Miniature Dual-Band Substrate Integrated Waveguide Slotted Antenna Array for Millimeter-Wave 5G Applications

A compact substrate integrated waveguide (SIW) antenna array that operates at 28GHz and 38GHz is proposed for fifth generation (5G) applications. *e proposed array consists of four SIW cavities fabricated on one single layer of substrate. Each cavity implements a rhombic slot and a triangular-split-ring slot, resonating on TE101 and TE102 modes at 28GHz and 38GHz, respectively. In comparison with dual-band SIW antennas in the literature, the proposed configuration depicts a miniature footprint (28.7× 30.8mm) without stacking substrates. To excite the four cavities with equal power, a broadband power divider that supports the propagation of TE10 mode is designed. Accordingly, the impedance bandwidths are 26.6–28.3GHz and 36.8–38.9 GHz. *e measured realized peak gain over the lower and higher bands is 9.3–10.9 dBi and 8.7–12.1 dBi, respectively. *e measured half-power beam widths (HPBWs) at 28GHz and 38GHz are 20.7° and 15.0°, respectively. Considering these characteristics, including dual bands, high gain, narrow beam widths, miniaturization, and single layer, the proposed antenna array is a suitable candidate for millimeter-wave 5G communication systems with the flexibility in switching operating frequency bands against channel quality variations.


Introduction
e millimeter-wave fifth generation (5G) communications have gathered great importance in recent years. As relatively broad bandwidths are available at 28 GHz and 38 GHz, millimeter-wave communications provide higher data capacity. To feed a millimeter-wave antenna, waveguide structures are particularly suitable due to low loss and high power handling capability. However, classical waveguides with solid walls have difficulty to be integrated with printed circuit boards (PCBs). In this situation, substrate integrated waveguide (SIW) feeds can be fabricated with PCBs seamlessly, showing the advantages of low material and fabrication cost, low profile, and excellent shielding properties.
us, SIW antennas depict great potential for 5G millimeter-wave applications.
Although considerable attention has been paid to the development of 5G millimeter-wave SIW antennas, earlier studies put more emphasis on single-band design at 28 GHz [1][2][3][4][5][6][7][8][9][10]. Simultaneous realization of 28-GHz and 38-GHz resonances can reduce the cost and space, but such dualband designs are relatively lacking [11][12][13][14]. e performance of these dual-band SIW antenna arrays is shown in Table 1. Several arrays are implemented through multiple layers of substrates [11][12][13], which increase the complexity and cost of integration. On the other hand, the single-layer design depicts relatively large dimensions, and the impedance bandwidths are too narrow to cover the 5G specification [14]. e goal of this paper is to present a miniature dual-band SIW slotted array fabricated on one single layer of substrate for millimeter-wave 5G applications.
is SIW antenna consists of four cavities arranged as a linear array. e novelty of the proposed antenna array is threefold. First, while SIW slotted arrays have been constructed by conventional thin slots [4, 6-8, 12, 14], square rings [3,5], quarter-mode resonant cavities [13], triangular-split-ring slots [10], and spoon-shaped slots [9], this study proposes a cavity that consists of new resonant topologies, including a rhombic slot and a triangular-split-ring slot.
is new topology miniaturizes the antenna footprint, providing straightforward integration with the microwave electronics of a transceiver. Next, the proposed antenna is a single-layer design, reducing cost and complexity from multilayer PCBs [11][12][13]. ird, although a miniature and low-profile antenna is developed, the antenna still provides sufficient  impedance bandwidths and high broadside gain. e operating mechanism of the antenna will be explained, and optimized performances supplemented with simulated and measured results will be demonstrated.

Unit Element
While the proposed antenna is a 1 × 4 linear array, this section presents the design and analysis of a unit element. Figure 1 exhibits the geometry and detailed dimensions of the proposed unit element. e antenna is fabricated on one layer of RT/Duroid 5880 substrate (dielectric constant ε r � 2.2 and loss tangent tanδ � 0.003). e SIW is fabricated by metallic via arrays, where the separation between adjacent vias is 0.675 mm and the diameter is 0.4 mm. A tapered SIWto-grounded co-planar waveguide (GCPW) transition is designed to enable dual-band features. e GCPW is terminated with a 50 Ω end launcher, which is utilized to excite the cavity. e proposed unit element consists of a SIW cavity with a rhombic slot and a triangular-split-ring slot. To explain the principle of dual-band resonances, Figure 2 exhibits the distributions of E-fields simulated using Ansoft HFSS. With the SIW-to-GCPW transition, the TEM mode in the GCPW can be converted into TE 10 fundamental mode in the SIW. More explicitly, Figure 3 demonstrates the E-field distributions of the cavity at 28 GHz and 38 GHz. e maximum magnitude at 28 GHz is associated with both the resonators. e E-field of the rhombic slot is closely coupled with that of the triangularsplit-ring slot, leading to mode bifurcation at the resonant frequency. e cavity resonates on TE 101 mode. At 38 GHz, the maximum magnitude is majorly radiated through the triangular-split-ring slot, and the cavity resonates on TE 102 mode. e operating principle explained above can be validated through parametric studies. Figure 4 indicates that the impedance characteristics at 28 GHz and 38 GHz are affected by the variation of the width of the triangularsplit-ring slot, namely, w t . e larger the w t , the higher the resonant frequencies are for both the dual bands. In addition, when the separation between the two slots, namely, s, is varied over 0.425 mm, 0.475 mm, and 0.525 mm, the impedance matching at 28 GHz is unaffected. Meanwhile, the resonant frequency at the higher frequency band can be controlled by tuning s; the associated resonant frequencies are 38.8 GHz, 38.6 GHz, and 38.2 GHz, respectively. According to these results, the design rule of the proposed unit element involves two steps. e first step is to tune the geometric parameters to cover the lower frequency band. When the bandwidth requirement of the lower band is met, the second step is to tune the geometric parameters of the triangular-split-ring slot so that the bandwidth of the higher frequency band covers the specification and that of the lower band remains unvaried. ese design guidelines offer us a systematic method to obtain the configuration of the rhombic slot and the triangular-split-ring slot. In addition, our parametric studies suggest that the maximum dimension of the triangular-split-ring slot should be larger than 3.2 mm to cover the lower frequency band. Similarly, the maximum dimension of the rhombic slot should be larger than 4.3 mm to cover the higher frequency band. e prototype of the proposed unit element is fabricated and tested. e impedance characteristics are tested using a vector network analyzer (VNA) N52227A from Keysight. Figure 5 shows the simulated and measured reflection coefficients. In terms of the -10 dB threshold, the simulated impedance bandwidths are 27.7-28.4 GHz and 37.7-40.1 GHz, whereas the measured ones are 26.7-27.7 GHz and 36.7-38.2 GHz.
ere are frequency shifts between the simulated and measured results. e discrepancy could be due to the manufacturing errors including the variation of the substrate thickness and the soldering of the end launcher. As the thickness of the substrate increases, the resonant frequency shifts to a lower level. e measured response agrees very well with the antenna designed with a 0.82 mm-height substrate. On the other hand, the model in full-wave simulation does not include the end launcher, which causes differences between the simulated and measured results. e sensitive nature of millimeter-wave cavity and resonator geometry also incurs additional errors. Figure 6 exhibits the simulated and measured realized peak gain and antenna efficiency. For the lower frequency band, the simulated (measured) realized peak gain is 7.6-8.5 (7.0-8.7) dBi. At the resonant frequency, the simulated and measured antenna efficiencies are 84% and 82%, respectively. For the higher frequency band, the simulated (measured) realized peak gain varies over 7.0-8.0 (3.9-6.8) dBi. At the resonant frequency, the simulated and measured antenna efficiencies are 76% and  Figure 7. e discrepancy between the simulated efficiency and the measured one is noticeable at the 38 GHz band. is difference is attributed to three issues. First, as will be shown in Figure 8, the pattern of the unit element is not highly directional, and the half-power beam widths (HPBWs) are greater than 70°.
us, mutual coupling between the AUT and the transmitting horn antenna may occur, reducing the radiation efficiency of the AUT. Second, serrated structures are not implemented in the available CATR anechoic chamber, so diffraction effects may reduce the measurement resolution and the antenna    International Journal of Antennas and Propagation efficiency. ird, the simulated efficiency does not include the loss due to the end launcher, but the imperfect conductivity of the connector may incur additional loss, especially at 38 GHz. More explicitly, Figure 8 demonstrates the simulated and measured radiation patterns. e main beam is directed at the broadside direction. e simulated (measured) HPBWs at the resonant frequencies of the lower and higher bands are 70.5°(76.0°) and 76.2°(74.7°), respectively. ese impedance and radiation characteristics validate the proposed unit element.

Antenna Array
After investigating the performance of the unit element, four identical unit elements are integrated as a linear array. Figure 9 exhibits the geometry and detailed dimensions of the proposed 1°×°4 slotted SIW antenna array. e four cavities are cascaded in a shunt-fed manner. All design parameters regarding a single cavity are the same as those described in the previous section.
A broadband power divider is designed to excite the four cavities. e design is created by the duplicate of multiple 2-way dividers. Considering a 2-way power divider, tailoring the vertical segments leads to the equality of output power at each port, and adjusting the spacing between horizontal segments results in an antiphase between adjacent ports. e E-field distributions of the power divider are shown in Figure 10. As compared to the E-field of the unit element, the proposed slotted array still can be excited by the TE 10 mode. In addition, this power divider avoids multilevel divider configurations, splitting input energy into 4-way output straightforwardly.
For verification, the proposed SIW slotted antenna array is tested in terms of impedance and radiation characteristics.   International Journal of Antennas and Propagation e simulated (measured) reflection coefficients, provided in Figure 11, indicate that the impedance bandwidths are 26.6-28.4 GHz and 37.8-39.3 GHz (26.6-28.3 GHz and 36.8-38.9 GHz). Minor discrepancies are observed, and they are attributed to the experimental manufacturing techniques. However, both the millimeter-wave 5G frequency bands can be covered by the proposed antenna array. Figure 12 depicts the realized peak gain and antenna efficiency. e simulated peak gain and measured peak gain over the lower frequency band are 8.5-11.8 dBi and 9.3-10.9 dBi, respectively, whereas the simulated and measured results over the higher frequency band are 13.5-14.3 dBi and 8.7-12.1 dBi. e simulated (measured) antenna efficiency at the resonant frequencies of the lower and higher bands is 66% (45%) and 72% (41%), respectively. Figure 13 exhibits radiation patterns at 28 GHz and 38 GHz. e broadside gain at 28 GHz and 38 GHz is 9.2 dBi and 8.7 dBi, respectively. In particular, at 28 GHz, the simulated and measured HPBWs on the H-plane are 20.4°a nd 20.7°, respectively. At 38 GHz, the simulated and measured HPBWs on the H-plane are 15.3°and 15.0°, respectively. Additional studies are performed to investigate the radiation performance of the proposed 1 × 4 layout and that of a 2 × 2 series-fed SIW slotted array. e proposed design demonstrates desirable broadside radiation, whereas the 2 × 2 series-fed layout suffers from the deflation angle of main beams. Since dual-band performance is pursued,  International Journal of Antennas and Propagation creating in-phase excitations for longitudinal adjacent elements on both the frequency bands is challenging. ese results indicate that the proposed antenna achieves dualband, high-gain, and narrow-beam features.

Conclusion
A dual-band SIW slotted antenna array that implements four cavities comprised of rhombic and triangular-split-ring slots has been presented. e proposed topology has three distinct features. First, the 1 × 4 antenna array demonstrates a compact footprint (28.7 × 30.8 mm 2 ) with single-layer construction, reducing manufacturing complexity and providing straightforward integration. Second, a systematic design method is developed for the proposed antenna. e resonance at the lower frequency band is generated by the mode bifurcation of the rhombic and triangular-split-ring slots, whereas the higher resonant frequency can be controlled by tuning the triangular-split-ring slot alone. ird, the proposed design not only shows manufacturing simplicity, but it also provides sufficient impedance bandwidths (26.6-28.3 GHz and 36.8-38.9 GHz) and high realized gain (10.9 dBi and 12.1 dBi for the lower and higher frequency bands, respectively). ese results make the proposed SIW slotted antenna array a suitable candidate for 5G dual-band communication systems.

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

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper. International Journal of Antennas and Propagation 9