A Metasurface-Based MIMO Antenna for 5G Millimeter-Wave Applications

This work presents a 4-element multiple-input multiple-output (MIMO) antenna with a single layered metasurface array for fifth generation (5G) millimeter-Wave (mm-Wave) communication systems. Each MIMO antenna element is composed of a $1\times 2$ array with a corporate feed network. Moreover, a metasurface array structure consisting of $9\times 6$ Circular Split Ring (CSR) shaped cells is employed to improve the gain and isolation between the MIMO antenna elements. The proposed antenna system is realized using 0.787 mm thick Rogers RT Duroid 5880 substrate. The performance of the antenna is investigated in terms of s-parameters, radiation characteristics, and MIMO parameters. The antenna operates at the mm-Wave frequency band ranging from 24.55 to 26.5 GHz. The incorporation of the metasurface layer enhanced the gain thus attaining a peak measured gain of 10.27 dBi. Additionally, isolation is also improved by 5dB after the employment of metasurface. The investigation of MIMO performance metrics such as Envelope Correlation Coefficient (ECC), Diversity Gain (DG), Channel Capacity Loss (CCL), and Mean Effective Gain (MEG) exhibits good antenna characteristics. The demonstrated radiation properties of the proposed antenna ascertain its suitability for the forthcoming 5G communication systems.

increases with the increasing number of radiators in an array, however larger array structures have inevitable losses across the power divider. Therefore, high gain antennas with a low profile are of substantial importance. Various antenna designs have been investigated to increase the gain maintaining a low profile. This includes metamaterial-based antennas, lens-coupled antennas, and Fabry-Perot cavity antennas [18]- [25]. Moreover, despite the attainment of high gain (necessary to mitigate the effects of increased attenuation and atmospheric absorptions), these antennas exhibit the same capacity as a single element due to a single feed port. To enhance the communication system capacity, MIMO technology is considered to be of great significance as it enables the multiple antennas to operate concurrently, thus improving the data rate, capacity, and reliability of the communication link. Several literary works reported MIMO antennas operating at mm-wave frequency band [26]- [32]. The antennas presented in these works demonstrate good MIMO performance with significant isolation. However, the gain exhibited by these MIMO antennas is comparatively low.
In recent years, metamaterials have been investigated extensively owing to their various electromagnetic characteristics not commonly present in natural materials [33]. Several works reported mm-wave MIMO antennas with metamaterial structures to enhance the antenna performance, particularly reduction in mutual coupling or antenna gain improvement [34]- [40]. The work in [34] proposed a 2 × 2 mm-wave MIMO antenna with bi-layered Frequency Selective Surface (FSS) superstrate for addressing the mutual coupling effects. Due to the FSS integration, a 6-12 dB improvement in isolation is obtained for the proposed MIMO antennas. Similarly, in [35] another technique is presented to decrease the coupling amid the MIMO antennas. In this work, metamaterial based corrugations are etched on the long edges of the substrate. As a result, enhancement of 12-15 dB isolation is observed. The presented antenna also obtains high gain. However, the substrate size of the proposed design is 21 mm × 85 mm, which is relatively large. In [36], a metamaterial based polarization-rotator (PR) wall is integrated to enhance isolation amid mm-wave DRAs. On average, 16 dB reduction in mutual coupling is attained by using this technique. The design proposed in [37] demonstrated an Electromagnetic band-gap (EBG) reflector positioned below the two-port MIMO antenna to enhance the antenna gain. The peak gain obtained for the proposed geometry is 11.5 dBi. Although the antenna exhibited good MIMO performance and high gain, however, the presented overall structure possesses comparatively large dimensions, which limit its suitability for the future miniaturized communication devices. Moreover, the bow-tie shaped mm-wave MIMO antenna in [38] is integrated with three pairs of metamaterial arrays to enhance gain. The maximum gain obtained by this antenna is 7.4 dBi. In [39], a two-port mm-wave MIMO antenna is reported with EBG structure for gain improvement. The introduction of EBG enhances gain by 1.9 dBi with a peak gain of 6 dBi. In [40], a DRA with four element MIMO configuration is proposed for mm-Wave applications. A metamaterial structure is printed on the top side of the DRA to enhance isolation. Due to the involvement of metamaterial, a 12-13 dB increase is observed in isolation between the MIMO elements. The maximum gain achieved for the operational frequency band is 7 dBi which is relatively low. This discussion reveals that these research studies either focus on gain enhancement or isolation improvement. However, both of these factors are important for the antenna to perform efficiently.
Considering the limitations as experienced by the antenna designs discussed above, this work proposed a four-element MIMO antenna for mm-wave 5G communication devices involving metasurface for gain and isolation enhancement. In order to enhance gain, a single element antenna is modified to array structure. Thus, each MIMO antenna comprises of 1 × 2 element array fed with a parallel feeding network. To further enhance gain and to decrease the coupling effects between the MIMO elements, a metasurface of periodically patterned 9 × 6 Circular Split Ring (CSR) shaped cells is placed above the MIMO antennas. The final proposed antenna covers the mm-Wave frequency band 24.55-26.5 GHz supporting 5G applications. A prominent increase in gain is observed with a peak gain value of 10.27 dBi. In addition, an improvement in isolation is attained after the introduction of the metasurface.

II. ANTENNA GEOMETRY AND DESIGN PROCEDURE
The proposed antenna design consists of a 4-element MIMO antenna structure with a substrate size of 30 mm × 43 mm. The substrate used is Rogers RT/Duroid 5880 with ε r = 2.2 and thickness of h = 0.787 mm. In addition, a metasurface array of periodically arranged 9 × 6 Circular Split Ring (CSR) shaped unit cells is placed above the MIMO antennas at the height of approximately 0.42 λ to optimize 51806 VOLUME 9, 2021 the antenna performance. The commercially available EM simulator CST Microwave Studio is used for the modeling and simulation of this design. Figure. 2 illustrates an overview of the stepwise design progression. The subsequent sections provide a detailed discussion on the stepwise design procedure, the optimization techniques employed, and relevant obtained results.

A. ANTENNA DESIGN
The modeling of the design starts from a single antenna element, as illustrated in Figure. 3 (a). In the first step, a rectangular patch antenna with inset feed and the full ground plane is obtained using well established mathematical equations [41], resonating for the 26 GHz frequency band with 0.9 GHz bandwidth, as shown in Figure. 3 (b). In the second step, the rectangular patch antenna is optimized by etching bow-tie shaped slots both vertically and horizontally. This optimization improves the impedance matching [42], [43], and a bandwidth of 1.18 GHz is obtained with a resonant band of 23.9-25.08 GHz. The third and final optimization step involves the etching of semicircular slots at the four corners and at the top side of the antenna. This antenna covers a band from 25-26.2 GHz with 1.2 GHz impedance bandwidth, as illustrated by the reflection coefficient curve in Figure. 3 (b). Although shifting of bands has been observed for three optimization steps due to change in electrical length of the antenna, however, the overall impedance bandwidth has improved. The optimized dimensions of the final single antenna are provided in Table 1.
In order to improve the radiation characteristics, the single antenna is modified to an array with two elements fed by a parallel feed line, as shown in Figure. 4 (a). The widths of the transmission lines are calculated using the characteristic equations given as (1-7) to match the impedance of the main feed line at 50 , whereas of the branched network at 100 [41]. where, where, where Z 0 represents the characteristic impedance of the transmission line, W a is the width of the feedline. ε r is the dielectric constant and ε reff is the effective permittivity. The length and width of the feed network are calculated by using the equations given below. Here, B is basically a constant used in the inverse design formula given in (5) for a microstrip line of the given characteristic impedance, and L a is the length of the feedline. The gap p a amid the two elements of an array is 1.437mm, nearly equal to 0.125λ at 26 GHz, which demonstrates the compactness of the design. Moreover, a vertical slot of length L c and width W c is incorporated at the ground plane. This design optimization from a single element to array enables the bandwidth enhancement, thus covering the frequency band 24.5-26.6 GHz, as exhibited in Figure. 4 (c).
The design is further progressed to obtain MIMO capability. The four antenna elements are placed along the opposite edges, with a substrate size of 30 mm × 43 mm × 0.787 mm, as illustrated in Figure. 4 (b). In addition, the ground plane has defected with horizontal slots exactly beneath the four MIMO antennas. The simulated reflection coefficient results plotted in Figure. 4 (c) for the MIMO antennas (S 11 , S 22 , S 33 , and S 44 ) exhibit nearly similar behavior, resonating for the band ranging from 24. 8-26.15 GHz. This demonstrates that the bandwidth obtained by the MIMO antennas is narrow as compared to the two-element array. This insignificant degradation in bandwidth is mainly due to the near field coupling effects. Despite the fact that bandwidth has narrowed for the MIMO structure, the attained bandwidth is sufficient enough to enable the mm-Wave operation. Figure. 4 (d) depicts the transmission coefficient plots, where significant isolation has been obtained for different MIMO antennas, where the minimum value near −33 dB is obtained between Ant1 and Ant2 over the entire operating frequency band.

B. METASURFACE ARRAY DESIGN
The reflection coefficient results of the MIMO antennas exhibit degradation in the antenna characteristics primarily due to coupling effects. In order to enhance the radiation performance of the antenna in terms of operating bandwidth, antenna gain, and isolation between MIMO elements, an eminent technique is to employ metasurface. A metasurface is basically a periodical placement of unit cells at a specific distance from each other.
The unit cell in this work is designed at 26 GHz. The substrate used is Rogers RT Duriod 5880 with 0.787 mm thickness. CST in the time domain is used to perform the simulations of a unit cell. The final optimized unit cell consists of two circular split ring resonators with a circular geometrical pattern in the centre, as illustrated in Figure. 5 (a). Figure. 5 (a) also depicts the step wise evolution of the unit cell. Figure 5 (b) illustrates the boundary conditions and the ports assignment to excite the unit cell. The reflection coefficient curve of the unit cell in Figure. 5(c) clearly exhibits unit cell is resonating at 26 GHz. Also the transmission coefficient plot shows that the unit cell loss is nearly zero at the resonating frequency, thus ascertains the full transmission at the desired band.
In order to further elaborate the behavior of metamaterial unit cell, the integral parameters such as permittivity (ε) and permeability (µ) are also investigated by applying the technique as provided in [44]. For permittivity and permeability extraction, equations (8-11) are used. (9) where S ii is the reflection coefficient, S ji is transmission coefficient, k o is the wavenumber, z is the normalized impedance, d is the thickness of the material, and η is the refractive index.
where ε is the relative permittivity, while µ is the relative permeability. Figure. 6 depicts the permittivity and permeability plots of the unit cell. It is observed that both permittivity and permeability tend to zero at the 26 GHz, which demonstrates zero refractive indexes [45]. Later, a metasurface array is obtained by periodically placing 9 × 6 unit cells at a spacing of g, as Figure. 7 (a) illustrates. The dimensions of the metasurface are W m × L m . The metasurface is stacked above the MIMO antennas, as illustrated in Figure. 7 (b), at a distance of h t to enhance the antenna performance. After incorporation of metasurface, the MIMO antennas show significant improvement in performance, as exhibited in Figure. 8 (a) and (b). The reflection coefficient plots of MIMO antennas before and after metasurface employment in Figure.  As other antennas exhibit the nearly same behavior, therefore the curves for only Ant1 and Ant4 are shown here. While the transmission coefficients curve in Figure. 8 (b) depicts the minimum isolation among Ant1 and Ant2 with metasurface, which is approximately −38 dB. At the same time, the Ant1 and Ant3 exhibit minimum isolation of −40 dB over the entire operational frequency band. The same behavior is observed for Ant2 and Ant4. The maximum isolation is achieved for Ant2 and Ant3 approaching a value of −45 dB. These transmission coefficient curves clearly provides evidence that minimum isolation is improved by 5dB after metasurface employment.
In order to further elaborate on the contribution of metasurface in isolation enhancement by repressing the surface waves [33], surface current distribution is investigated at 26 GHz. The surface current distribution is shown in Figure. 9 exhibits that the near-field coupling effect is cancelled as the current coupled to unit cell traverses in the opposite direction in half of the ring as well as in adjacent rings. It is also observed that without metasurface when Ant1 is excited while other antennas are terminated with 50 load, the current is mainly coupled to Ant2, whereas lower coupling is noticed for Ant3 and Ant4. With metasurface placed above, the coupled field between antenna elements is primarily distributed on the unit cells of the metasurface. Consequently, coupling between antenna elements is minimized and, the isolation is improved significantly.
Similarly, Figure. 10 provides a comparative analysis of the simulated broadside gain of antenna for different steps of design evolution at various frequencies of the operating band. It is observed that the gain of the antenna is increased at the first step by modifying the single element antenna to a two element array structure, from 6.8 dBi to 8.1 dBi at 26 GHz. The antenna gain for the MIMO configuration nearly remains the same, however after employing metasurface, gain increases by approximately 2.2 dBi. Here the metasurface acts as a superstrate layer, which generates a resonant cavity effect resulting in gain enhancement of the antenna system [46], [47]. Thus, the peak simulated gain obtained is 10.37 dBi for Ant1 and 10.15 dBi for Ant4 at 26 GHz.

III. EXPERIMENTAL RESULTS
The proposed MIMO antenna with metasurface is fabricated, as shown in Figure. 11 (a), for experimental validation. In order to stack the metasurface above the antennas, Nylon spacers are used. The comparative investigation of simulated and measured results is provided in the succeeding sections.

A. REFLECTION COEFFICIENT AND ISOLATION
For S-parameters measurement, the Rohde and Schwarz ZVA 40 Vector Network Analyzer (VNA) is used. Simulated  and measured S-parameter results are demonstrated in Figure. 12 (a), (b). The measured reflection coefficient results for Ant1 and Ant4 are illustrated in Figure. 12 (a) demonstrates that Ant1 resonates for the 24.7-26.4 GHz frequency band with −10 dB impedance bandwidth of 1.7 GHz. Similarly, the operational frequency band for Ant4 ranges VOLUME 9, 2021    Figure 12 (b) shows the measured transmission coefficient curves demonstrating isolation of Ant1 with Ant2 and Ant3. It is observed that measured and simulated results have minimal discontinuities, mainly due to flaws in fabrication and losses across the cable.

B. RADIATION PATTERN AND GAIN
The far-field results of the antenna are obtained in the anechoic chamber at the Electromagnetic Wave Technology Institute, Seoul, Korea. A standard gain horn antenna (SGH-series) is used as a transmitter (TX), while the proposed antenna is measured as a receiver (RX). In order to provide stable power reception, amplifiers were used. The antenna under test is rotated to obtain measurements at different orientations. The radiation patterns in the XZ and YZ plane for Ant1 and Ant4 at 26 GHz are shown in Figure 13 (a) and (b). It is observed that in the XZ plane, Ant1 exhibits maximum radiation at θ = 25 • , whereas Ant4 demonstrates the main beam in the direction of θ = −16 • .
The measured and simulated antenna gain values of the proposed antenna with metasurface at different frequencies are tabulated in Table 2. It is demonstrated that the Ant1 attained the measured peak gain of 10.21 dBi at 26 GHz. Whereas, for Ant4, the maximum value of gain measured is  10.27 dBi. Thus, coherence is observed amid the measured and simulated results.

IV. MIMO PERFORMANCE
In order to analyze the MIMO performance of an antenna, different parameters, including ECC, DG, MEG, and CCL, are of great significance. For the proposed antenna design, these parameters are discussed in the following subsections.

A. ECC AND DG
ECC for any MIMO system determines the correlation between the individual elements in terms of their individual characteristics. For ECC calculation, equation (12) is used as given in [48], [49].
where S ii is the reflection coefficient of antenna i while S ij (where i = j) is the transmission coefficient for antennas i and j. The ECC plot for the proposed design is illustrated in Figure. 14 clearly shows that the obtained values are far below the practically acceptable value, which is 0.5 for wireless systems. Similarly, diversity gain is another substantial MIMO performance metric. It demonstrates the extent of transmission power reduction after using a diversity scheme. Diversity gain is obtained using the relation (13), as provided in [50]. Figure. 14 also depicts the plots of diversity gain. Diversity gain of approximately 9.98 dB is attained for Ant1 & Ant3,

B. MEG
Mean Effective Gain also has significance in analyzing the performance of MIMO antennas, which is actually the ratio of mean received power to the mean incident power. Equation (14) is used to estimate the value of MEG as given in [32].
The standard value of the Mean effective gain (ME f G) for good diversity performance should be The numerical values of MEG obtained for the proposed MIMO antennas are given in Table 3.

C. CHANNEL CAPACITY LOSS (CLL)
In general, with the increase in a number of antenna elements in the MIMO system, the channel capacity also increases. CCL determines the channel's capacity loss due to the correlation between the MIMO links. CCL can be calculated using the equations (15)-(18) as given in [32].
where a k is correlation Matrix. a k = σ 11 σ 12 σ 21 σ 22 (16) where, where σ ii and σ ij are the correlation coefficients between the antenna elements ii and ij of MIMO antenna. CCL for the proposed MIMO design is depicted in Figure. 15. It is clearly shown that CCL is below the practically acceptable value of 0.4 bit/s/Hz over the operating band. Thus the high throughput of the proposed system is ascertained.

V. COMPARISON WITH THE RELATED WORK
The proposed 5G mm-Wave MIMO antenna with metasurface array is compared with the recently reported literary works, as shown in Table 4. It is observed that the proposed MIMO antenna with metasurface array outperforms the other reported works in terms of isolation. Moreover, the antenna gain exhibited by proposed MIMO antenna is significantly higher than the works [32], [34], [36], [38]- [40]. Although the works presented in [35], [37] achieved higher gain as compared to the proposed antenna, however, these MIMO structures have only two elements, and also MIMO performance analysis is not provided. The presented antenna design demonstrates improved MIMO performance in comparison to other antennas discussed here. Consequently, the pertinence of the proposed MIMO antenna is ascertained for current and future communication devices. YASAR AMIN (Senior Member, IEEE) received the B.Sc. degree in electrical engineering with a focus on telecommunication and the M.B.A. degree in innovation and growth from the Turku School of Economics, University of Turku, Finland, and the M.Sc. degree in electrical engineering with a focus on system on chip design and the Ph.D. degree in electronic and computer systems from the KTH Royal Institute of Technology, Sweden, with the research focus on printable green RFID antennas for embedded sensors. He is currently a Professor and the Chairman of the Department of Telecommunication Engineering, University of Engineering and Technology Taxila, Pakistan. He also serves as the Director of Embedded Systems Research and Development Centre. He is the founder of Agile Creative Technologies for Smart Electromagnetic Novel Applications (ACTSENA) Research Group. He has authored or coauthored more than 100 international technical papers in conferences and journals. His research interests include the design and application of multiple antenna systems for next generation mobile communication systems, millimeter-wave and terahertz antenna array, implantable and wearable electronics, and inkjet printing technology in microwave applications. He is a member of more than a dozen international professional societies and the Fellow of PAE. VOLUME 9, 2021