. Nagy, Automobile Antennas, McGraw-Hill,
requirement and dimensions. New York, 2007.
777 ACES JOURNAL, Vol. 36, No. 6, June 2021

[6] K. L. Wong, Planar Antennas for Wireless [17] Y. Liu, Z. Ai, G. Liu, and Y. Jia, “An Integrated
Communications, Wiley Interscience, Hoboken, shark-fin antenna for MIMO-LTE, FM, and GPS
NJ, Jan. 2003. applications,” IEEE Antennas and Wireless
[7] V. Rabinovich, N. Alexandrov, and B. Alkhateeb, Propagation Letters, vol. 18, no. 8, pp. 1666-1670,
Automotive Antenna Design and Applications, Aug. 2019.
CRC Press, Taylor & Francis, Boca Raton, FL, [18] N. Guan, H. Tayama, M. Ueyama, Y. Yoshijima,
Dec. 2017. and H. Chiba, “A roof automobile module for
[8] S. Arianos, G. Dassano, F. Vipiana and M. Orefice, LTE-MIMO antennas,” 2015 IEEE-APS Topical
“Design of multi-frequency compact antennas for Conference on Antennas and Propagation in
automotive communications,” IEEE Transactions Wireless Communications (APWC), Turin, pp.
on Antennas and Propagation, vol. 60, no. 12, pp. 387-391, Sep. 2015.
5604-5612, Dec. 2012. [19] S. Hastürkoğlu, M. Almarashli, and S. Lindenmeier,
[9] G. Artner, W. Kotterman, G. Del Galdo, and M. A. “A compact wideband terrestial MIMO-Antenna
Hein, “Automotive antenna roof for cooperative set for 4G, 5G, WLAN and V2X and evaluation
connected driving,” IEEE Access, vol. 7, pp. of its LTE-Performance in an urban region,”
20083-20090, Jan. 2019. 2019 13th European Conference on Antennas and
[10] H. J. Song, A. Bekaryan, J. H. Schaffner, T. Talty, Propagation (EuCAP), Krakow, Poland, pp. 1-5,
D. Carper, E. Yasan, and A. Duzdar, “Evaluation Mar. 2019.
of vehicle-level MIMO antennas: capacity, total [20] O. Jonah, “5G antenna for automotive applications,”
embedded efficiency, and envelope correlation,” 2020 IEEE International Symposium on Antennas
2014 IEEE-APS Topical Conference on Antennas and Propagation and North American Radio
and Propagation in Wireless Communications Science Meeting, pp. 1493-1494, Web, July 2020.
(APWC), Palm Beach, FL, pp. 89-92, Aug. 2014. [21] T. Jiang, T. Jiao, and Y. Li, “A low mutual
[11] O. Kwon, R. Song, and B. Kim, “A fully integrated coupling MIMO antenna using periodic multi-
shark-fin antenna for MIMO-LTE, GPS, WLAN, layered electromagnetic band gap structures,” The
and WAVE applications,” IEEE Antennas and Applied Computational Electromagnetics Society
Wireless Propagation Letters, vol. 17, no. 4, pp. (ACES), vol. 33, no. 3, Mar. 2018.
600-603, Apr. 2018. [22] F. Liu, J. Guo, L. Zhao, G. Huang, Y. Li, and Y.
[12] V. Franchina, A. Michel, P. Nepa, M. Gallo, R. Yin “Ceramic superstrate-based decoupling method
Parolari, A. P. Filisan, and D. Zamberlan, “A for two closely packed antennas with cross-
compact 3D antenna for automotive LTE MIMO polarization suppression,” IEEE Transactions on
applications,” 2017 IEEE-APS Topical Conference Antennas and Propagation, vol. 69, no. 3, pp.
on Antennas and Propagation in Wireless 1751-1756, Mar. 2021.
Communications (APWC), Verona, Italy, pp. 326- [23] S. Luo, Y. Li, Y. Xia, G. Yang, L. Sun, and L.
329, Oct. 2017. Zhao, “Mutual coupling reduction of a dual-band
[13] A. Heiman, A. Badescu, and A. Saftoiu, “A new antenna array using dual-frequency metamaterial
multiple input multiple output V2V automotive structure,” The Applied Computational Electro-
antenna for long term evolution band applications,” magnetics Society (ACES), vol. 34, no. 3, pp. 403-
2018 International Symposium on Fundamentals 410, Mar. 2019.
of Electrical Engineering (ISFEE), Bucharest, [24] A. Mansoor and R. Amiri, “Mutual coupling
Romania, pp. 1-5, Nov. 2018. reduction of closely spaced MIMO antenna using
[14] D. Preradovic and D. N. Aloi, “Cross polarized 2x2 frequency selective surface based on meta-
LTE MIMO system for automotive shark fin materials,” The Applied Computational Electro-
application,” The Applied Computational Electro- magnetics Society (ACES), vol. 32, no. 12, pp.
magnetics Society (ACES), vol. 35, no. 10, pp. 1064-1068, Dec. 2017.
1207-1216, Oct. 2020. [25] K. Yu, Y. Li, and X. Liu, “Mutual coupling
[15] C. Demien and R. Sarkis, “Design of shark fin reduction of a MIMO antenna array using 3-D
integrated antenna systems for automotive novel meta-material structures,” The Applied
applications,” 2019 PhotonIcs & Electromagnetics Computational Electromagnetics Society (ACES),
Research Symposium - Spring (PIERS-Spring), vol. 33, no. 7, pp. 758-763, July 2018.
Rome, Italy, pp. 620-627, June 2019. [26] S. Luo, Y. Li, Y. Xia, and L. Zhang “A low mutual
[16] A. Thiel, L. Ekiz, O. Klemp, and M. Schultz, coupling antenna array with gain enhancement
“Automotive grade MIMO antenna setup and using metamaterial loading and neutralization line
performance evaluation for LTE-communications,” structure,” The Applied Computational Electro-
2013 International Workshop on Antenna Tech- magnetics Society (ACES), vol. 34, no. 3, pp. 411-
nology (iWAT), Karlsruhe, pp. 171-174, Mar. 2013. 418, Mar. 2019.
KHALIFA, YACOUB, ALOI: COMPACT 2X2 AND 4X4 MIMO ANTENNA SYSTEMS 778

[27] J. Jiang, Y. Xia, and Y. Li “High isolated X-band Ahmad M. Yacoub received the
MIMO array using novel wheel-like metamaterial B.S. degree in Electrical Engineering
decoupling structure,” The Applied Computational from Princess Sumaya University
Electromagnetics Society (ACES), vol. 34, no. 12, for Technology Amman, Jordan, in
pp. 1829-1836, Dec. 2019. 2014 and the M.S. degree in Electrical
[28] S. F. Beegum and S. K. Mishra, “Compact WLAN and Computer Engineering from
band-notched printed ultrawideband MIMO antenna Oakland University, Rochester,
with polarization diversity,” Progress in Electro- Michigan, in 2018. He is currently
magnetics Research C, vol. 61, pp. 149-159, Jan. pursuing the Ph.D. degree in Electrical and Computer
2016. Engineering at Oakland University.
[29] K. Sreelakshmi, P. Bora, M. Mudaliar, Y. Dhanade, He served as a Research Assistant at Applied EMAG
and B. T. P. Madhav, “Linear array Yagi-Uda 5G and Wireless lab at Oakland University from 2016-2018.
antenna for vehicular application,” International He has been employed in Molex LLC, Grand Blanc,
Journal of Engineering & Technology, vol. 7. pp. Michigan from 2018 until present. Yacoub’s research
513-517, Dec. 2017. interests reside in area of applied electromagnetics with
[30] D. Potti, Y. Tusharika, M. G. N. Alsath, S. emphasis on antenna measurements, antenna modeling/
Kirubaveni, M. Kanagasabai, R. Sankararajan, S. analysis and antenna design. He is an inventor/co-
Narendhiran, and P. B. Bhargav, “A novel optically inventor of 2 patents.
transparent UWB antenna for automotive MIMO
communications,” IEEE Transactions on Antennas Daniel N. Aloi received his B.S.
and Propagation, vol. 69, pp. 3821-3828, July 2021. (1992), M.S. (1996) and Ph.D. (1999)
degrees in Electrical Engineering
from Ohio University, located in
Athens, Ohio, USA. He served as a
Research Assistant from 1995-1999
Mohamed O. Khalifa received the in the Avionics Engineering Center
B.S. degree in Electrical and within the School of Engineering
Electronic Engineering from and Computer Science at Ohio University; Summer
University of Khartoum, Khartoum, Intern at Rockwell International in Cedar Rapids, Iowa,
Sudan, in 2010 and the M.S. degree and Senior Project Engineer at OnStar, Incorporated, a
in Electrical Engineering from King subsidiary of General Motors from 2000-2002. He
Fahd University of Petroleum and has been employed in the Electrical and Computer
Minerals, Dhahran, KSA, in 2015. Engineering Department at Oakland University in
He is currently pursuing the Ph.D. degree in Electrical Rochester, Michigan from 2002 until present. He is the
and Computer Engineering at Oakland University. Founder and Director of the Applied EMAG and
He served as a Research Assistant at King Fahd Wireless Lab at Oakland University.
University of Petroleum and Minerals, Visiting graduate Aloi’s research interests reside in area of applied
Intern at I-Radio Lab within the School of Engineering electromagnetics with emphasis on antenna measurements,
at University of Calgary in Alberta, Canada and antenna modeling/analysis and antenna design. He is a
Research Assistant at Applied EMAG and Wireless Lab member of the Institute of Navigation and is a senior
at Oakland University from 2012-2018. He has been member of the Institute of Electrical and Electronics
employed in Ficosa North America, Madison Heights, Engineers (IEEE). He has received in excess of $4M in
Michigan from 2018-2019 then in Molex LLC, Grand research funding from a variety of federal and private
Blanc, Michigan from 2019 until present. His research entities including the Federal Aviation Administration,
interests reside in area of Power Amplifier design and Defense Advanced Research Program Agency (DARPA)
linearization techniques and applied electromagnetics and the National Science Foundation (NSF). He has
with emphasis on antenna measurements, antenna authored/co-authored over 100 technical papers and is an
modeling/analysis and antenna design. He has authored/ inventor on 5 patents.
co-authored around 10 technical papers and is an inventor
on 3 patents.