1. Introduction
Over recent years, the evolution of mmWave communication systems has led to more rigorous requirements for antenna designs such as high gain, wide and multiband operation, as well as pattern diversity. DRAs have the potential of addressing these requirements due to well-known advantages such as high radiation efficiency, wide impedance bandwidth, and design flexibility. Therefore, mmWave DRAs have received increased research interest with a focus on broadside radiation [
1,
2,
3,
4]. On the other hand, omnidirectional radiation is desired for 5G and Beyond 5G (B5G) communication systems to increase the coverage area in various applications such as on-body communications as well as device-to-device short-distance communications [
5]. Therefore, several studies have been reported on the design of mmWave omnidirectional antennas [
6,
7]. However, an omnidirectional mmWave DRA has not been reported previously, which is in sharp contrast with the numerous published studies on the design of omnidirectional DRAs at lower frequencies with a focus on exciting specific transverse magnetic (TM) and transverse electric (TE) modes to achieve the required pattern.
For example, an omnidirectional cylindrical DRA was proposed by exciting the
and
resonance modes at 3.87 GHz and 4.02 GHz, respectively, using a central coaxial probe feed in [
8]. Moreover, a multiband, multisense, circularly polarized hybrid patch/DRA omnidirectional antenna was reported by exciting the
and
resonance modes for the patch at 2 GHz and DRA at 2.6 GHz, respectively [
9]. In a more recent study, a wideband filtering omnidirectional cylindrical DRA was presented using a hybrid feed that consisted of a coaxial probe and metallic disk to excite the
and
DRA resonance modes at 2.19 GHz and 3.37 GHz, respectively, in [
10]. Further, a coaxial probe-fed omnidirectional hemispherical DRA was proposed by exciting the
resonance mode at 3.7 GHz in [
11]. Another probe-fed omnidirectional hemispherical DRA was designed for a wireless capsule endoscope system by exciting the
and
resonance modes for multipolarization at 2.45 GHz in [
12]. Moreover, a probe-fed omnidirectional rectangular DRA with a square cross section was designed by exciting the quasi-TM
011 mode at 2.4 GHz for linear and circular polarizations in [
13]. Furthermore, innovative rectangular multifunction glass DRAs were reported with the capability of achieving linearly and circularly polarized omnidirectional radiation patterns by exciting the quasi-TM
011 mode at 2.4 GHz in [
14]. Similarly, the higher-order degenerate
and
modes were excited at 3.6 GHz, with equal amplitude and phase, to achieve omnidirectional radiation from a rectangular DRA in [
15,
16]. Subsequently, a multiband probe-fed omnidirectional rectangular DRA was proposed, where the
and
resonance modes were excited at 3.5 GHz together with the
and
resonance modes at 5.8 GHz, in [
16].
It should be noted that in all the above-mentioned studies, omnidirectional radiation was attained using a centrally located coaxial feeding probe to excite the required resonance modes. On the other hand, owing to their capability of supporting various types of modes when placed above a ground plane, cylindrical DRAs have successfully been utilized recently with planar feed networks to achieve omnidirectional radiation patterns. For example, an omnidirectional cylindrical DRA with a planar feed of a shorted microstrip cross was demonstrated by exciting the
and
resonance modes at ~2.4 GHz to achieve circular polarization diversity [
17]. The first attempt to utilize a ring-slot aperture to feed an omnidirectional cylindrical DRA was proposed by exciting the
fundamental resonance mode at 2.4 GHz [
18]. Furthermore, a pattern diversity cylindrical DRA was proposed using a meander line-loaded annular slot to excite the TM
01δ mode in combination with a differential strip to excite the HEM
11δ for omnidirectional and broadside radiations, respectively, at 2.4 GHz [
19]. Moreover, four linear stubs were utilized to excite the
and
resonance modes of an omnidirectional cylindrical DRA to achieve circular polarization at 5.8 GHz [
20]. In a more recent study, arched-aperture feeding was employed in the design of a wideband omnidirectional cylindrical DRA by merging the bandwidth of the excited TM
01δ and TM
02δ resonance modes at ~5.8 GHz [
21].
A rectangular DRA with a square cross section supports quasi-TM modes [
22] that have been traditionally excited using a coaxial feeding probe to achieve omnidirectional radiation [
13,
14,
15,
16]. It is well-known that a coaxial feeding probe requires a hole to be drilled inside the DRA, which is impractical at mmWave frequencies due to the physically small DRA size. Furthermore, the probe’s reactance can be large at millimeter-wave frequencies. Moreover, the power handling capacity of the probe is reduced at higher frequencies, leading to signal degradation and power dissipation [
18,
23]. Therefore, aperture–slot coupling is preferred to excite a DRA at higher operating frequencies as it provides a high level of isolation between the antenna and the planar feed network. On the other hand, compared to the cylindrical counterpart, a rectangular DRA offers an additional degree of design freedom together with simpler fabrication due to the planar sides. Therefore, an alternative noninvasive feeding approach needs to be utilized to design a mmWave omnidirectional rectangular DRA. Such a design is proposed in this study, where a ring-slot aperture is utilized to excite the required modes. In addition to the planar feed, the proposed antenna offers another advantage of multiband operation with two types of radiation patterns: broadside and omnidirectional. The first is achieved by exciting the fundamental TE
111 mode at 17.5 GHz as well as a slot resonance at 23 GHz. An omnidirectional pattern is achieved by exciting the
and
higher-order degenerate modes. It should be noted that all the reported dual-band DRA designs radiate either broadside or omnidirectional patterns in both bands. As a result, the proposed DRA can be employed simultaneously for off-body and on-body applications, for example, by utilizing the broadside and omnidirectional patterns, respectively. A common problem with existing on-body antennas is the reduced radiation efficiency due to the impact of the human body, especially at mmWave frequencies. However, the utilization of on-body omnidirectional DRA can help in achieving more efficient on-body antennas, which necessitates the design of a planar feeding network.
This article is organized as follows:
Section 2 presents the proposed DRA configuration.
Section 3 investigates the excitable DRA modes at a frequency range of 20–30 GHz.
Section 4 is focused on the design of the planar feed network.
Section 5 presents an analysis of the performance of the on-body mmWave DRA.
Section 6 presents the measured results that agree closely with the simulations and
Section 7 is focused on the conclusions. All the simulations are implemented using CST microwave studio.
2. Antenna Configuration
The DRA was designed using a square cross section to facilitate the excitation of the required degenerate modes for omnidirectional radiation. In addition, alumina with a dielectric constant of
εd = 9.9 and a loss tangent of tanδ = 0.0001 was chosen as the DRA’s material.
Figure 1 illustrates the utilized configuration in which the DRA was placed on a square ground plane with a size of
Gs = 12.5 mm. The feed network also involved a square Rogers substrate, Ro4003, that was located at the lower side of the ground plane. The substrate had a thickness of
hs = 0.308 mm, dielectric constant of
εr = 3.5, and loss tangent of 0.0027. Additionally, a 50 Ω microstrip feedline was printed on the substrate’s lower surface with a respective length and width of
lt = 6.25 mm and
wt = 0.3 mm.
The design was evolved by noting that an electrically small vertical probe and a ring-slot that is etched in a metal ground plane represent the duals of a planar loop antenna with equivalent size. Therefore, the small ring-slot provided the same fields as an electrically short vertical probe and, hence, could be considered as an option to create the required planar feeding network that incorporated a rectangular ring-slot as a natural choice to feed a rectangular DRA, as can be observed from
Figure 1. However, the ring-slot size may need to be increased depending on the field distribution of the required DRA mode. Furthermore, the utilized ring-slot consisted of
x- and
y-directed slot arms with side lengths of
ls1 and
ls2. These slots behave as magnetic currents that excite the required magnetic fields inside the DRA. Since the ring-slot was positioned at the interface between the alumina DRA and the Rogers substrate, the circumference needed to be calculated in terms of the effective wavelength λ
eff = λ
0/√
εeff, where λ
0 is the free space wavelength and
εeff is defined as [
24]
In order to design an optimum feed, the supported DRA modes need to be identified over the frequency range of interest, as illustrated in the next section.
3. Supported Modes of the Proposed DRA
Based on the dielectric waveguide model (DWM) [
25], the DRA dimensions were chosen to support the required degenerate modes for omnidirectional radiation at ~28.5 GHz when the DRA is located above a metal ground plane. Therefore, the DRA’s length, width, and height were determined as
a =
b = 3.8 mm and
h = 1.7 mm, respectively. These dimensions offer a compact DRA size that allows easy integration. The resonance frequencies of the
modes can be determined using the DWM as [
25]:
where
k0 is the free space wave number. Owing to the square cross section of the DRA, the resonance frequencies of the
and
modes are equal. Therefore, the required
and
higher-order modes can be simultaneously excited at the same frequency. This results in a total magnetic field distribution that is similar to that of the HEM
21δ mode of a cylindrical DRA that generates an omnidirectional pattern.
Table 1 summarizes the supported resonance modes for the chosen rectangular DRA dimensions over a frequency range of 15–30 GHz based on the DWM.
The excitation of the required modes can be achieved by studying the magnetic field distributions of the supported resonance modes inside an isolated DRA which are illustrated in
Figure 2. For example, from the
mode’s magnetic field distribution, it can be observed that the H-field is null when
y = 0.5
b, where
b is the DRA size. Therefore, the utilization of a centrally located
x-directed slot aperture will suppress this mode. Hence, the slot needs to be shifted along the
y-axis to a strong H-field point to excite this mode effectively. Similarly, for the
mode, in which the H-field is null at
x = 0.5
b, an off-set
y-directed slot is needed at a strong H-field point for effective mode excitation. However, for omnidirectional radiation, the degenerate modes need to be excited simultaneously. Therefore, a ring-slot aperture was utilized, which involved
y and
x-directed slot arms that acted as magnetic current components exciting the aforementioned modes. Furthermore, the chosen DRA dimensions also support the fundamental broadside TE
111 mode at 17.5 GHz and, hence, it would be beneficial if the same ring-slot aperture excites the fundamental TE
111 mode as well as the
and
modes. As mentioned earlier, the interaction between these degenerate modes can result in a field distribution that is similar to the cylindrical HEM
21δ mode [
22]. For a rectangular DRA, such a mode is defined as a quasi-HEM
21δ mode, which offers the required omnidirectional radiation pattern.
Having identified the supported DRA resonance modes and understood the corresponding fields’ distribution, the design of the required ring-slot needed to be implemented as described in the next section together with the achieved DRA performance.
5. DRA Performance Next to a Human Body
For on-body applications, mmWave omnidirectional antennas are widely used. Therefore, it is important that the antenna’s performance is assessed when the proposed antenna is placed next to the human body, as illustrated in
Figure 16. In line with the literature, we assessed the DRA’s performance next to three body areas: arm, chest, and stomach [
27], where a three-layer phantom was utilized. The utilized parameters for the different tissue layers are illustrated in
Table 2. The thicknesses of the three different body parts were based on those reported in [
28]. The return losses when the DRA was placed next to arm, chest, and stomach are presented in
Figure 17, where it can be noted that the presence of the ground plane minimized the impact of the human body on the resonance frequencies. In addition, the omnidirectional pattern was preserved in the presence of the chest, as demonstrated in
Figure 18, which can also be attributed to the presence of the ground plane. However, reflections from the utilized phantom reduced the back lobes considerably and hence increased the omnidirectional realized gain from 4.33 dBi in free space to 5.8 dBi in the proximity of the human body tissue. On the other hand, the presence of the phantom reduced the radiation efficiency from 95% to 84%. However, this did not impact the gain as the increased directivity compensated for any loss due to the slightly reduced radiation efficiency.
The Specific Absorption Rate (SAR) indicates the safety threshold at which radio-frequency energy can be absorbed by human body tissue [
29] The SAR must be assessed to ensure compliance with safety limits set by the Federal Communications Commission (FCC) and the International Commission for Non-Ionizing Radiation Protection (ICNIRP) standards. These standards define SAR thresholds of 1.6 and 2 W/kg for 1 g and 10 g tissues, respectively [
29]. Unfortunately, the above guidelines do not offer dosimetric information or suggestions for mmWave frequencies [
29,
30] However, at 28 GHz, a 5 mm space is recommended between the antenna and the human body with input power levels of 15 dBm, 18 dBm, or 20 dBm at 28 GHz [
30]. Subsequently, the proposed omnidirectional DRA was simulated next to a layered phantom, as demonstrated in
Figure 16. The conducted SAR simulations confirmed that the radiation from the proposed antenna meets the safety requirements, as illustrated in
Figure 19 and
Figure 20 for 1 g and 10 g tissues, respectively. It is worth noting that this SRA example is for the scenario of an antenna placed on the chest phantom.
6. Measured Results
The alumina DRA and planar feed network incorporating a rectangular ring-slot were fabricated by T-ceramics [
31] and Wrekin [
32], respectively. At the mmWave frequency range, a precise alignment between the DRA and the feeding slot poses significant challenges. To overcome these challenges, a solution involving mapping out the DRA position on the ground plane was implemented during the fabrication stage [
33]. The resulting fabricated feed network, which includes the outlined DRA position, is presented in
Figure 21a. Following the outlining of the DRA position, ultrathin double-sided adhesive copper tape with a thickness of 0.08 mm was utilized to bond the antenna to the ground plane, ensuring secure assembly. The assembled DRA prototype is presented in
Figure 21b, including the utilized ELF50-002 SMA connector that was attached using screws. In addition, the prototype was measured without experiencing any alignment or bonding issues. The implementation of this approach is critical in ensuring the mmWave measurements’ accuracy, where even slight deviations can significantly affect the performance [
34]. All measurements were carried out using the UKRI National Millimeter-Wave Facility [
35], where an N5245B vector network analyzer (VNA) was employed to measure the return losses following a standard calibration procedure. Based on the analyzed data, the return losses were determined. On the other hand, an NSI-MI Technologies system was utilized in conducting the far-field measurements. By employing this specialized measurement system, various parameters, including the radiation pattern as a function of ϕ and θ, were accurately measured and visualized. To cover the elevation angle range of θ = −90° to θ = 90°, the arm of the NSI-MI spherical system was set up to rotate across the upper hemisphere. Additionally, the gain of the antenna under test was determined with respect to a reference horn antenna.
As demonstrated in
Figure 22, the measured and simulated return losses shared almost the same resonance frequencies of 17.5 GHz, 23 GHz, and 28.5 GHz for the TE
111, ring-slot, and quasi-HEM
21δ modes, respectively. In addition, the measured and simulated −10 dB impedance matching bandwidth of the lower band was 3.4%. In terms of the middle band that corresponds to the ring-slot resonance, the −10 dB impedance matching bandwidth was 1.8 GHz, demonstrating a good agreement between the measured and simulated percentage impedance bandwidths of 7.7% and 7.5%, respectively. However, a slight discrepancy can be noted in the omnidirectional mode’s simulated and measured bandwidths of 1.9% and 3%, respectively. This discrepancy can be attributed to measurement uncertainties, including measurement setup as well as fabrication and calibration errors. In addition, the utilization of bulky SMA and fittings could have contributed to the discrepancy between simulated and measured results. It should be noted that the achieved impedance bandwidth of the omnidirectional mode was narrower than that of a probe-fed omnidirectional rectangular DRA. For example, impedance bandwidths of 22% were reported in [
15] by merging the bandwidths of the DRA mode and that due to the feeding probe’s resonance, which also offers an omnidirectional pattern. However, such a hybrid operation is not possible in the proposed configuration since the feeding ring-slot has broadside radiation, i.e., different from that of the excited DRA mode. Therefore, a feeding ring-slot with an omnidirectional pattern needs to be utilized for bandwidth enhancement. Alternatively, a dielectric coat layer [
10], or concentric rectangular ring-slots, can be utilized to achieve a wider bandwidth.
Figure 23 presents the measured and simulated radiation patterns at 17.5 GHz, where the TE
111 broadside mode was excited. Close agreement can be observed between the simulated and measured broadside patterns. As mentioned earlier, the feeding slot’s resonance was achieved at 23 GHz and the corresponding far field patterns are demonstrated in
Figure 24, with reasonable agreement between the measurements and simulations. For example, the respective measured beamwidths were 88° and 108° in the E- and H- planes compared to 90° and 107° in the simulations. In addition, the simulated and measured omnidirectional radiation patterns are presented in
Figure 25 for both the elevation and azimuth planes at 28.5 GHz. The results demonstrate close agreement between the simulated and measured radiation patterns, where an omnidirectional radiation pattern was achieved with a main beam direction at θ = 40°, as demonstrated in
Figure 23a. The measured and simulated beamwidths of the omnidirectional patterns were 61.2° and 60.6°, respectively. However, a slight asymmetry can still be noted in the ϕ = 90° plane cut of
Figure 25a, owing to the asymmetrical feed point position. An improved roundness of the azimuthal plane pattern can be observed in
Figure 25b, which suggests that the rectangular ring-slot arms were placed at equally strong magnetic field points. Furthermore, the copolarized field component was considerably stronger than the cross-polarized component in all cases. The azimuthal plane gain variation presented in
Figure 26, where it is evident that the variation was reduced considerably to ~0.85 dB, resulted in a more stable omnidirectional pattern with close agreement between the measurements and simulations. Additionally, the gain and radiation efficiency of the rectangular ring-slot-fed DRA are illustrated in
Figure 27, where it can be noted that the maximum achieved gains were 6.56 dBi, 5.2 dBi, and 4.33 dBi for the TE
111 mode, ring-slot resonance, and the quasi-HEM
21δ mode, respectively. Furthermore, a high radiation efficiency of ~90% was attained in the three operating bands.
A comparative analysis of the ring-slot-fed DRA performance with respect to the reported on-body mmWave antenna designs is presented in
Table 3. As mentioned earlier, there is no reported study on the on-body mmWave DRA in the open literature; hence, a comparison was made with respect to different antenna types that are available in the literature. The comparison was conducted with respect to the size, bandwidth, gain, and radiation efficiency. It is evident from
Table 3 that the electrical size of the proposed antenna was smaller than most of the reported designs, except that of [
36]. In addition, the utilized simple geometry resulted in simple and low-cost fabrication. On the other hand, a triple-band operation was achieved, which was also the case in [
36]. owever, the individual bandwidths in the presented design were wider with higher gains compared to those in [
36]. At the same time, the other antennas in
Table 3 offer single-band operation, albeit with wider bandwidths. Furthermore, the proposed DRA outperformed the reported counterparts in terms of radiation efficiency.