High‐gain, high‐isolation, and wideband millimetre‐wave closely spaced multiple‐input multiple‐output antenna with metamaterial wall and metamaterial superstrate for 5G applications

Mohammed Amin Honarvar, Department of Electrical Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Isfahan, Iran. Post code: 8514143131 Email: Amin.Honarvar@pel.iaun.ac.ir Abstract A method to significantly increase the gain and isolation of a wideband multiple‐input multiple‐output (MIMO) antenna using metamaterial structures is reported. The proposed metamaterial medium includes just one wall and one superstrate, which are fixed in xz and xy planes of the MIMO antenna, respectively. The dimensions of the proposed MIMO antenna are 13�13�3.5 mm at 30 GHz. A comparison between a simulated MIMO antenna and a conventional microstrip antenna demonstrates the good performance of the proposed metamaterial environment. The corresponding return‐loss of the antenna is better than 10 dB over 28–32 GHz for 5G applications. The maximum simulated gain of the antenna is 17.1 dB at 30 GHz, generating a maximum gain enhancement of 11.8 dB in comparison with a MIMO antenna without any metamaterial structures. The isolation (insertion‐loss) is 36.7 dB at 30 GHz, which has improved by more than 29.4 dB compared to the conventional one.


| INTRODUCTION
Multiple-input multiple-output (MIMO) systems are a more appropriate choice than single-input single-output systems because of high spectral efficiency and high reliability [1]. In these systems, multiple antennas in the transmitter and receiver are used to create independent paths to transmit multiple signals simultaneously. Since the probability of the signal fading in all these paths is low, by combining the signals from the different paths, the overall fading rate can be dramatically reduced [2]. It is common to use microstrip structures to design antennas [3]. But the low gain and the mutual coupling between elements in closely spaced MIMO designs are the most important challenges.
For high-speed data transfer, operating at high frequencies is inevitable. But due to high losses because of atmospheric absorption, high-gain antennas are required [4]. On the other hand, the modern telecommunication packages generally design as integrated circuits. So, the proximity of the antenna elements increases the mutual coupling. But most researches focus on only one of the aspects above, meaning the improvement of both aspects at the same time is not covered.
In recent years, methods such as microstrip antenna arrays [5], use of parasitic elements [6], substrate integrated waveguide technology [7], and electromagnetic band-gap (EBG) structures [8] have been used to enhance the gain of microstrip antennas. Also, to reduce the MIMO antennas mutual coupling, methods such as defected ground structure (DGS) method [9], parasitic networks [10], EBG structures [11] and current localization concept [12] are also employed.
Over the past few decades, the use of metamaterial structures to increase gain and reduce mutual coupling has become common. The utilization of metamaterial slabs [13][14][15] and surface metamaterial arrays [16][17][18] to increase gain have been reported. The authors have designed a microstrip highgain bow-tie antenna with the help of a novel metamaterial region for H-band applications in [13]. In that design, with the use of six rotated two-sided planar metamaterial slabs, a peak gain of 10.21 dB was achieved at 7.5 GHz.
A method for simultaneously increasing gain and reducing mutual coupling is presented. Also, to increase the bandwidth for the 5G application (28-32 GHz), the triple lines method, has been used. To enhance the gain and isolation, arrays of a novel metamaterial unit-cell, Combined Metamaterial Structure and Parasitic Ring (CMSPR), were used. The utilization of the proposed unit-cells yields a maximum gain of 17.1 dB at a central frequency of 30 GHz. Also, the mutual coupling (S 21 ) is less than -36.7 dB at 30 GHz. The final dimensions of the proposed MIMO antenna are equal to 13�13�3.5 mm 3 .

| GAIN, ISOLATION, AND BANDWIDTH ENHANCEMENT METHODS
To design a high-gain, high-isolation, and wideband MIMO electric dipole antenna (EDA), a metamaterial environment consists of the proposed CMSPR unit-cells and the triple lines are employed as shown in Figure 1. The proposed metamaterial environment is divided into two media. The metamaterial superstrate medium (CMSPR superstrate) on top of the antenna (xy plane) horizontally and the metamaterial wall (CMSPR wall) in the yz plane. To excite the Lorentzian magnetic resonance, the H-field should be perpendicular to the axis of CMSPR unit-cells, which is realized due to using the EDA. In this work, an EDA has selected, which can be considered as a quasi-transverse electromagnetic (TE) source, where the H-field is along the propagation direction. The metamaterial wall acts as a spatial filter that prevents the transfer of electromagnetic waves between the antenna elements [13].
According to Figure 1, the E-field is along the x-axis. So, the transfer function (TF ) will be equal to [22]: ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ε CMSPR � μ CMSPR p � 1 − r 2 exp � − j2πf z 0 ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where μ CMSPR is permeability, ε CMSPR is permittivity, and f shows operating frequency. t 1 , t 2 , and r correlate with the transmission and reflection coefficients at z = 0 and z = z 0 , respectively. Figure 2 shows the phase change of Equation (1), in terms of the thickness of metamaterial ambiances (z 0 /λ 0 ) with a different value of permeability. The phase change in the proposed CMSPR medium is much lower than the free space (μ CMSPR = μ 0 ) and the electromagnetic fields propagate at a higher phase velocity inside the CMSPR medium. Thus, the interactions between CMSPR superstrate increase the antenna's aperture size, which results in improved antenna gain. The interference of electromagnetic waves in a MIMO antenna, which reduces its quality performance, can be minimized by using metamaterial structures because of their bandgap [22]. So, the electromagnetic energy will not transfer anymore due to these waves being suppressed by the metamaterial wall. As shown in Figure 1, the metamaterial wall is embedded between two EDA elements. The existence of CMSPR wall in the space between EDA elements, relies on its band-gap, and can seriously decrease the mutual coupling.
Finally, the triple lines method mentioned in [23] is utilized to increase the proposed antenna bandwidth. Using the optimization algorithms or parametric study, the optimum distance of the triple lines can be determined. With this value, the maximum bandwidth is achieved. For the proposed MIMO antenna, the optimum distance between the triple lines is 0.4 mm which a bandwidth of 4 GHz (28-32 GHz) is obtained for 5G applications.

| THE PROPOSED METAMATERIAL UNIT-CELL (CMSPR UNIT-CELL) ANALYSIS
In this section, the proposed CMSPR unit-cell to create a metamaterial environment is presented and is shown in Figure 3. This metamaterial unit-cell, derived from Peace-Logo Planar Metamaterial (PLPM) design, was introduced in [24]. This structure contains a Modified PLPM (MPLPM) and a Parasitic Ring (PR). The PR was used to increase the stop F I G U R E 1 Configuration of the proposed MIMO antenna with triple lines and metamaterial structures. CMSPR, combined metamaterial structure and parasitic ring; MIMO, multiple-input multiple-output F I G U R E 2 Phase of transfer function for different values of μ CMSPR 380bandwidth (band-gap) of this unit-cell. Using this part, the band-gap is increased to about 5 GHz compared to the nonring sample. In this study, CST Microwave Studio 2019 is used full-wave for simulation of the structures on a Rogers RT/Duroid 5880 dielectric with ε r = 2.2, tan δ = 0.0009, and thickness of h = 0.254 mm.
The initial equivalent circuit model of the proposed unitcell is shown in Figure 4a. The MPLPM part is modelled by the C 0 and L 0 elements. This part acts like an LC resonator. C 0 is a parallel capacitor due to the presence of dielectrics and MPLPM metal strips. L 0 is the MPLPM inductor due to applying a magnetic field and the creation of surface current. The PR part is also modelled by the capacitor C PR and the inductor L PR elements. Also, S is the magnetic coupling coefficient. To simplify the calculations, the equivalent circuit shown in Figure 4b is determined. M is the mutual impedance, which can be calculated according to Equation (2).
The  Figure 4c. It is observed that assuming |S 21 | < -10 dB, a wide band-gap is obtained from approximately 25 to 40 GHz.
For full-wave analysis, two waveguide ports are adjusted in the x-direction, and perfect electric conductor, as well as perfect magnetic conductor boundaries, are assigned in the xz and xy planes, respectively. The S 11 (return-loss) and S 21 responses of the CMSPR unit-cell are shown in Figure 5a. It is observed that two magnetic resonance frequencies are generated at 27 and 34 GHz. Wide bandwidth is one of the most important advantages of this structure. In Figure 5a, the effect of the PR part is visible. Without the ring, this unit-cell is single-band, and assuming |S 21 | < -10 dB, its band-gap is equal to 5 GHz (28-33 GHz). But with the use of PR, the stop bandwidth has increased dramatically. So, the bandwidth obtained is 10 GHz To calculate the relative permeability and permittivity parameters the Equations (3) and (4) presented in [24] are used. In these equations, β 0 is the wave number and h is the substrate thickness. The achieved permeability and permittivity are depicted in Figure 5b. According to Figure 5b, both the μ r and ε r parameters have values less than zero in the frequency range 25-35 GHz.
In Figure 5c, the dispersion diagram of the proposed CMSPR unit-cell is shown. It is observed that a wide band-gap is created between the frequencies of 25 to 35 GHz.
In Figure 6a, the S 21 of the proposed unit-cell is shown when ports are placed in different directions. Also, in Figure 6b, the μ r and ε r parameters are illustrated. It is seen that the figures correspond with each other in all the diagrams. Therefore, the unit-cell can be used both horizontally and vertically depending on the MIMO EDA.
The return-loss, insertion-loss, gain, and radiation efficiency of the proposed MIMO EDA in comparison with the conventional one without any metamaterial element are shown in Figure 8. According to Figure 8a, the isolation has been improved by 29.4 dB with the proposed idea. The S 21 parameter at 30 GHz for the conventional and the proposed antenna are −7.3 and −36.7 dB, respectively. The maximum and minimum isolation obtained by the proposed MIMO antenna in the operating frequency band are 55 and 23 dB, respectively. However, for the conventional antenna, these values are 11 and 4 dB respectively. Based on Figure 8b, for the conventional MIMO antenna, the gain is 5.3 dB at 30 GHz. But for the proposed antenna, the gain is 17.1 dB. Therefore, the gain has enhanced by 11.8 dB. Obviously, use of the CMSPR wall does not harm the antenna gain. Also, according to Figure 8c, the radiation efficiency at 30 GHz, for the proposed antenna and the conventional one, are 87.5% and 96.3%, respectively. Reduction in radiation efficiency is generally due to dielectric and ohmic losses. Figure 9a shows the H-field of the antenna and its effect on the CMSPR superstrate. The antenna H-field should be in the +z direction which is perpendicular to the CMSPR superstrate plane as shown in Figure 9a. As a result, it induces a surface current distribution on the metamaterial superstrate elements as shown in Figure 9b. This indicates that the CMSPR superstrate acts as a parasitic element that affects the radiation performance of the EDA. Therefore, this effect has enhanced the proposed antenna gain.
The Current density distribution over the two EDA elements with the CMSPR wall at 30 GHz is shown in Figure 10. The surface current is suppressed by introducing the metamaterial wall between the antenna elements. This confirms that the proposed CMSPR wall acts as an effective decoupling structure. Figure 11a shows the E-field distribution for both conventional and CMSPR-loaded MIMO antenna. The E-field strength at the edges of the lines and substrate of the conventional EDA is greater than the proposed antenna. Moreover, the E-field strength at the surface of the exciting antenna in both cases is approximately equivalent. Therefore, use of the CMSPR medium does not have a significant effect on the gain and directivity of the antenna. Figure 11b also shows the distribution of the H-field at 30 GHz. As expected, the H-field is perpendicular to the unit-cells embedded in the various planes. The proper performance of the proposed antenna indicates that unit-cell placement in different directions does not have a destructive effect on improving the antenna characteristics.

| PARAMETRIC STUDY
In this section, parametric studies on the effects of the proposed metamaterial environments on the MIMO antenna performance are presented. The first is to determine the optimal distance between the triple lines (d 1 ). This method is used to increase the MIMO antenna bandwidth. The bandwidth variations for different values of d 1 are shown in Figure 12a. Based on [23] where with the help of the triplelines idea, a wideband power divider is designed, by changing the distance between lines, the maximum possible bandwidth of the antenna can be achieved. As shown in Figure 12a, the bandwidth has also changed as the d 1 varies from 2 to 4 GHz. It is observed that for d 1 = 0.4 mm, the maximum bandwidth (4 GHz) is achieved. Therefore, in the following and for other simulations, this parameter will be considered constant and equal to 0.4 mm. To investigate the effect of the proposed CMSPR superstrate on the gain, the gain modifications are also reported at 30 GHz in Figure 12b. According to this figure, the gain varies from 5.2 to 5.6 dB. In Figure 12c, the effect d 1 changes on the antenna cross-polarization at 30 GHz are shown. It is observed that these changes do not have a detrimental effect on the radiation pattern of the proposed MIMO antenna. It can be seen that the triple lines can also help slightly increase gain.
So far, a parametric study is provided which is based on the number of CMSPR rows of the metamaterial wall. Depending on the size of the MIMO antenna, along the x-axis, we can just can eventually use 5 CMSPR unit-cells. In Figure 13(a), the variation of the isolation for 1 to 5 unit-cells is reported at 30 GHz. According to this figure, by using just a unit-cell, the isolation improves about 7.3 dB. It is observed that with the increasing number of unit-cells, isolation also improves. Note that 5 CMSPR unit-cells are selected to improve isolation. Also, this selection does not increase the size of the antenna.  Figure 13b, it is also shown that how the gain at 30 GHz is affected if the number of wall unit-cells increases. It can be seen that this wall has no detrimental effect on the gain. For one unit-cell, the gain is 5.5 dB. According to Figure 1, although there is a limit along the x-axis, it is possible to increase the number of metamaterial rows along the z-axis. So, the effect of increasing the number of CMSPR rows on the proposed metamaterial wall on the gain is illustrated in Figure 13c. Given this figure, the isolation improves at a rate of 3.5 dB by increasing the number of rows from 1 to 10 at 30 GHz. But note that under these conditions, antenna dimensions will be more than doubled. Therefore, by selecting just one row, isolation is achieved about 37 dB at 30 GHz. However, the dimensions of the antenna have not increased.
The parametric investigation of the metamaterial superstrate is now discussed. In Figure 14(a), the effect of increasing the number of metamaterial rows on the gain at 30 GHz is reported. Due to the limited size of the superstrate, only five metamaterial rows can be added. According to Figure 14a, it can be seen that with 5 rows, the gain 17.1 dB is obtained.
Note that the gain is also calculated for CMSPR rows numbers 6, 7, and 8 (with increasing dimensions). But this large increase in antenna size improves the gain by just 0.9 dB. It is therefore advisable to select a 5�4 array of CMSPR unit-cells for the superstrate.
Finally, in Figure 14b, the effects of increasing the metamaterial rows of the proposed superstrate on the isolation at 30 GHz are shown. It can be seen that these changes do not have a significant effect on isolation.

| FABRICATION AND MEASUREMENT RESULTS
Finally, to affirm the wideband, high-gain, and high-isolation performance introduced in the past sections, the proposed MIMO antenna is fabricated and is shown in Figure 15.  Figure 16a. Assuming |S 11 | < -10 dB, the bandwidth of the proposed MIMO antenna is 28-32 GHz for 5G applications. The measurements were conducted using Agilent network analyzer HP872. Figure 16b shows the simulated and measured MIMO antenna gain. The simulated and measured gain at 30 GHz, is 17.1 and 17 dB, respectively. Figures 17a,b demonstrated the simulated and measured normalized radiation patterns for both Eand Hplanes at 30 GHz. To validate these results the measurement setup in Figure 17c was used. According to this figure, the horn antenna is the transmitter and the proposed MIMO antenna is the receiver. Correlation between the simulated and measured outcomes is good.
In the end, to provide a comprehensive comparison between the proposed MIMO antenna and other designs reported in the articles, Table 1 is prepared. It is obvious that the proposed MIMO antenna has significant benefits such as high-gain, high-isolation, wideband, and compact size.

| CONCLUSION
A compact high-gain, high-isolation, wideband MIMO antenna has been designed and fabricated. The proposed EDA consists of just one type of metamaterial unit-cell. The metamaterial environment included a superstrate and a wall for gain and isolation enhancement with the help of the proposed metamaterial unit-cells. The simulated peak gain of the proposed MIMO antenna equals 17.1 dB at 30 GHz. Also, the isolation was about 36.7 dB at 30 GHz, which improved 29.4 dB compared to the conventional MIMO antenna one.  -387