Engineering Science and Technology, an International Journal 64 (2025) 102026

Contents lists available at ScienceDirect

Engineering Science and Technology, an International Journal

journal homepage: www.elsevier.com/locate/jestch

Full length article

Swarm intelligence driven inverse design of slot patterns for sub-6 GHz 5G
MIMO antennas in virtual reality applications
Shadman Fuad Bin Faruquee , Himel Bhuiyan , Md. Shafiul Alam, Sharnali Islam,
Khaleda Ali ∗
Department of Electrical and Electronic Engineering, University of Dhaka, Dhaka 1000, Bangladesh

ARTICLE INFO ABSTRACT

Keywords: This paper presents an efficient approach for developing miniature sub-6 GHz 5G MIMO antennas with
5G enhanced bandwidth and high isolation. Utilizing the capabilities of particle swarm optimization (PSO)
MIMO antenna on a pixelated surface, an objective function is formulated to minimize the reciprocal of the bandwidth
Mutual coupling
while ensuring that the reflection coefficient (|𝑆11 |) remains below -10 dB and the transmission coefficient
Path loss
(|𝑆21 |) is below -20 dB. This methodology effectively eliminates the necessity for incorporating complex
Virtual reality
Particle swarm optimization
decoupling networks to enhance isolation. The inverse design concept is independently implemented in the slot
arrangements of two different configurations of 2-element MIMO antennas: one in which the individual antenna
components are placed in an antiparallel orientation, and the other in which they are arranged orthogonally.
The optimized designs, measuring 24 × 48 (0.4𝜆𝑜 ×0.8𝜆𝑜 ) 𝑚𝑚2 in size, yield operating spectra spanning from
4.53 GHz to 7.01 GHz for the orthogonal scenario and 3.95 GHz to 5.15 GHz for the antiparallel orientation in
free space. An ECC <0.0173 and MEG <-6 dB, observed in the operating bandwidth, supports satisfactory MIMO
antenna diversity performance. Having negligible current coupling between ports, these MIMO antennas, with
a diversity gain of close to 10 dB are well-suited for integration into virtual reality (VR) headsets. In situ path
loss measurement campaign was executed by integrating the antenna into participants’ virtual reality headsets.
The performance of the proposed MIMO antenna has been tested through numerical analysis, equivalent circuit
modeling, and measurement campaigns using fabricated prototypes. Experimental results from testing the
prototypes closely align with the simulations, confirming the effectiveness of the proposed designs both in
free space and in proximity to the human body.

1. Introduction consumption while simultaneously ensuring accelerated and secure

radio protocols with better precision [7] brings an added benefit that
The constant drive to improve channel capacity and data rate in extends the usability of sub-6 GHz 5G MIMO antennas in diverse appli-
wireless communications has led to the realization of 5G standards. cations [8–13]. In [12], 10 elements and in [13], 18 elements MIMO
The transformative impact of 5G networks on smart communications is antennas are reported that are suitable for 5.7-inch mobile handsets.
characterized by the integration of physical objects into cyberspace and Escalation in antenna elements in these massive MIMO antennas unde-
the seamless connectivity of low-powered portable handheld devices
niably enhances data transmission rates. However, such complex and
[1–3]. Multiple input multiple output (MIMO) antennas, notably ultra-
expensive configurations may also cause greater channel correlation
wideband types, hold significant opportunities in this context. These
challenges, primarily due to the heightened inter-element interference.
antennas in fifth generation (5G) wireless systems, operating in either
sub-6 GHz or millimeter wave (mm-Wave) frequency bands, present In addition, the larger physical dimensions of massive MIMO antenna
distinct advantages and limitations in their respective implementations. arrays provide significant integration issues for portable devices. On
At sub-6 GHz frequencies, MIMO antennas offer more extensive cov- the other hand, rapid advancement of immersive computing and small
erage than the ones at millimeter-wave frequencies, which, although IoT gadgets, particularly for surveillance and wearable communica-
possessing greater capacity, are more susceptible to physical obstruc- tions [6], has generated a substantial demand for miniature antenna
tions and environmental attenuation [4–6]. Besides, reduced power

∗ Correspondence to: Curzon Hall, University of Dhaka, Dhaka 1000, Bangladesh.

E-mail addresses: [email protected] (S.F.B. Faruquee), [email protected] (H. Bhuiyan), [email protected] (M.S. Alam),
[email protected] (S. Islam), [email protected] (K. Ali).

https://doi.org/10.1016/j.jestch.2025.102026
Received 29 September 2024; Received in revised form 21 February 2025; Accepted 24 February 2025
Available online 8 March 2025
2215-0986/© 2025 The Authors. Published by Elsevier B.V. on behalf of Karabuk University. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

systems that can accommodate elevated data rates and expanded band-
widths. This tendency is further aggravated by the projected significant
rise in virtual reality industry revenue [14], necessitating the incorpo-
ration of low-latency antenna systems into VR headsets [15]. This is due
to the fact that, apart from spatial diversity, MIMO systems’ low-latency
qualities reduce temporal discrepancies between user actions and visual
responses, minimizing discomfort and preserving perceptual cohesive-
ness. The concept of 5G devices with smaller and lighter profiles,
however, runs counter to the idea that greater isolation requires a larger
gap between radiators. By limiting the inter-element distance in tightly
coupled antennas, not only are the bandwidths reduced but also the
mutual coupling is increased [16]. As a trade-off, designing a compact
MIMO antenna that achieves both enhanced bandwidth and reduced
mutual coupling with improved diversity behaviors poses a significant
challenge. Common methods to address these concerns involve the
installation of parasitic elements and the varying lengths or placements
of the radiators [17–20]. In [20], Qu et al. introduced shunt capacitors Fig. 1. Scopes of MIMO antenna when incorporated with VR in immersive communi-
for higher port-to-port isolation, achieving results exceeding 12 dB. To cations.
keep mutual coupling below −20 dB [21] shows a fan-shaped parasitic
decoupler, placed between radiating elements. However, decoupling
networks or parasitic element systems often require a larger surface 2. Two element antenna design
area or lead to a more crowded antenna layout, potentially causing
unwanted interactions with other elements and extra power loss. Liter-
This segment presents the primary architecture and the sequential
ature also demonstrates metamaterial-based superstrates, incorporated
development of a 2-element MIMO antenna. Assessment of impedance
to reduce mutual coupling and the envelope correlation coefficient
bandwidth and mutual coupling between array elements are two criti-
(ECC). Insertion of complementary split-ring resonators has also been
cal parameters in identifying the desired design. CST Microwave Studio
considered for further enhancement of the antenna bandwidth [22,23].
Nevertheless, the use of complementary split-ring resonators (CSRRs) Suite has been used as a full-wave 3-D EM solver to perform the
and metamaterials necessitates high-precision fabrication and is more numerical investigation. A 1.575 mm (ℎ𝑠 ) thick FR-4 substrate with
susceptible to manufacturing tolerances. In [24], characteristic mode a dielectric constant (𝜖𝑟 ) of 4.3 and a loss tangent (tan𝛿) of 0.002 is
analysis is incorporated for designing multi-port antennas. Multiple- designed as the base with a metal ground etched on its bottom. Initially,
evolutionary algorithms, including artificial intelligence-based and hy- a uniform hexagonal patch is chosen for single element antenna due to
brid approaches, have been examined for the optimization of MIMO its potential for achieving wider impedance bandwidth compared to
antenna design [25–28]. Although multiobjective optimization algo- the rectangular ones [31]. Moreover, six equal sides of such patches
rithms utilizing decomposition techniques were implemented in [26], distribute fringing fields uniformly, thereby increasing radiation effi-
the resultant port isolation was limited, failing to surpass −24 dB across ciency and generating radiating fields that extend beyond its physical
a restricted frequency range. A recent study in [29] explored inverse boundaries [32]. The side length of the uniform hexagonal radiator (R)
modeling approaches for dual-port MIMO antennas, including L-shaped has been estimated using Eq. (1) at resonance frequency (𝑓𝑐 ) [32]. A
shorted pins. Nonetheless, the deep learning approach was verified microstrip edge feed technique is employed with the width of a 50 ohm
using only one antenna configuration, yielding a narrow-bandwidth feedline stated using Eq. (2) [33].
of 110 MHz. In contrast, swarm intelligence approaches provide an √
efficient exploration of the solution space and faster convergence to 𝑐 2
𝑅= (1)
optimal or near-optimal solutions. Particle swarm optimization (PSO), 4 ⋅ 𝑓𝑐 𝜖𝑟 + 1
in addition, requires fewer computational resources and provides better 60 ℎ 𝑊 𝑓
𝑍0 = √ ln 8 𝑠 + 0.25 (2)
interpretability compared to machine learning techniques [30]. In this 𝜖𝑟 𝑊𝑓 ℎ𝑠
work, taking advantage of this evolutionary algorithm, we propose to
The calculated value of the dimension of R and 𝑊𝑓 are 0.139 𝜆0
achieve reduced mutual coupling at enhanced bandwidths in smaller
and 0.046 𝜆0 respectively considering the electrical length of 5 GHz.
packages of MIMO antennas. A notable feature of the presented an-
tenna working at sub-6 GHz band is its capability for straightforward However, these parameters have been further optimized based on the
integration with virtual reality headset systems. This frequency range 3-D EM solver solutions and the optimized dimensions are given in
provides an optimal balance between coverage and data capacity, Table 1. In order to enhance impedance matching, a closer look at Fig.
offering sufficient bandwidth to support the high data rates required for 2 indicates that grounds 1 and 2 exhibit high port reflection, with |𝑆11 |
VR applications while maintaining reliable signal propagation through exceeding −10 dB throughout the spectrum. In contrast, the resonance
various obstacles. As depicted in Fig. 1, the proposed antenna’s utility frequency band transitions to a higher range with the addition of a stub
extends across a diverse spectrum of potential applications, spanning in the ground plane, and ground 3 exhibits a slightly reduced reflection
healthcare, remote driving, and personal entertainment. Through the coefficient. Ground 4 exhibits superior impedance matching by incorpo-
use of an inverse design approach to improve slot distribution topology rating stubs, resulting in an inverted F-shaped ground plane geometry
on the radiating patches, we attained mutual coupling suppression that also improves the bandwidth. The side lengths of the stubs in
below −20 dB throughout an expanded bandwidth. To the best of our the inverted F-shaped ground plane are calculated using Eqs. (3) and
knowledge, this method for improving isolation and bandwidth-which (4) [32]. Here, 𝑓𝑠 is the resonance frequency after adding the stubs.
operates without any dedicated decoupling structures-is the first to be √
proposed and has been validated for two different sub 6 GHz MIMO 𝑐 2
𝐿1 = (3)
antenna configurations. Both the proposed optimized 5G antennas with 2 ⋅ 𝑓𝑠 𝜖𝑟 + 1
small footprints of (0.4𝜆𝑜 × 0.8𝜆𝑜 ) find themselves suitable to be inte- 𝐿1 ≈ 𝐿2 + 𝐿3 (4)
grated into VR headsets due to its improved bandwidth,isolation, and
low SAR. Experimental campaigns involving real human subjects with Quantitative analysis reveals that the implementation of ground-4 re-
the presented MIMO antenna-incorporated headsets have also been sults in an around 6 dB improvement in return loss compared to
carried out pathloss across different onbody links are investigated. the antenna utilizing ground-3. This configuration keeps a significant

2
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

Fig. 2. Reflection coefficients (|𝑆11 |) for modified ground incorporated single element
antenna.
Fig. 4. Schematic diagram of (a) Radiating element of MIMO antenna (b) pixelated
pattern of the patch.

Table 1
Optimized parameter of the proposed antenna.
Parameters Dimensions (𝜆0 ) Parameters Dimensions (𝜆0 )
W 0.8 L1 0.2
L 0.4 L2 0.1
Wg 0.4 L3 0.09
Lg 0.05 𝑅 0.13
Wf 0.043 g 0.025
Lf 0.084 t = t1 0.03

Fig. 3. Reflection coefficient (|𝑆11 |) for (a) antiparallel and (b) orthogonal arrangement
of antenna elements in the MIMO antenna. Here, d is measured in millimeters. Here, 𝑥𝑚 indicates the position of 𝑚𝑡ℎ particle; 𝑤𝑙𝑜𝑐 and 𝑤𝑔𝑙𝑜𝑏 stand
for two positive cognitive and social factors, respectively; 𝑟1 and 𝑟2
designate uniformly distributed random numbers between 0 and 1.
Once the velocity has been ascertained, it is easier to transport the
portion of the patches uncovered, leading to enhanced impedance particle to its next location.
matching and a better reflection coefficient.
The optimization of impedance bandwidth is achieved by adjusting 𝑥𝑖𝑚 = 𝑥𝑖−1 𝑖
𝑚 + 𝑣𝑚 (6)
the inter-element distance (d) for two distinct MIMO antenna con- 1
figurations: orthogonal and antiparallel (see inset of Fig. 3(a) and 𝑇 𝑓 (𝑣𝑖𝑚 ) = (7)
1 + 𝑒𝑥𝑝(−𝑣𝑖𝑚 )
(b)). Here, gaps have been increased in steps from 18 mm to 24 mm 𝑥𝑖𝑚 = 1, 𝑤ℎ𝑒𝑛 𝑟𝑖1 < 𝑇 𝑓 (𝑣𝑖𝑚 )
where 24 mm is the highest possible gap offered in the proposed (8)
= 0, 𝑤ℎ𝑒𝑛 𝑟𝑖1 > 𝑇 𝑓 (𝑣𝑖𝑚 )
antenna. 24 mm gap between the two patches is considered for both
configurations for the rest of the study. However, this gap worsens the Eq. (6) can be used to explain the 𝑚th particle’s updated posi-
mutual coupling between the antenna elements. To alleviate this issue, tion mechanism. Typically, PSO, in its original form is employed for
a BPSO algorithm-based inverse design concept has been implemented, real-value, considering our emphasis on enhancing slot distribution
strategically optimizing the slot configurations, which makes it a pixe- topology in this work, BPSO is a more suitable approach for this work.
lated structure. The next section elaborates on the concept of the BPSO In BPSO, the position of each particle is represented by a binary string
algorithm. It is worth mentioning that a parametric study performed to of length 𝑁 while particle velocities are kept real-valued. Updating
determine the optimal design parameters is excluded for the brevity of a particle’s position indicates changing a bit to either 1 or 0; hence,
the article. The specifications for the baseline antenna configuration, the position of each particle in a swarm can be represented as 𝑥𝑚 =
shown in Fig. 4(a), are presented in Table 1 in terms of wavelength 𝑥1 , 𝑥2 , …, 𝑥𝑁 . Similarly, the velocity represents the probability of the
considering the electrical length of 5 GHz frequency. bit being 1 or 0 as 𝑣𝑚 = 𝑣1 , 𝑣2 , …, 𝑣𝑁 . All real-valued velocities are
transformed to probabilistic values using a sigmoid transfer function
3. Concepts of Particle Swarm Optimization (PSO) algorithm (𝑇 𝑓 ) as in Eq. (7) [37]. Finally, the modified position that updates
Eq. (6) is provided in Eq. (8) [38].
Particle Swarm Optimization (PSO) is an adaptive optimization
technique that is popularly used for antenna design-related issues [34]. 4. BPSO incorporated miniature MIMO antenna
This multidimensional search technique, inspired by the social cooper-
ation of flocking birds, begins with a population of random solutions The primary objective of implementing the BPSO algorithm in
called particles that can move with dynamically adjusted velocities (𝑣) MIMO antenna inverse design is to develop a miniature pixelated
in an N-dimensional search space. The positions of the particles are antenna with ultra-wide bandwidth and high isolation between el-
updated towards the optimum or near-optimum solution (position) by ements. For forward modeling, we have utilized the full-wave EM
a velocity, 𝑣. The velocities of these particles are determined using two simulation package in CST Microwave Studio. The simulations have
optimum parameters: 𝑝𝑚 , the best position (the position that gives the been performed on a server equipped with 32 GB of RAM. Before
best fitness value) of the particle 𝑚𝑡ℎ obtained so far, and g, the best starting the inverse design process, a target region was selected on the
position obtained by any particle [35]. The velocity of the 𝑚𝑡ℎ particle hexagonal patch where the BPSO algorithm would be applied. To be
at the 𝑖th instant, 𝑣𝑖𝑚 , can be estimated as Eq. (5) [36]. specific, this region is pixelated into 32 cells, each cell occupying an
area of 1 × 1 mm2 (as in Fig. 4(b)) and assigned to a binary bit of 0 or 1.
𝑣𝑖𝑚 = 𝑣𝑖−1 𝑖−1 𝑖−1
𝑚 + 𝑤𝑙𝑜𝑐 ⋅ 𝑟1 ⋅ (𝑝𝑚 − 𝑥𝑚 ) + 𝑤𝑔𝑙𝑜𝑏 ⋅ 𝑟2 ⋅ (𝑔
𝑖−1
− 𝑥𝑖−1
𝑚 ) (5) A binary value of 0 indicates the presence of a slot, while 1 indicates its

3
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

Fig. 7. (a) Outcome of inverse design process and (b) Simulated |𝑆11 | with different
patch structures in the MIMO antenna.

Fig. 8. Optimized antenna schematics for (a) antiparallel and (b) orthogonal orienta-
tion.
Fig. 5. Step-by-step process of the BPSO incorporated MIMO antenna design.

process. With a swarm size of 30 agents, convergence to a stable

solution is reached after 72 iterations for the orthogonal arrangement
and 52 iterations for the antiparallel arrangement (as shown in Fig. 6).
The EM solver processed a total of 2160 and 1560 requests for
orthogonal and antiparallel orientations, respectively. The global op-
timum value of the objective function converged in 0.4016 ns for the
orthogonal configuration and 0.833 ns for the antiparallel configura-
tion. Each individual forward simulation requires approximately one
minute to complete, which leads to a total computation time of about
1.5 h per generation. Furthermore, the developed BPSO algorithm was
executed in several optimization trials with identical criteria and objec-
tives, consistently yielding the same optimal solutions, as illustrated in
Fig. 6. Convergence curve of the objective function during optimization. Fig. 6. The evolution of the hexagonal patch from initial design to final
optimized design by BPSO is shown in Fig. 7(a), while (b) illustrates
the comparative |𝑆11 | results for different patches.
absence. The particle matrix, divided into 32 cells, is symmetrically dis-
tributed between the two halves of the patch, ensuring 180◦ rotational 5. Prototype realization and characteristics analysis
symmetry (refer to Fig. 4(b)). Hence, the independent unique pixels
are reduced to 16. To avoid pixelation at the edges, four rectangular The final optimized MIMO antenna for both antiparallel and orthog-
slots with lengths ofp = 2.25 mm and q = 1.85 mm are taken into onal orientations is shown in Fig. 8. To further validate the performance
account for each patch. The optimization problem intends to find a set of the proposed MIMO antenna, an equivalent circuit model has been
of parameters that minimizes the objective function provided in Eq. (9). developed and antenna prototypes with both antiparallel and orthogo-
( ) nal orientations are fabricated. Characteristics are investigated in terms
1
𝐹𝑜𝑏𝑗 = 𝑛𝑠 (9) of scattering properties, radiation pattern, and diversity features.
𝑓𝑚𝑎𝑥 − 𝑓𝑚𝑖𝑛

Here, 𝑓𝑚𝑎𝑥 and 𝑓𝑚𝑖𝑛 are identified as the maximum and minimum 5.1. Antenna performance analysis with equivalent circuit model
frequencies of the operating band where |𝑆11 | ≤ −10 dB and insertion
loss, |𝑆21 | ≤ −20 dB. The objective function, as in Eq. (7), is set to We have developed an equivalent circuit model to validate the
minimize finding the optimal pattern of a 16-bit binary string. Here, performance of the proposed UWB MIMO antenna using the Keysight
the velocity for each particle is initialized in the finite boundary for ADS circuit simulator (see Fig. 9). Here, the radiating antenna ele-
a higher convergence rate, and the two coefficients (𝑤𝑙𝑜𝑐 and 𝑤𝑔𝑙𝑜𝑏 ) ments are represented as RLC networks with similar properties. The
used in Eq. (5) are set to +2. Fig. 5 illustrates a flowchart providing a operating mechanism of these RLC resonators primarily originates from
step-by-step breakdown of how BPSO is applied in the patch pixelation the principles of multisection filter circuits [39]. Multisection resonant

4
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

Fig. 9. Equivalent circuit model of the proposed antenna.

Table 2
Optimized values of lumped elements in the equivalent circuit model for both antiparallel and orthogonal antennas.
Parameters R1 (Ω) C1 (pF) L1 (nH) Cp1 (pF) Lp1 (nH) R2 (Ω) C2 (pF) L2 (nH) Cp2 (pF) Lp2 (nH) R3 (Ω) C3 (pF)
Anti-parallel 22 6 0.1 2.5 0.35 10 3.15 0.7 1.5 1.2 0.3 20.5
Orthogonal 40.4 15 0.65 1.65 0.85 0.5 0.68 1.85 0.7 15 0.15 0.1
Parameters L3 (nH) Cp3 (pF) Lp3 (nH) R4 (Ω) C4 (pF) L4 (nH) Cp4 (pF) Lp4 (nH) Cp (pF) Lp (nH) Cg (pF) Lg (nH)
Anti-parallel 0.85 0.75 6.5 53 0.1 8.65 0.8 2 0.7 2.45 12.4 1.75
Orthogonal 6.65 0.75 2 27.5 0.25 3.4 1.85 3.7 3.45 0.16 6.2 0.25

Fig. 11. Experimental setup with real-time S-parameter measurement.

Fig. 10. Front and back plane of the fabricated prototype of the proposed MIMO
antenna for (a) antiparallel, and (b) orthogonal orientation.
simulation, equivalent circuit results and the measurement campaign
performed in free space, which validates the design of the proposed
MIMO antenna.
circuits have been used to cover the wide frequency range. Also, such
As in Fig. 12, the distribution of |𝑆11 | indicates that the impedance
filter circuit has been transformed from the low-pass structure into a
bandwidth of <−10 dB for the antiparallel arrangement is 3.55 GHz
bandpass layout [40]. Here, the resistance represents conductive loss of
(3.95–7.5 GHz), which is marginally greater than that of an orthogonal
the transmission line. The circuit model that accounts for the mutual
case of 3.25 GHz (4.53–7.78 GHz). However, in the same order, ac-
coupling between two radiating elements is derived using a network
cording to the distribution of |𝑆21 |, to maintain high isolation (>20 dB)
of series and shunt capacitors and inductors [39,41,42]. The values of
between two antenna elements the frequency range achieved through-
capacitance and inductance in the circuit model are primarily estimated
out the bandwidth is 3.95–5.15 GHz and 4.53–7.01 GHz for antiparallel
using the following Eqs. (10)–(12) [43].
orientation and orthogonal orientation, respectively (see Fig. 12(b) and
𝑊𝑓 ∗ 𝜖𝑟 (d)). Hence, the operating bandwidths where low reflection and high
𝐶= (10)
𝑅 ∗ ℎ𝑠 ∗ 𝜔 0 2 isolation have been maintained are 2.49 GHz and 1.20 GHz for orthog-
1 onal and antiparallel orientations, respectively. A wider bandwidth for
𝐿= (11)
𝐶 ∗ 𝜔0 2 orthogonal positioning is obtained, since isolation increases when the
√
𝜔0 = 𝜔𝐻 ∗ 𝜔𝐿 (12) radiators are placed at 90◦ to each other.
Here, 𝜔𝐿 and 𝜔𝐻 are lower and higher angular frequencies of the band- The integration of the Binary Particle Swarm Optimization (BPSO)
width respectively. Following parametric optimization of the lumped algorithm to optimize the slot distribution on the patch has resulted
components, the finalized metrics are provided in Table 2. Variations in in a significant improvement in bandwidth while maintaining high
coupling lead to orthogonal and antiparallel configurations displaying isolation. Specifically, this approach resulted in a 25% increase in
unique characteristic values. bandwidth for antiparallel orientation and a 112% increase in or-
thogonal arrangement compared to the Multiple-Input Multiple-Output
5.2. Optimized antenna fabrication & measurement (MIMO) system that lacks optimized slot distribution. Furthermore,
the minimum return loss value throughout the band has decreased by
Utilizing the optimal slot configurations illustrated in Fig. 8, a 123% from the initial design to the optimized design in the antiparallel
prototype of the improved MIMO antenna is constructed. (see Fig. 10). orientation and by 66% in the orthogonal orientation (see Figs. 3 and
Fabrication was carried out using a 1.6 mm thick FR-4 dielectric 12).
substrate coated with double conducting layers. A photolithographic To gain a more comprehensive understanding of the results, we
chemical etching process was adopted to produce the device. The setup analyze the surface current patterns depicted in Fig. 13.
for the real-time measurement scenario for the S parameter is shown These current distributions illustrate a substantial reduction of the
in Fig. 11. In the experimental campaign, a ZVH-8 vector network coupled current flow between antenna elements when a single antenna
analyzer from RHODE & SCHWARZ was used to measure |𝑆11 | and port is activated at a time at 5 GHz. The reduction reflects a decrease
|𝑆21 | (as in Fig. 12). A reasonable agreement is observed between the in mutual coupling that results in enhanced isolation between antenna

5
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

Fig. 12. Simulated and measured and Eq. circuit responses of (a) |𝑆11 |, (b) |𝑆21 | for
antiparallel and (c) |𝑆11 |, (d) |𝑆21 | for orthogonal orientations in free space.

Fig. 15. Simulated and measured radiation patterns of MIMO antenna configurations:
for co-polarization in an antiparallel arrangement at frequency 4.36 GHz (a)–(b)
and at 5 GHz (c)–(d); for cross-polarization in antiparallel arrangement at frequency
4.36 GHz (e)–(f) and at 5 GHz (g)–(h); for co-polarization in orthogonal arrangement at
frequency 4.84 GHz (i)–(j) and at 5 GHz (k)–(l); for cross-polarization in an orthogonal
arrangement at 4.84 GHz (m)–(n) and at 5 GHz (o)–(p).

Fig. 13. Surface current distribution on the patch surface of (a) antiparallel and (b)
orthogonal arrangements of MIMO antenna at 5 GHz while only one port is energized. Specifically, In coplanar polarization, the azimuthal radiation pattern
behaves as nearly directional as compared to the bidirectional cross-
polarized counterpart. The elevation plane radiation patterns exhibit
different characteristics: coplanar polarization causes dumbbell-shaped
configurations for both antenna types, while cross-polarization results
in a distorted bidirectional morphology. The observed minor tilt in the
elevation plane across all frequencies (illustrated in Fig. 15(b), (d), (f),
(h), (j), (l), (n), and (p)) can be attributed to asymmetrical current
distribution resulting from the finite ground plane configuration. A
comparatively wide beamwidth in all cases highlight the antennas’ suit-
ability for immersive computing applications, where broad coverage
and simplified alignment are crucial for effective communication with
Fig. 14. Simulated gain and Efficiency of (a) antiparallel and (b) orthogonal arrange- moving objects.
ments of MIMO antenna.

5.3. MIMO diversity performance

elements. This arrangement effectively mitigates electromagnetic inter-
actions, significantly reducing parasitic current transfer and potentially To analyze the diversity features of the proposed MIMO antenna,
improving overall system performance. the envelope correlation coefficient (ECC), diversity gain (DG), channel
The radiation efficiencies and gains of the antennas are illustrated capacity loss (CCL), mean effective gain (MEG) and total active reflec-
in Fig. 14. Results demonstrate that the antiparallel and orthogonal tion coefficient (TARC) have been evaluated (see Fig. 16). Firstly, ECC
configurations exhibit maximum radiation efficiencies of 71.85% and value has been computed from the equation provided in [44,45]. ECC
71.54%, respectively. While maintaining satisfactory efficiency levels, quantifies the degree of correlation between the far-field patterns of
the MIMO antenna system featuring orthogonally oriented elements the individual elements within the MIMO antenna. In ideal conditions,
achieves a peak gain of 2.96 dB, coupled with a directivity of 4.56 dB.
the ECC of a MIMO antenna is 0 due to completely uncorrelated
The free space, far-field radiation characteristics of the optimized
signals. However, in practical scenarios, an ECC value of <0.5 is con-
MIMO antenna are demonstrated in Fig. 15. The antennas were tested
sidered satisfactory for effective MIMO performance. Owing to the high
at 5 GHz and at one of their respective resonant frequencies, with
one port active and the other port terminated using 50-ohm loads. isolation, the envelope correlation coefficient (ECC) between the radi-
Here, the chosen resonant frequency for antiparallel and orthogonal ating patches remain below 0.0137 dB and 0.0173 dB throughout the
arrangements are 4.36 GHz and 4.84 GHz respectively. Both co and operating bandwidth, for antiparallel and orthogonal configurations,
cross-polarization of the antennas have been considered in this study respectively.
and all values are normalized to peak values. A reasonable agree- Diversity Gain (DG) is another metric used to analyze the transmis-
ment is observed between the simulated and experimental results. sion power losses associated with the implementation of a diversity

6
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

Table 3
Comparison between the initial and optimized design.
Characteristics parameter Initial design Optimized design
Anti-parallel Orthogonal Anti-parallel Orthogonal
Max. return loss (dB) 16.2 25.6 36.8 38.9
Isolation (dB) >15 >20 >20 >20
Operating frequency band (GHz) 6.54-7.5 6.55–7.72 3.95–5.15 4.53–7.01
Maximum efficiency (%) 67.36 65.6 71.85 71.54
Envelope correlation coefficient <0.045 <0.039 <0.0137 <0.0173

Fig. 17. SAR for 10 g of tissue (a) anti-parallel antenna and (b) orthogonal antenna.

MEG values are calculated from s-parameters using equation shown

in [48]. As illustrated in Fig. 16(d) MEG values are in the accept-
able range for the proposed MIMO antenna in both anti-parallel and
orthogonal arrangements throughout the operating frequency band.
Total Active Reflection Coefficient (TARC) is another important
diversity parameter which evaluates the overall reflection of a MIMO
antenna system considering coupling between elements and active
port excitations. TARC can be evaluated from s-matrix using equa-
tion in [48]. As shown in Fig. 16(e) and (f), the TARC value is
<−10 dB throughout the operating bandwidth, both for anti-parallel
and orthogonal configurations, respectively.
Table 3 provides an extensive comparison study illustrating the
dominance of the improved antenna design relative to its initial con-
figuration across multiple key performance indicators.

6. Optimized antenna integration with VR: Experimental analysis

Fig. 16. Diversity parameters (a) ECC-DG (antiparallel), (b) ECC-DG (orthogonal), (c)
6.1. SAR analysis
CCL, (d) MEG, (e) TARC (antiparallel) and (f) TARC (orthogonal).

The Specific Absorption Rate (SAR) quantitatively measures the

absorption of electromagnetic energy by biological tissues, when sub-
scheme. This parameter, being a function of the correlation coeffi- jected to electromagnetic field exposure. Since the antennas are in-
cient between antenna signals, has been evaluated using the formula tended for integration with a VR headset, carrying out a SAR analysis
presented in [45]. is essential to evaluate their effects in the worst-case scenario. A flat,
The proposed optimized MIMO antennas achieve diversity gains inhomogeneous model of 150 × 150 mm2 , comprising three layers
exceeding 9.93 dB and 9.913 dB in free space across the operational of body tissue (skin, fat, and muscle), was developed in CST MWS
frequency range for antiparallel and orthogonal orientations, respec- software to perform the SAR analysis using the suggested antennas. The
tively. thickness of skin, fat and muscle tissues are 1 mm, 3 mm and 13 mm
The maximum rate at which information can be reliably transmitted respectively [49]. The dielectric properties of these layers, specifically
through a communication channel is known as channel capacity. Chan- relative permittivity (𝜖), conductivity (𝑆∕𝑚) at the designated spectrum
nel Capacity Loss (CCL) is another metric that quantifies the reduction are obtained from [50] and density (kg/m3) from [51]. According to
in the theoretical maximum rate of information transfer through a IEEE 𝐶95.1 − 2005 standard [52] the SAR value for any 10 g tissues
media, beyond which the probability of error-free transmission be- should be less than 2 W/kg. The highest specific absorption rate (SAR)
comes negligible [46]. The calculations of CCL for a two-element MIMO value in 10 g of average mass is 1.74 W/kg for the anti-parallel antenna,
system (as illustrated in Fig. 16(c)) are performed numerically using whereas for the orthogonal antenna, it is 1.57 W/kg (see Fig. 17).
equations provided in [46]. It can be observed from the graph that
CCL does not exceed 0.3 bps/Hz across the entire working frequency 6.2. On body performance analysis
in both antenna orientations, which agrees with the characteristics of
any efficient MIMO antenna system [47]. This section presents a brief analysis on the performance of the
The Mean Effective Gain (MEG) measures the efficiency of an proposed MIMO antennas when integrated into head-mounted virtual
antenna in receiving power from a multipath environment relative to reality (VR) displays. Assessing the human body’s impact on antenna
its isotropic counterpart. MEG values typically ranges from −3 dB to operation necessitates a comprehensive examination of two crucial
−12 dB for balanced MIMO systems. For the proposed MIMO system, properties: the phenomenon of frequency detuning and the variability

7
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

Table 4
Performance comparison of sub-6 GHz MIMO Antennas.
Ref. Num. of Antenna size (𝜆20 ) Bandwidth (GHz) Isolation (dB) ECC CCL (bps/Hz) Decoupling methods
ports
[20] 4 0.36 × 0.1 3.4–3.6 >12 <0.15 – Shunt capacitor
[22] 4 1.065 × 0.72 5.6–5.92 >20 <0.26 – Metamaterial superstate
[48] 4 0.36 × 0.1 2.76–4.3 >23 <0.001 <0.4 Strip slots
[53] 2 0.85 × 0.5 3.4–3.6 >16 <0.03 <0.4 Parasitic element (hollow stub)
[54] 2 0.85 × 0.42 2.394–2.53 >15.4 <0.007 – Lumped inductance
[55] 4 0.44 × 0.44 4.25–5.13 >13.7 <0.14 – Grounded short branches
2.37–2.65
3.25–3.85
[56] 2 3.33 × 3.33 >20 – – Metallic stubs
5.0–6.1
7.15–8.5
[57] 8 0.6 × 1.2 2.4–3.8 >13 <0.03 – Ground slots
2 0.3 × 0.46 2.61–5.52 >16
[58] <0.004 <0.29 Nutralized line
4 1.03 × 0.79 2.80–5.58 >15
3.4–3.6
[59] 4 0.39 × 0.39 >19.2 <0.002 <0.4 Defected ground structure with decoupling filters
5.15–5.35
3.95-5.15, <0.0137
This work 2 0.4 × 0.8 >20 <0.3 Particle Swarm Optimization
4.53–7.01 <0.0173

Fig. 18. Variation in S-parameter at free space and on-body (a) antiparallel and (b)
orthogonal arrangement of antenna.

in path loss. Frequency detuning scrutinizes the shifts in resonant fre-

quencies due to the subject’s presence. In contrast, path loss, evaluated
across various frequencies and spatial configurations, aids in character- Fig. 19. Setup for path loss measurement using vector network analyzer.
izing the antenna’s in situ coverage. To quantitatively measure various
attributes, we have conducted two different measurement protocols.
The first approach involved positioning the antennas parallel to the
trunk. Fig. 18 demonstrates the measured values of |𝑆11 | and |𝑆21 | for
the antenna placed at a fixed distance of 5 mm from the human par-
ticipant. It is observed that for antiparallel orientation of the antennas,
bandwidth remains between 4–4.6 GHz which indicates that frequency
variation is less than 4% in the antiparallel configuration. However, in
the same position, the orthogonal MIMO antenna offers dual bandwidth
near the body. These variations may be attributed to the synergistic
influence of multiple factors, including the lossy dielectric properties
inherent to human tissue, the directional characteristics of the antenna
radiation pattern, and the specific spatial orientation of the antenna Fig. 20. Measured path loss of the proposed antenna (a) antiparallel and (b) orthogonal
relative to the body. It is important to note that, near the body, the orientation.
mutual coupling is reduced for both the antenna organizations. The
alteration in scattering characteristics and consequently in bandwidth
of the antennas in the presence of a subject suggests that an antenna measurements were conducted at each frequency, and the correspond-
positioned parallel to the body is suboptimal for transmission. ing average path loss value was calculated. Fig. 20 showcases the
The second method involved affixing the antenna to the under- average path loss values for 4 different locations at their respective
side of a virtual reality (VR) headset worn by the human subject free space resonant frequencies and at 5 GHz. At both the frequencies,
while standing. Here, the transmitting antenna maintained a perpen- path loss values span between −30 dB to −50 dB for antiparallel setup
dicular orientation to the individual’s posture; for receivers, identical where as for the latter, average path loss is 5 dB higher at every
antennas were strategically positioned at several locations, including location. Moreover, results indicate that the best link performance
the chest, arm, wrist, and waist (see Fig. 19). The propagation path was obtained when the receiving antenna was placed on the chest,
loss characteristics for distinct on-body channels at the concerned fre- compared to the positions of the arm, waist, or wrist, for all cases. The
quency spectrum were quantitatively evaluated through 𝑆21 parameter better performance in the chest position can be attributed to the shorter
measurements acquired via VNA [60]. At each receiver location, five distance between the transmitting and receiving antennas. The distance

8
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

between the VR transmitter and the arm receiver is considerably greater References
than that between the VR transmitter and the wrist receiver. However,
the variation in path loss between these two links is not substantial. [1] H. Hao, D. Hui, D. Lau, Material advancement in technological development for
the 5G wireless communications, Nanotechnol. Rev. 9 (1) (2020) 683–699.
This empirical evidence suggests that the creeping wave mechanism
[2] S.A. Ali, M. Wajid, A. Kumar, M.S. Alam, Design challenges and possible solutions
dominates signal propagation more than the Friis transmission equation for 5G SIW MIMO and phased array antennas: A review, IEEE Access 10 (2022)
typically applied in Line-of-Sight (LoS) scenarios. A comparison of the 88567–88594.
proposed MIMO antenna with recent literature work is shown in Table [3] S.K. Ibrahim, M.J. Singh, S.S. Al-Bawri, H.H. Ibrahim, M.T. Islam, M.S. Islam, A.
Alzamil, W.M. Abdulkawi, Design, challenges and developments for 5G massive
4, indicating that our optimized antenna design is advantageous in size,
MIMO antenna systems at sub 6-GHz band: a review, Nanomaterials 13 (3)
isolation, and channel capacity loss over a broad operating frequency (2023) 520.
range. [4] B.T. Malik, S. Khan, S. Koziel, Beam steerable MIMO antenna based on conformal
passive reflective metasurface for 5G millimeter wave applications, Sci. Rep. 14
(1) (2024) 24086.
[5] S. Khan, M. Khalily, A. Ali, H. Ali, C. Tanougast, A wideband circularly polarized
7. Conclusion
linear parasitic dielectric resonators antenna array for 5G applications, in: 2020
IEEE 5th International Symposium on Smart and Wireless Systems Within the
Conferences on Intelligent Data Acquisition and Advanced Computing Systems
In this work, a computationally efficient evolutionary algorithm has
(IDAACS-SWS), IEEE, 2020, pp. 1–5.
been introduced to implement the inverse design of MIMO antennas [6] R.N. Tiwari, O.S. Sai, D. Sharma, M.S. Kumar, P. Singh, P. Kumar, C. Sree-
tailored for immersive communications. Specifically, BPSO algorithm- manya, S. Rajasekaran, A low-profile dual-band millimeter wave patch antenna
controlled slot distribution topology on the radiators has been proposed for high-speed wearable and biomedical applications, Results Eng. 24 (2024)
103212.
to maximize bandwidth while maintaining high isolation in miniature
[7] S. Nikolaou, A. Quddious, Antennas for UWB applications, in: UWB
MIMO antennas. The optimal value of the global objective function and Technology-Circuits and Systems, IntechOpen, 2019.
the corresponding solution have been achieved by repeatedly running [8] M. Zada, I.A. Shah, H. Yoo, Integration of sub-6-GHz and mm-wave bands with
the PSO algorithm on two-element MIMO antennas featuring hexagonal a large frequency ratio for future 5G MIMO applications, IEEE Access 9 (2021)
11241–11251.
radiators. PSO, as it reduces the design time and allows for quick [9] A. Desai, M. Palandoken, J. Kulkarni, G. Byun, T.K. Nguyen, Wideband flexi-
evaluation of different antenna configurations, leads to faster and more ble/transparent connected-ground MIMO antennas for sub-6 GHz 5G and WLAN
efficient design iterations. Numerical results have been compared and applications, IEEE Access 9 (2021) 147003–147015.
validated through a dual verification approach, incorporating both [10] A. Baz, D. Jansari, S.P. Lavadiya, S.K. Patel, Miniaturized and high gain
circularly slotted 4× 4 MIMO antenna with diversity performance analysis for
equivalent circuit modeling and experimental measurements, thereby 5G/Wi-Fi/WLAN wireless communication applications, Results Eng. 20 (2023)
establishing the robustness. The optimized footprint of 24 × 48 mm2 101505.
of the MIMO antenna offers a bandwidth of 1.2 GHz and 2.49 GHz [11] X. Fang, G. Wen, D. Inserra, Y. Huang, J. Li, Compact wideband CPW-fed
meandered-slot antenna with slotted Y-shaped central element for Wi-Fi, WiMAX,
for antiparallel and orthogonal orientations, respectively, where mutual
and 5G applications, IEEE Trans. Antennas and Propagation 66 (12) (2018)
couplings are as low as −20 dB in free space. The proposed antenna per- 7395–7399.
formances were evaluated post-integration with VR displays to assess [12] N. Jaglan, S.D. Gupta, B.K. Kanaujia, M.S. Sharawi, 10 element sub-6-GHz multi-
suitability for Sub-6 GHz 5G communication applications. band double-T based MIMO antenna system for 5G smartphones, IEEE Access 9
(2021) 118662–118672.
[13] N. Jaglan, S.D. Gupta, M.S. Sharawi, 18 element massive MIMO/diversity 5G
smartphones antenna design for sub-6 GHz LTE bands 42/43 applications, IEEE
CRediT authorship contribution statement Open J. Antennas Propag. 2 (2021) 533–545.
[14] S. Brodsky, Why immersive computing could be the next big thing, 2023,
URL https://www.lifewire.com/why-immersive-computing-could-be-the-next-big-
Shadman Fuad Bin Faruquee: Writing – review & editing, Writing thing-7112403.
– original draft, Software, Resources, Methodology, Investigation, Data [15] W. Su, P.V. Prasannakumar, Y. Li, G. Ye, J. Zhu, Wearable antennas for virtual
curation, Conceptualization. Himel Bhuiyan: Writing – review & edit- reality cross-body links, IEEE Open J. Antennas Propag. 4 (2023) 207–215.
ing, Writing – original draft, Software, Resources, Methodology, Formal [16] M.S. Sharawi, O. Hammi, Design and Applications of Active Integrated Antennas,
Artech House, 2018.
analysis, Data curation, Conceptualization. Md. Shafiul Alam: Writing [17] C.F. Ding, X.Y. Zhang, C.-D. Xue, et al., Novel pattern-diversity-based decoupling
– review & editing, Software, Methodology, Formal analysis. Sharnali method and its application to multielement MIMO antenna, IEEE Trans. Antennas
Islam: Writing – review & editing, Supervision, Project administration, and Propagation 66 (10) (2018) 4976–4985.
[18] V. Dhasarathan, T.K. Tran, J. Kulkarni, B.A. Garner, Y. Li, et al., Mutual coupling
Formal analysis. Khaleda Ali: Writing – review & editing, Visual-
reduction in dual-band MIMO antenna using parasitic dollar-shaped structure for
ization, Validation, Supervision, Project administration, Investigation, modern wireless communication, IEEE Access 11 (2023) 5617–5628.
Formal analysis. [19] K. Ding, C. Gao, D. Qu, Q. Yin, Compact broadband MIMO antenna with parasitic
strip, IEEE Antennas Wirel. Propag. Lett. 16 (2017) 2349–2353.
[20] L. Qu, H. Kim, K.-Y. Jung, Y. Liu, Compact four-port MIMO antenna module with
Declaration of competing interest decoupling components for 5g terminal devices, IEEE Antennas Wirel. Propag.
Lett. 23 (7) (2024) 2224–2228.
[21] M.M. Hassan, M. Rasool, M.U. Asghar, Z. Zahid, A.A. Khan, I. Rashid, A. Rauf,
The authors declare that they have no known competing finan- F.A. Bhatti, A novel UWB MIMO antenna array with band notch characteristics
using parasitic decoupler, J. Electromagn. Waves Appl. 34 (9) (2020) 1225–1238.
cial interests or personal relationships that could have appeared to
[22] R. Mark, H.V. Singh, K. Mandal, S. Das, Mutual coupling reduction using near-
influence the work reported in this paper. zero 𝜀 and 𝜇 metamaterial-based superstrate for an MIMO application, IET
Microw. Antennas Propag. 14 (6) (2020) 479–484.
[23] M.A.R. Ohi, Z. Hasan, S.F.B. Faruquee, A.A.M. Kawsar, A. Ahmed, Wideband
Acknowledgments Minkowski fractal antenna using complementary split ring resonator in modified
ground plane for 5G wireless communications, Eng. Rep. 3 (9) (2021) e12388.
[24] H. Li, Y. Chen, U. Jakobus, Synthesis, control, and excitation of characteristic
The authors would like to thank the University Grant Commission of modes for platform-integrated antenna designs: A design philosophy, IEEE
Antennas Propag. Mag. 64 (2) (2022) 41–48.
Bangladesh for providing grants to teachers at the University of Dhaka.
[25] M.E. Aydın, R. Durgut, Efficient and adaptive operator selection in swarm
Special thanks to Raunak Al Jami Md. Fahim Faisal for his support intelligence using machine learning approaches, Artif. Intell. Theory Appl. 4 (1)
during fabrication and measurement. (2024) 22–32.

9
S.F.B. Faruquee et al. Engineering Science and Technology, an International Journal 64 (2025) 102026

[26] X. Wang, G. Wang, D. Wang, Q. Zhang, Ensemble-learning-based multiobjective [45] R.N. Tiwari, D. Sharma, P. Singh, P. Kumar, Design of dual-band 4-port flexible
optimization for antenna design, IEEE Trans. Antennas and Propagation 71 (2) MIMO antenna for mm-wave technologies and wearable electronics, IEEE Access
(2022) 1295–1303. 12 (2024) 96649–96659.
[27] F. Zardi, P. Nayeri, P. Rocca, R. Haupt, Artificial intelligence for adaptive and [46] M.N. Hasan, S. Chu, S. Bashir, A DGS monopole antenna loaded with U-shape
reconfigurable antenna arrays: A review, IEEE Antennas Propag. Mag. 63 (3) stub for UWB MIMO applications, Microw. Opt. Technol. Lett. 61 (9) (2019)
(2020) 28–38. 2141–2149.
[28] Q.-Q. Li, Q.-X. Chu, Y.-L. Chang, Design of compact high-isolation MIMO antenna [47] T. Raj, R. Mishra, A. Kapoor, D. Pandey, S. Kundu, A compact quad port
with multiobjective mixed optimization algorithm, IEEE Antennas Wirel. Propag. MIMO antenna with gain augmentation using angularly stable and polarization
Lett. 19 (8) (2020) 1306–1310. insensitive FSS for 5G communication, AEU-Int. J. Electron. Commun. 190 (2025)
[29] D.-L. Wu, X.J. Hu, J.H. Chen, L.H. Ye, J.-F. Li, AI deep learning optimization for 155636.
compact dual-polarized high-isolation antenna using backpropagation algorithm, [48] M. Salehi, H. Oraizi, Wideband high gain metasurface-based 4t4r MIMO antenna
IEEE Antennas Wirel. Propag. Lett. 23 (2) (2024) 898–902. with highly isolated ports for sub-6 GHz 5G applications, Sci. Rep. 14 (1) (2024)
[30] A. Pandey, A.P. Singh, V. Kumar, Design and optimization of circularly polarized 14448.
dielectric resonator-based MIMO antenna using machine learning for 5G Sub-6 [49] B. Yin, M. Ye, Y. Yu, J. Gu, A dual-band, miniaturized, AMC-based wearable
GHz, AEU-Int. J. Electron. Commun. 162 (2023) 154558. antenna for health monitoring applications, Prog. Electromagn. Res. C 112 (2021)
[31] N.K. Darimireddy, R.R. Reddy, A.M. Prasad, A miniaturized hexagonal-triangular 165–177.
fractal antenna for wide-band applications, IEEE Antennas Propag. Mag. 60 (2) [50] R. Andreuccetti, C. Petrucci, An Internet resource for the calculation of the
(2018) 104–110. dielectric properties of body tissues in the frequency range 10 Hz - 100 GHz.
[32] C.A. Balanis, Antenna Theory: Analysis and Design, John wiley & sons, 2015. URL http://niremf.ifac.cnr.it/tissprop/htmlclie/htmlclie.php.
[33] D.M. Pozar, Microwave Engineering: Theory and Techniques, John wiley & sons, [51] B.R. Steensma, E.F. Meliadò, P. Luijten, D.W. Klomp, C.A. van den Berg, A.J.
2021. Raaijmakers, SAR and temperature distributions in a database of realistic human
[34] M.-S. Kang, Y.-J. Won, B.-G. Lim, K.-T. Kim, Efficient synthesis of antenna pattern models for 7 T cardiac imaging, NMR Biomed. 34 (7) (2021) e4525.
using improved PSO for spaceborne SAR performance and imaging in presence [52] IEEE standard for safety levels with respect to human exposure to radio frequency
of element failure, IEEE Sensors J. 18 (16) (2018) 6576–6587. electromagnetic fields, 3 khz to 300 GHz, in: IEEE Std C95. 1- 2005 (Revision
[35] R. Eberhart, J. Kennedy, A new optimizer using particle swarm theory, in: IEEE Std C95. 1- 1991), 2006, pp. 1–238.
MHS’95. Proceedings of the Sixth International Symposium on Micro Machine [53] S.N.R. Rizvi, M. Abu Sufian, W.A. Awan, Y. Choi, N. Hussain, N. Kim, A closely
and Human Science, Ieee, 1995, pp. 39–43. spaced two-port MIMO antenna with a radiation null for out-of-band suppressions
[36] T.S.C. Amma Palanisamy, M. Murugappan, Joint optimisation of ground, feed for 5G sub-6 GHz applications, PloS One 19 (7) (2024) e0306446.
shapes with material distributive topology of patch in UWB antennas using [54] L. Sun, Y. Li, Z. Zhang, Decoupling between extremely closely spaced patch
improved binary particle swarm optimisation, IET Microw. Antennas Propag. 12 antennas by mode cancellation method, IEEE Trans. Antennas and Propagation
(12) (2018) 1967–1972. 69 (6) (2021) 3074–3083.
[37] M.-J. Haji-Ahmadi, V. Nayyeri, M. Soleimani, O.M. Ramahi, Pixelated checker- [55] X. Tian, Z. Du, Wideband shared-radiator four-element MIMO antenna module
board metasurface for ultra-wideband radar cross section reduction, Sci. Rep. 7 for 5G mobile terminals, IEEE Trans. Antennas and Propagation 71 (6) (2023)
(1) (2017) 11437. 4799–4811.
[38] J.-S. Pan, P. Hu, V. Snášel, S.-C. Chu, A survey on binary metaheuristic [56] N. Hussain, N. Kim, Integrated microwave and mm-wave MIMO antenna module
algorithms and their engineering applications, Artif. Intell. Rev. 56 (7) (2023) with 360 pattern diversity for 5G internet of things, IEEE Internet Things J. 9
6101–6167. (24) (2022) 24777–24789.
[39] B. Kasi, Y.E. Jalil, C.K. Chakrabarty, Circuit modeling for UWB-MIMO antenna [57] H.T. Chattha, M.K. Ishfaq, B.A. Khawaja, A. Sharif, N. Sheriff, Compact multiport
array, in: 2013 IEEE International RF and Microwave Conference, RFM, IEEE, MIMO antenna system for 5G IoT and cellular handheld applications, IEEE
2013, pp. 66–70. Antennas Wirel. Propag. Lett. 20 (11) (2021) 2136–2140.
[40] I. Hunter, Theory and Design of Microwave Filters, Iet, 2001, 48. [58] P. Sharma, R.N. Tiwari, P. Singh, Circumferential slots based decoupled wide-
[41] J.-S.G. Hong, M.J. Lancaster, Microstrip filters for RF/microwave applications, band MIMO antenna for 5G and sub-6 GHz applications, Phys. Scr. 99 (6) (2024)
John Wiley & Sons, 2004. 065012.
[42] S. Jayant, G. Srivastava, S. Kumar, Pattern diversity and isolation enhancement [59] K.D. Ayinala, P.K. Sahu, Isolation enhanced compact dual-mode 4-port MIMO
of UWB MIMO antenna based on characteristic modes for mobile terminals, Int. design using slot-based switchable DGS decoupling filters, Wirel. Pers. Commun.
J. Microw. Wirel. Technol. 15 (5) (2023) 793–804. 135 (2) (2024) 805–833.
[43] A. Moradikordalivand, T. Rahman, S. Ebrahimi, S. Hakimi, An equivalent circuit [60] K. Ali, A. Brizzi, A.N. Khan, Y. Hao, On-body nlos radio channel at millimeter-
model for broadband modified rectangular microstrip-fed monopole antenna, wave frequencies, IEEE Trans. Antennas and Propagation 71 (2) (2022)
Wirel. Pers. Commun. 77 (2014) 1363–1375. 1783–1792.
[44] M.A. Sufian, N. Hussain, A. Abbas, J. Lee, S.G. Park, N. Kim, Mutual coupling
reduction of a circularly polarized MIMO antenna using parasitic elements and
DGS for V2X communications, IEEE Access 10 (2022) 56388–56400.

10