A Dual Wide-Band Mushroom-Shaped Dielectric Antenna for 5G Sub-6-GHz and mm-Wave Bands

The design and implementation of a dual wide-band mushroom-shaped antenna are presented in this paper. The antenna consists of a cylindrical dielectric resonator (cDRA), a cylindrical dielectric rod as a waveguide (cDR), and a dielectric lens (DL). The cDRA in conjunction with the DL acts as a sub-6-GHz antenna. At the mm-wave band, the small cDR acts as a waveguide, which transfers the wave from the feed toward the larger cDRA and the DL in order to produce a high gain. A dual wide-band antenna is designed and fabricated with different dielectric constants for the cDRA/cDR and the DL. 3D printing is utilized to precisely control the dielectric constant of printed component. Measurement results show a maximum gain of 6.4 dB at 5.3 GHz with a 21% 3-dB gain bandwidth and 12.7 dB at 31.5 GHz with a 26.2% 3-dB gain bandwidth. The sub-6-GHz band exhibits a measured 10-dB return loss bandwidth of 21% (centered at 5.15 GHz), and the mm-wave frequency band demonstrates a measured 10-dB return loss bandwidth of 26.2% (centered at 30.5 GHz).


I. INTRODUCTION
G ROWING wireless data traffic requires high-rate data transmission, which necessitates a substantial increase in communication frequencies. This demand has led to the inception of fifth-generation (5G) wireless standards operating at sub-6-GHz and mm-wave band, which provides a 10 to 100 times increase in the data transmission rate. As a result, attempts have been made to develop wideband/multi-band antennas to 5G and beyond for telecommunication systems. Multi-band antennas have various applications in point-topoint and satellite communication systems. Moving toward the mm-wave band necessitates the design of high-gain antennas to compensate for signal losses.
DRAs are known for their high efficiency, wide bandwidth, low fabrication cost, and simplicity of excitation. Previous works employed a single DRA element covering several bands close to one another by exploiting special feeding mechanisms and by employing the fundamental and highorder DRA modes [1], [2], [3], [4], [5]. Recently, a high-gain dual-band hemispherical DRA with resonant frequencies of 7.65 and 10.04 GHz has been presented [6]. The antenna is fed by a waveguide. The impedance bandwidth of the antenna is 4.98% and 12.68% at the lower and higher bands, respectively. However, this approach does not apply to 5G in which the frequency bands are far apart.
Dual-band operation with a high-frequency ratio is essential in the new generation of wireless communication systems and various designs have been proposed for that [7], [8], [9], [10]. The approaches mainly rely on employing two distinct elements; one for the lower frequency band and the other for the upper-frequency band [11], [12], [13], [14], [15], [16], [17], [18]. For wide-band operation, it is necessary to increase the overall size of the antennas. In order to decrease the overall size while maintaining the wide-band operation, a single radiator with two backto-back folded plates was proposed to design a compact dual-band antenna [19], [20]. This antenna provides bandwidths of 9.7% and 2.1% at center frequencies of 2.4 and 24 GHz (ISM bands), respectively. A dual-band antenna that integrates a microwave hollow DRA with a high-gain mmwave dielectric Fabry-Perot resonator antenna (FPRA) was proposed in [21]. This design provides a bandwidth of 4.67% at the mm-wave band which is again not sufficient for 5G mm-wave communications. The feed at the mm-wave band is a rectangular metallic waveguide, which is not easy to integrate antenna on RF printed circuit boards. In another study, a microwave DRA and an mm-wave metallic FPRA were integrated to form a dual-band antenna [22]. At the lower band, the antenna exhibits good performance. However, the antenna has a low radiation efficiency at the upper band due to the conductor loss. A large ground plane and conventional feeding for the lower frequency band is used. Most recently, the idea of encapsulated dielectric resonator antennas (E-DRAs) is introduced in [23], which offer efficient performance for dual wideband applications and deliver bandwidths of 33% and 27% centered around frequencies of 3.6 GHz and 30.5 GHz, respectively.
This work presents a novel wideband, high gain, single dielectric antenna element covering two widely separated bands. The design methodology enables the design of a dualband antenna at the desired frequency ranges. The dielectric antenna used in this work is mushroom-shaped. It consists of a large cDRA on top of a smaller dielectric cylinder acts as a dielectric rod (cDR). In addition, a dielectric lens (DL) is placed on top of the larger DRA. The large dielectric cylinder and the DL serve as a dielectric resonator antenna at sub-6-GHz frequencies. At the mm-wave band, the small cDR acts as a dielectric rod waveguide while the large DR and the DL form a radiating aperture to enhance the gain of the mm-wave signals. Different feeding schemes are utilized at the two bands. A microstrip-fed aperture is used to feed the antenna in the millimeter wave band. At sub-6-GHz, the DRA is fed by a substrate-integrated cavity (SIC) through an annular slot. The same technique has been used in [24] to improve the bandwidth at sub-6-GHz and stabilize the gain patten. However, in contrast to [24], the SIC feed is single-ended in this work (compared to a differential one used before). The antenna presented here is a dual band antenna operating at sub-6-GHz and mm-wave bands but the one in [24] is operating at only sub-6-GHz band. Thus, the contribution in this work is different than that in [24] from an antenna perspective (microwave and mm-wave bands in this work) and the use of the DL.
The remaining sections of this paper are structured as follows. The design of a SIC-backed DRA at sub-6-GHz is described in Section II. In Section III, the designed DRA at sub-6-GHz is modified to act like a lens at mmwave. A dual-band wide-band mushroom-shaped dielectric antenna is described in Section IV. Characterizations of 3D printing materials and the fabrication process are described in Section V. In Section VI, a detailed analysis of the measurement and simulation results is presented. Finally, a conclusion is drawn in Section VII. Figure 1 shows the proposed slot-coupled cDRA backed by an SIC. Compared to simple slot coupling, using a cavity to feed the DRA improves the operating bandwidth and reduces the back lobe of the antenna [24]. For the cDRA designed in this paper, adding the SIC increased the 10-dB return loss bandwidth from 13% to 21%.

II. DESIGN OF A SIC-BACKED DRA AT SUB-6-GHZ BAND
The first step in the design process is to choose the appropriate substrate material. Here, Rogers RO3035 with thickness of 1.524 mm is used, which is a low-loss substrate with dielectric constant and loss tangent of 3.5 and 0.0015, respectively. As shown in Fig. 1(a), the cavity consists of two layers of RO3035 substrates, with thicknesses of h 1 and h 2 .
Two different techniques are used for DRA fabrication, CNC machining, and 3D printing. For CNC machining, the C-Stock AK material from a Cuming Microwave with a relative permittivity of 10 and loss tangent of 0.002 is used. For 3D Printing the material is PREPERM 3D ABS DK 10.0 with nominal relative permittivity of 10.0 ± 0.35 and loss tangent of 0.003 at 2.4 GHz.
The resonant frequency of the fundamental mode (HEM11δ) for a cDRA can be calculated by [35]: where r is the DRA radius, h is the DRA height, and c is the speed of light in free space. The initial value of the radius of the cDRA (r 1 ) is set equal to one-quarter of the wavelength inside the substrate material at the lowest frequency of the desired band and by using (1), the initial value for the height of cDRA is calculated. For the selected relative permittivity (= 10), the initial values for radius and height of the DRA at 4.6 GHz are selected 6.6 mm and 8.7 mm, respectively. Starting with these initial values, the dimensions are varied to maximize the bandwidth. A substrate-integrated square cavity specified by the parameters X 1 and X 2 and with a depth of h 1 +h 2 is shown in Fig. 1. A short-circuited stripline of length L 1 and width W 1 is printed on the bottom metal layer of the upper substrate. The stripline is shorted with a metallic slot via the top metal layer of the upper substrate of the cavity as shown in Fig. 1(b). The annular slot on the top metal layer of the upper substrate has an inner radius of r 1 and width of W 3 . A tapered microstrip line with the length L 2 and width of W 2 is printed on the bottom layer of the top substrate. It is passed through an opening in a sidewall of the cavity and connected to the stripline to feed the antenna at the lower band, as shown in Fig. 1(c).
For the SIC, slot vias were used because of fabrication limitations. Dimensions of the metallic slot vias (V 1 and V 2 ) and edge-to-edge spacing between them (S 1 ) are selected based on the fabrication limitations and to ensure that the waves remain confined within the cavity over the operation bandwidth [24]. Initially, the inner dimension of the square cavity (X 1 and X 2 ) is set to be equal to the one wavelength inside the substrate material at the center frequency of the sub-6-GHz band (5.15 GHz) [36].
In the next step, the DRA dimensions (h c1 and r 1 ) and short-circuited striplines and the annular slot geometrical   parameters (W 1 , L 1 , r 1 , W 3 ) are optimized in ANSYS HFSS to get the best impedance matching over the desired frequency band.
The final values of the parameters are listed in Table 1. In this design, a cylindrical dielectric resonator with single permittivity is used as shown in Fig. 2(a). For the same feeding structure and without changing the SIC dimensions, other shapes/materials for the dielectric resonator shown in Fig. 2 provide similar performance. The designs shown in Figs. 2(b) and 2(c) behave like a DL at the mm-wave band. In the design in Fig. 2(b) we used ε r1 = ε r2 = 10 and in the design in Fig. 2(c) we used ε r1 = 10 and ε r2 = 8. The reflection coefficients of the antenna structure with these three DRAs are also shown in Fig. 3, and similar performance can be observed for the sub-6-GHz band. The simulated reflection coefficient of the antenna shows a 10-dB return loss bandwidth of 21% centered at 5.15 GHz. More details on these designs and their performance will be discussed in the next section.

III. MODIFICATION OF THE LOW BAND DRA TO ACT AS A MM-WAVE DL
The DRA used in this work is mushroom-shaped, which consists of a large cDRA on top of a smaller cDR (Fig. 4). A dielectric lens (DL) is placed on top of a large cylinder. The smaller cDR is attached to the bottom of the larger dielectric cylinder, as a part of the feed at the mm-wave band. Both the small cDR and the large cylinder have a relative permittivity of ε r1 and the DL has a relative permittivity of ε r2 . This mushroom-shaped dielectric is a modified version of the sub-6-GHz structure with the same performance presented in the last section ( Fig. 2(a)), that is now able to operate at mm-wave frequencies to improve the gain.
To change the sub-6-GHz DRA in Fig. 2(a) to a DL ( Fig. 2(b) or 2(c)) at mm-wave band without any change in the performance at sub-6-GHz band, the profile of the top surface of the lens should be modified. This is done by using a simple ray propagation model. To act like a lens, the fields propagating from the focal point to a plane on top of the lens (Fig. 4) must have the same phase. T(d) defines the profile of the top surface of the lens at the distance of d from the center and L(d) is the distance from the focal point to the interface between the larger cylinder and the bottom surface of the lens. The phase of waves propagating from the focal point to a plane at top of the lens can be described by: where k 0 is free-space propagation constant.
and a constant phase at a plane at top of the lens, i.e., ϕ(0) = ϕ(d), the profile of the lens can be calculated using as In (3), To reduce the thickness of the lens, it is desired to set T(d) equal to zero when d = r 1 so that: This provides the maximum thickness (T(0)) of the top part of the DL: Hence, the profile of the top surface of the dielectric lens is calculated based on its permittivity, r 1 , and the focal length (L(0)). Given the approximation on A(d) above, T(d) does not accurately represent the DL profile but it can provide a good estimation of that to find the initial values for the dimensions of the DL for further optimizations. Initially, the focal point is placed at the top of the small cDR (aperture of the waveguide). hc3 is the height of the cylindrical part of the DRA and L(0) is the focal length for the dielectric lens which is not qual to hc3 in general. Since the focal point of the dielectric lens antenna is not known from the beginning, the initial value for focal length is approximately h c3 . The initial value for focal length should be optimized based on the shape of patterns at different frequencies within the mm-wave band.

IV. DESIGN OF THE DUAL-BAND DIELECTRIC ANTENNA
In this section, the cavity is modified to accept the mushroom-shaped dielectric antenna. The disassembled view of the mushroom-shaped antenna and the modified cavity are shown in Fig. 5(a). The modified cavity has a hole at the center to feed the antenna at the mm-wave band through a slot, as shown in Fig. 5(c). The cavity feeds the antenna at the sub-6-GHz band as discussed in Section II. The mmwave feeding slot is at the center of the ground layer and its dimensions are specified by L s , and W s , as shown in Fig. 5(c). A 50-microstrip line with a width of W 4 and length of L 3 is used to feed the slot. The small dielectric cylinder (or rod) acts as a waveguide. The height of the cavity is fixed based on substrate selection for the sub-6-GHz band (h 1 +h 2 ). As can be observed in Fig. 5a, the hole where the dielectric rod is placed is only plated in the top substrate. Based on the simulations, such plating results in a better side lobe level at mm-wave bands. So, the upper part of the rod forms a circular metallic waveguide filled with the dielectric. The minimum diameter for the dielectric rod is determined by the cut-off frequency of this guide. With the selected diameter, the cut-off frequency is about 15 GHz which is well below the desired mm-wave band of operation of the antenna. The bottom part of the rod embedded in the lower substrate forms a dielectric guide with no (i.e., zero) cut-off frequency.
In the next step, the mm-wave feed dimensions (W s , L s , L 3 , r 2 , and r 3 ) are optimized to obtain the best matching over the mm-wave band. Figure 6 displays the antenna's reflection coefficient as a function of frequency within the mm-wave band, with various values of W s , L s , r 2 and r 3 while keeping other design parameters fixed at the values listed in Table 2. In the next step, the cavity with the small dielectric rod at the center is fine-tuned around the geometrical values provided in Section II. Figure 7 shows the reflection coefficient  of the antenna versus frequency for different values of L 1 , r 1 , h c3 , and X 2 at the sub-6-GHz band. These results are for a mushroom-shaped DRA with ε r1 = 10 and = ε r2 = 8. Other parameters are set according to the values shown in Table 2. In that Table 2, parameter values in parenthesis are for another design of the sub-6-GHz antenna, covering the 3.5-4 GHz band. This illustrates the capability of this design to cover different bands of the 5G sub-6 spectrum.
In the next step, optimization is done for the top profile of the lens at the mm-wave frequency to get the best radiation characteristics and matching. As shown in Fig. 8b, the lowest side lobe level (SLL) is obtained when L(0) = 11.4 mm. By changing the L(0), only the profile of the top surface of the dielectric lens will be changed and h c3 is fix. Fig. 8(c) and (d) also show that varying this parameter has no significant effect on sub-6-GHz performance.
To reduce the reflection from the lens-air interface, it is better to select a smaller dielectric constant for the top curved

FIGURE 8. (a) Reflection coefficient of the antenna at mm-wave band versus frequency for different values L(0), and (b) H-plane realized gain pattern at 30 GHz, (c) Reflection coefficient of the antenna at sub-6-GHz versus frequency for different values L(0), and (d) normalized radiation pattern at 5.15 GHz.
part of the lens, so the value 8 or smaller is better than 10 for the lens performance. In addition, in this design, since the dielectric constant of the top part of the DL and the larger cylindrical dielectric (bottom part of DL) are very close (i.e., 8 and 10, respectively), the reflection from their interface is negligible. Fig. 9 presents the effect of varying ε r2 based on equations (4) and (6) at the mm-wave band. It can be observed that the gain/pattern/SLL is better when the top part of DL is made of a material with a relative permittivity of 8 as compared to the case where DL is completely made of a material with a relative permittivity of 10. For the four values of permittivity shown in Figure 9, the lens profile was re-optimized based on equations (4) and (6). The optimum  geometrical parameter values based on the simulation results are listed in Table 2.
The antenna design presented in this study offers versatility for various applications that demand an impedance bandwidth exceeding 20% in both sub-6-GHz and mm-wave frequency bands, exemplified by the proof-of-concept 5G antenna capable of supporting sub-6-GHz bands such as n46/n79 and mm-wave bands such as n257/n261 [23]. For the design of the antenna at a lower frequency band at sub-6-GHz, the size of the cavity and the larger cDRA must be increased.
The electric and magnetic field distributions inside the sub-6-GHz DRA at center frequency are shown in Fig. 10. The principal mode in the DRA is HEM 11δ based on the presented field distributions, which provides broadside radiation. This field distribution is not perfectly symmetric due to unbalanced feeding, but it still provides a broadside radiation pattern. As shown in [24], by feeding the DRA using an SIC, the DRA over the desired sub-6-GHz band has a single-mode operation. The E-field magnitude inside the mushroom-shaped antenna at the center frequency of the mm-wave band is shown in Fig. 11(a). It can be seen that the beam is collimated toward the z-axis which indicates improved radiation gain at broadside direction. Fig. 11(b) shows the phase of the electric field inside the mushroomshaped antenna at the same frequency. It can be observed that a flat wavefront is achieved in the "constant phase plane" defined in Fig. 4.

V. FABRICATION OF THE MUSHROOM-SHAPED DRA
The mushroom-shaped part of the proposed design has two different relative permittivities of 10 and 8. The small and the large cylindrical dielectrics are fabricated by machining a slab of the C-Stock AK from Cuming Microwave. The top-shaped dielectric part is 3D printed utilizing FFF technology which is an extrusion-based additive manufacturing technology. In FFT, a polymer-based filament is fed into an extruding nozzle and then the filament is melted and extruded based on the input geometry and deposited layer by layer.
Based on the required relative permittivity (8), commercial PREPERM ABS1000 R material is used for fabrication, which is a special compound based on Premix proprietary ABS technology. The nominal relative permittivity of the material is 10 and it is stable over a wide range of frequencies and temperatures, with a low loss tangent of 0.003.
3D printing enables tuning the permittivity of the fabricated part by controlling the infill percentage. The actual relative permittivity of printed parts is found to be less than the nominal relative permittivity of the filaments even for the 100 percent infill. This is because of the presence of voids [23], [32], [37]. Previous studies show that Bruggeman's model [38] can approximate the relative permittivity of 3D printed samples with an acceptable accuracy [23], [37]. Here, the same model is adopted to estimate the infill percentage needed to obtain the desired permittivity.
The print settings and parameters are selected according to the filament manufacturer instructions and to achieve the highest quality samples with minimum defects. A summary of the AM parameters and settings for printing are listed in Table 3. The extrusion temperature is 20 • C higher than the temperature recommended by the filament manufacturer in order to achieve a higher surface quality. Also, a concentric infill pattern is selected mainly to avoid any anisotropic dielectric properties in the samples.

VI. SIMULATION AND MEASUREMENT RESULTS FOR THE FABRICATED ANTENNA
The designed dual-band mushroom-shaped DRA was simulated at both sub-6-GHz and mm-wave frequency bands in ANSYS HFSS and the results are presented in this section. The dimensions of the fabricated antenna are presented in Table 2. Fig. 12 presents the image of the fabricated antenna  and the measurement setups. Fig. 13 shows the measured and simulated reflection coefficients of the antenna. The simulation and measurement results indicate a significant level of agreement. For the sub-6-GHz band, the measured −10 dB reflection coefficient bandwidth is 21%, centered at 5.15 GHz. Similarly, for the mm-wave band, the measured −10 dB reflection coefficient bandwidth is 26.5%, centered at 30.5 GHz. The isolation between the two antenna ports (each is used for one band) is better than 35 dB and 37 dB in sub-6-GHz and mm-wave bands, respectively, as can be seen in Fig. 13(c) and (d).
According to Fig. 5, the polarization at both bands is linear with the direction along x at sub-6-GHz and along y at the mm-wave band. In Fig. 14, the radiation patterns of the antenna at 4.6 GHz, 5.15 GHz, and 5.7 GHz for sub-6-GHz are presented. The results of simulation and measurement for co/cross-polarization E-and H-plane radiation patterns (normalized to the maximum of the co-polarization) at 28, 30.5 and 33 GHz are presented in Fig. 15. The results obtained from measurements and simulations exhibit a high level of agreement. The SLL achieves values lower than −12 dB and −15 dB in the E-plane and H-plane, respectively, while the cross-polarization level remains below −20 dB in the broadside direction. The main result for the difference between simulation and measurement is degradation in cross-polarization levels in the experimental results, which is possibly due to misalignment.
The measured and simulated realized gain versus frequency at sub-6-GHz and mm-wave bands of the antenna are shown in Fig. 16. According to Fig. 16(a), at sub-6-GHz band, the maximum gain is 6.4 dBi at 5.3 GHz. The measured 1-dB and 3-dB gain relative bandwidths are 15% and 21%. According to Fig. 16(b), at the mm-wave band, the maximum gain is 12.7 dBi (at 31.5 GHz) and the measured 3-dB gain relative bandwidths are 26.2%. Fig. 17 shows the radiation efficiency at the antenna in both bands. In this figure, the radiation efficiency is measured by Satimo Starlab near-field chamber for sub-6-GHz. The radiation efficiency is calculated by dividing the measured realized gain by the simulated directivity [39] for the mmwave band. The minimum total efficiency is 95% over the sub-6-GHz band and 77% over the mm-wave band. A comparison of the measured performance of the proposed antenna with the previous works on dual-band DRAs is given in Table 4.
The idea behind the proposed antenna was to use a new feeding technique for the dual-band dielectric antenna at both sub-6GHz and mm-wave frequency bands. The DRA was modified as an mm-wave lens without changing its size, resulting in high gain and wideband antenna performance at mm-wave frequency. According to Table 4, the proposed antenna has a smaller size compared to [5], [6], [21], and [22]. Tough [17] has a smaller size, and its impedance bandwidth is much less than that of the proposed antenna. Compared to other works, the proposed antenna achieved the most significant wide bandwidth at mm-wave frequency and high bandwidth for the lower frequency band. The antenna is made of low-loss dielectric materials for both bands and thus exhibits a high radiation efficiency in both bands. The main part of the antenna (the inner part of the cavity) has a lateral size of 32.5 mm x 32.5 mm (Table 4).

VII. CONCLUSION
A novel dual wide-band mushroom-shaped antenna has been proposed for operation at sub-6-GHz and mm-wave frequency bands with a high-frequency ratio. For improving the bandwidth at sub-6-GHz, a cavity is used to feed the DRA. The cavity supports two modes to provide almost the same excitation for the DRA over a wide bandwidth resulting in a stable radiation characteristic for the antenna. The sub-6-GHz DRA also acts as a dielectric lens at the mm-wave and so improves the gain at the mm-wave band. The measurement results show the maximum gain of 6.4 dBi at sub-6-GHz and 12.7 dBi at mm-wave. The proposed antenna shows more than 20% 10-dB impedance bandwidth at both sub-6-GHz and mm-wave bands. In the fabrication of the antenna, a 3D printing technique has been used which enabled control of the dielectric constant and shape of the fabricated lens.  He is a Co-Director of the Laboratory for Multiscale Mechanics. Several important scientific contributions (∼ 120 refereed papers) in prestigious journals, such as Nature Materials, Advanced Materials, and Additive Manufacturing, and multiple innovations (ten Patents), originated from the research activities of his research team. He worked or is working on many collaborative research projects with partners such as Safran Group, Bombardier Aerospace, ArianeGroup, Bell Textron Canada, 3M, Dyze Design, NanoXplore, and the Canadian Space Agency. His research interests are mainly related to additive manufacturing of advanced materials and multifunctional composites. Past contributions include the development of innovative additive manufacturing processes (e.g., freeform printing), the design and fabrication of nanocomposite materials for advanced aerospace applications and the 3-D printing of carbon fiber-reinforced thermoplastic composites.