Dual-Band 2× 2 MIMO Antenna with Compact Size and High Isolation Based on Half-Mode SIW

School of Communication and Information Engineering, Shanghai University, Shanghai 200444, China Key Laboratory of Specialty Fiber Optics and Optical Access Networks, (e Shanghai Institute for Advanced Communication and Data Science, Joint International Research Laboratory of Specialty Fiber Optics and Advanced Communication, Shanghai University, Shanghai 200444, China School of Engineering and Digital Arts, University of Kent, Canterbury CT2 7NZ, UK


Introduction
Antenna systems that support multiple wireless standards and have compact size are highly desirable for future portable devices. Multiple-input multiple-output (MIMO) antenna technology has become essential for future portable devices owing to its ability to increase data throughput without increasing the power and bandwidth [1]. erefore, developing compact multiband antennas to support multifunctional MIMO systems is necessary. High isolation among the elements of a compact dual-band MIMO antenna is important for the performance of a communication system. Some dual-band MIMO antennas obtained using different decoupling techniques have been introduced in the literature; these are generally summarized into two types. e first type involves decoupling antenna elements without inserting any decoupling structures; here, the decoupling could be realized using spatial [2], pattern [3], and polarization diversities [4]. Spatial diversity can be achieved by arranging two or more antenna elements such that there is a wide space between them; therefore, such MIMO antennas have a large size. Pattern diversity can be realized by discriminating antenna elements over an angled space, whereas polarization diversity can be achieved by orthogonally arranging the antenna elements. Pattern and polarization diversities are preferred because the elements can be arranged close to each other. However, the coupling is difficult to control when more than two closely spaced antennas are employed. e second type of decoupling for MIMO antenna involves inserting a decoupling structure between the antenna elements to block or suppress the coupling, which includes defected ground structure [5], complementary split-ring resonator [6], electromagnetic bandgap [7], neutralization line [8], T-shaped stub [9], and ring strip [10]. However, these structures increase the space between antenna elements, which results in larger antenna sizes. High isolation can be achieved by combining the two methods above [11][12][13][14][15].
In this paper, a dual-band antenna element is proposed by loading a rectangular patch outside the aperture of a halfmode substrate integrated waveguide (HMSIW) cavity. By combining the spatial and polarization diversities and using small neutralization lines (SNLs), a close-spaced 2 × 2 MIMO antenna with high isolation and dual-band operation is suggested. To validate the proposed antenna, a prototype is fabricated and its performance is measured. is paper is arranged as follows. In Section 2, the design procedure of the proposed dual-band antenna element is discussed. In Section 3, four-element MIMO antennas are introduced and their corresponding decoupling mechanisms are explained. e fabrication and measurement results are presented in Section 4. e conclusion of the proposed work is presented in Section 5.

Dual-Band Antenna Element
A dual-band antenna element obtained by loading a rectangular patch outside an HMSIW cavity is proposed, as shown in Figure 1.
e antenna element has only one substrate layer, and the bottom side of the substrate is a metallic ground plane. e HMSIW antenna is fed by a coaxial probe, whereas the patch antenna is excited through proximity coupling by the radiating aperture of the HMSIW antenna. e HMSIW antenna generates the lower band, whereas the loading patch generates the upper band. e geometrical size of the proposed dual-band antenna element is shown in Figure 2. e size of the HMSIW cavity is W × W and the dimensions of the patch are L p × W p . e coupling gap is designated as G. To be equivalent to a conventional metallic cavity, the diameter d and pitch s of the HMSIW cavity via holes must meet the conditions of d/ s ≥ 0.5 and d/λ 0 ≤ 0.1 [16]. e antenna element was designed on a single-layer substrate of Rogers Duroid 5880 with a relative dielectric constant of 2.2, tangent loss of 0.0009, and thickness of 1.57 mm. A high-frequency structure simulator was used to analyze and simulate the antenna. e optimum parameters of the proposed antenna element are W � 40 mm, L p � 18 mm, W p � 20 mm, and G � 0.4 mm. e simulated reflection coefficient of the proposed dual-band antenna is depicted in Figure 3. e two simulated resonances appear at 4.46 and 5.38 GHz. Figure 4 depicts the simulated electric field distributions at f L � 4.46 GHz and f H � 5.38 GHz. Figure 4(a) illustrates that the power is concentrated along the edge of the aperture of the HMSIW cavity, which implies that the resonant mode at f L � 4.46 GHz is generated by the HMSIW cavity. In contrast, the electric field is mainly concentrated along the two edges of the patch at f H � 5.38 GHz, as observed in Figure 4(b).

Four-Element Antenna
Configuration. An initial configuration of the array is proposed by replicating and rotating the antenna element described in Section 2. As shown in Figure 5, the element pairs (1, 4) and (2, 3) provide spatial diversity, whereas the orthogonal arrangement of the pairs (1, 2) and (1, 3) results in polarization diversity. Good isolation was achieved by exploiting the spatial and polarization diversities. e ground planes of the antenna elements were separated by distance g � 2 mm to further improve the isolation. e dimensions of the MIMO antenna are 82 × 82 mm. e simulated S-parameters of the four-element MIMO antenna are depicted in Figure 6. Owing to the symmetrical arrangement, |S 12 | � |S 42 | � |S 34 | � |S 13 | and |S 14 | � |S 23 |; therefore, only |S 11 |, |S 12 |, |S 13 |, and |S 14 | are presented here. S 11 demonstrated that the antenna resonated at 4.46 and 5.38 GHz, which are the same resonance frequencies as those of the element in Figure 3. e minimum isolation levels were better than 25 dB in both bands. Figure 7, four SNLs were inserted between the radiators of the MIMO antenna to further reduce the coupling current without increasing the size of the original MIMO antenna. e SNLs created additional current paths to transfer some of the current from one element and feed it back to another element with a suitable magnitude and an opposite phase, which eliminates the coupling. Each SNL was placed away from the open end of the antenna element in a low impedance area, where the current has the highest intensity.

Isolation Enhancement of the Four-Element Array. As shown in
e decoupling current can be controlled by adjusting the location of the SNLs. Figure 8 shows the effects of location d nl on the isolation. Isolation higher than 35 dB in both bands was achieved when the location d nl was 5 mm.
To further understand the contribution of the SNLs to isolation enhancement, the simulated current distributions with and without SNLs at 4.46 and 5.38 GHz are illustrated in Figure 9. e current intensity in the other elements (2, 3, and 4) is reduced when element 1 is excited and the SNLs are introduced. Figure 9 shows that the SNLs are more effective in the lower band than in the upper band. Figure 10 presents the simulated S-parameters of the 2 × 2 MIMO antenna with the SNLs structure.
e two simulated resonances appear at 4.46 and 5.38 GHz and the isolation in both bands is higher than 35 dB. is means the isolation in both bands increased by around 10 dB in comparison with the MIMO antenna without SNLs.

Fabrication and Measurement
As shown in Figure 11, a prototype of the proposed MIMO antenna was fabricated to verify its effectiveness.

2
International Journal of Antennas and Propagation 4.1. S-Parameters. An Agilent N5227A vector network analyzer was used to measure the S-parameters. As shown in Figure 12, the measured S 11 resonated at 4.43 and 5.39 GHz. e measured minimum isolation levels in both bands are higher than 35 dB. ese results are in good agreement with the simulation results.

Radiation Patterns.
e radiation patterns of the fabricated antenna were measured in an anechoic chamber and compared with the simulation results. Owing to the symmetry of the antenna structure, only the radiation patterns of element 1 are shown. As shown in Figure 13, the simulated and measured radiation patterns at 4.46 and 5.38 GHz are in good agreement. e measured gains are 6 and 6.4 dBi, respectively, and the cross-polarization levels are lower than −14 dB in both bands. ese results are in good agreement with the simulation results.

Diversity Performance.
e diversity performance of the proposed MIMO antenna is described in terms of envelope  ECC is considered a paramount factor in the analysis of MIMO antennas. For better performance of the MIMO antenna, ECC < 0.5 across the operating frequency is essential.
e ECC can be computed by using either the S-parameters [15] or three-dimensional far fields [12]:   where F i (θ, φ) is a complex vector demonstrating the field radiating from the ith element. e symbols "·" and " * " indicate the Hermitian product and the complex conjugate, respectively. In this paper, the ECC is computed using both methods. For brevity, only ρ 12 and ρ 14 are presented here. As shown in Figures 14 and 15, the computed ECC is less than 0.005 across the two operating bands, which satisfies the requirement ECC < 0.5 and indicates good system performance. e 3D farfield method for finding ECC gives more accurate results [17]. DG determines the enhancement achieved by using multiple antenna systems over using single antenna systems. e relationship between ECC and DG can be described as [12] DG � 10 (2) It can be noted from (2) that a low ECC leads to ensure high DG. In an ideal case (ρ ij � 0), the DG is equal to 10 dB. As shown in Figure 16, the computed DG for the highest ECC values in the lower and upper operating bands is greater than 9.99 dB which guarantees good diversity performance.
TARC is the ratio of the square root of total reflected power divided by the square root of the total incident power. It can be computed using the following relation [18]: Frequency (GHz) S11 S12 S13 S14  -30 S12 and S14 (dB) Figure 8: Simulated S12 and S14 of the MIMO antenna with variations in d nl .   International Journal of Antennas and Propagation S11 S12 S13 S14  Frequency (GHz) S11 S12 S13 S14 International Journal of Antennas and Propagation where b i and a i represent reflected and incident signals, respectively, and N is the number of antenna elements. e computed TARC is shown in Figure 17. As can be seen, the TARC is better than -10 dB across the two operating bands.
MEG is defined as the ratio of the mean received power to the mean incident power of the antenna. If the antennas are 100% efficient, the maximum MEG is -3 dB [19]. It can be computed using the following equation [20]: where G θ and G φ are, respectively, θ and φ polarized components of the antenna power gain pattern. Γ is the cross-polarization discrimination of the incident field. e ratios of the MEG should satisfy the following condition in order to guarantee good channel characteristics [21]: In our design, average MEG of -3.2 dB is achieved for all the four elements. e ratios of MEG i /MEG j are around 0.937 which satisfies condition (5).  International Journal of Antennas and Propagation CCL caused by the correlation can be expressed as [22] C loss � − log 2 det Ψ R , where Ψ R is the correlation matrix. Figure 18 shows that the proposed MIMO antenna achieves lower than 0.2 CCL values (ideally, CCL should be < 0.4 bits/s/Hz) across the two operating bands. We can conclude that the diversity performances of the proposed MIMO antenna are suitable for a high-quality MIMO antenna system. e performance of the proposed MIMO antenna is compared with that of the recent four-element MIMO antenna in Table 1. Our 2 × 2 MIMO antenna has the highest isolation and gain. e proposed antenna also has the smallest edge-to-edge element spacing. e size of this work is smaller than that of the antenna in [14] and approximately equal to antenna in [15] and larger than that of the antennas in [11][12][13]. Our design has the smallest envelope correlation coefficient which guarantees good diversity performance. In this design, we used the coaxial probes to excite the dual- band antenna elements. In summary, our proposed 2 × 2 MIMO antenna has the highest isolation, highest gain, smallest edge-to-edge element spacing, and lowest envelope correlation coefficient.

Conclusion
In this paper, a close-spaced dual-band 2 × 2 MIMO antenna is proposed and fabricated. Good isolation was achieved by   exploiting spatial and polarization diversities. An isolation enhancement of around 10 dB was achieved by inserting SNLs at the center of the MIMO antenna between the radiators of the antenna elements without increasing the size of the original MIMO antenna. With only 0.036 λ 0 edge-toedge distance between elements, the measured isolation in both bands was higher than 35 dB. Moreover, the diversity performances of the proposed MIMO antenna are suitable for a high-quality MIMO antenna system.

Data Availability
All simulated and measured data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that they have no conflicts of interest.