The bare version, i.e., facing the ambient environment, dual-band nested patch array was designed using electromagnetic full-wave simulations, using CST Studio Suite 2023 [
20]. Upon fabrication, measurements were carried out in a shielded anechoic chamber with ±1 dB measurement uncertainty across the frequency ranges of interest; a snapshot of the measurement setup is depicted in
Figure 6a. The array was mounted to a plastic holder and bolted to the rotating metallic positioner, covered with absorbing material to avoid unwanted artefacts in the measured radiation pattern. The array was then embedded into the rear spoiler and re-measured as shown in
Figure 6b. The nested array was designed and integrated such that it remains hidden in the rear spoiler.
4.1. Results for the 3.5 GHz Array
The impedance matching of the bare simulated (red curves), bare measured (blue), and embedded measured versions (green) across the entire n78-band are compared in
Figure 7a. The |S
11|
2 data for the embedded state follow the simulated trend but with smoothened ripple due to additional conductor losses. The operational bandwidth of Δf
op ≈ 100 MHz around f = 3.5 GHz is covered with a favorable input matching of |S
11|
2 < −10 dB for the bare simulated version, and while the matching level degraded to the still acceptable level |S
11|
2 < −6 dB for the bare measured version in the same bandwidth. The measurement in the embedded state covered |S
11|
2 < −6 dB over a bandwidth of 160 MHz, most likely reflecting the additional attenuation caused by the rear spoiler material.
The frequency variation of the axial ratio is shown in
Figure 7b, which confirms a favorable polarization symmetry with AR < 3 dB around f = 3.5 GHz for all three data sets. The embedded array exhibits a 3-dB axial ratio bandwidth B
AR ≈ 100 MHz. The angular variation of the axial ratio is shown in
Figure 7c, indicating an elevation range Θ from −40° to +30° where AR < 3 dB for the bare simulated version. These elevation ranges shifted for the measured bare and embedded versions to Θ = −45° … +50° and Θ = −50° … +25°,respectively. All angular ranges cover the 3- dB beamwidth Θ
UE = ±30° of the proposed embedded array and appear feasible for the intended applications.
The simulated realized gain and the radiation efficiency of the array were found to be G
Uesim = 9 dBi and η
sim = 74%, respectively. The radiation efficiency decreased slightly to η
meas = 59% for the measurement of the bare antenna, associated with the reduced gain of G
Uemeas = 8.8 dBi. The radiation patterns are displayed in
Figure 7d,e for the three versions. The normalized vertical cuts at Φ = 0° and 90° of the simulated and measured antenna in its bare state agree well with each other, while the pattern measured for the embedded array shifted towards 0° and a side lobe developed in the direction of excitation for Φ = 0°. Upon embedding, apparently, the plastic material covering the patch elements deteriorates the radiation performance of the antenna array, e.g., through degrading the symmetry and affecting the phase balance between the elements. In addition, the realized gain of the embedded array increased by 1 dB to G
emb = 9.9 dBi in comparison to the measured bare antenna array as the embedded measured directivity is 1.5 dB higher than simulated array. The embedded array offers a 3-dB beamwidth of Θ
UE = −30° … +15° and an impedance bandwidth B
emb ≈ 160 MHz due to an increase in the overall thickness of the assembly.
All simulated and measured results are summarized in
Table 4. The data rate was calculated according to
Section 2. An uplink data rate of DR = 21 Mbit/s is anticipated for the embedded measured n78-band 2 × 2 square patch array, feasible for numerous wireless connected-vehicle applications.
4.2. Results for the 28 GHz Array
The impedance matching of the simulated and measured bare and embedded version across the entire n257-band is depicted in
Figure 8a. The bare and embedded minimum |S
11|
2 values shifted slightly by 0.3 GHz to a lower frequency of 29.3 GHz compared to the simulations, while staying well below −10 dB. The manufactured array dimensions were measured with an optical profilometer with a measurement uncertainty below 1 μm, revealing deviations of 2% from the design parameters, which are considered negligible. The frequency shift, which is often observed in similar studies, is hence associated with variations of the dielectric permittivity. The back side of the embedded patch array remained exposed to the free space as depicted in
Figure 5d, and displayed almost the same resonant frequency as the bare version.
The frequency variation of the axial ratio is shown in
Figure 8b, achieving AR < 3 dB around f = 28 GHz for all three data sets. The embedded array exhibits a wide 3-dB axial ratio bandwidth B
AR ≈ 1500 MHz. The angular variation of the axial ratio is shown in
Figure 8c, indicating an elevation range from Θ = −18° … +15° with AR < 3 dB for the bare measured version, which slightly shifted to Θ = −11° … +17° for embedded version.
The simulated realized gain and radiation efficiency of the array amounted to G
UEsim = 15.4 dBi and η
sim = 85%, respectively. The radiation efficiency decreased to η
meas = 60% for the measured bare antenna, associated with a lower gain of G
UEmeas = 14 dBi. The radiation patterns are shown in
Figure 8d,e for the three assembly versions. The normalized vertical cuts at Φ = 0° and 90° of the simulated and measured antenna in its bare state agree well with each other, while the pattern measured for the embedded array displays more pronounced ripples, in agreement with the findings for the lower frequency. In addition to the impact of dielectric embedding, adverse effects may result from the transition from the coaxial feed cable to the microstrip feed line. As a consequence, the realized gain of the embedded array decreased by 0.3 dB to G
emb = 13.7 dBi in comparison to the measured bare antenna array. Thus, the radiation efficiency is decreased to η
meas = 56% for the measured embedded antenna, associated with a lower gain of G
UEmeas = 13.7 dBi. The embedded array offers a 3-dB beamwidth of Θ
UE = ±12° and an impedance bandwidth B
emb > 4 GHz due to an increase in the overall thickness of the assembly.
All simulated and measured results are summarized in
Table 5. An uplink data rate of D
UL = 7 Mbit/s is anticipated for the measured free-space n257-band 4 × 4 circular patch antenna array. This value decreases to 6 Mbit/s for the embedded array due to the lower realized gain which, however, is still favorably suitable for the intended applications.
To address the advantages of the proposed UE antenna array, the performance comparison is tabulated in
Table 6. In [
21], the hybrid design of the slot and DRA is employed to realize dual-band circularly polarized function. However, the lower gain values will lead to a reduced data rate, which is not suitable for various 5G automotive mobility applications at the Ka band. In [
22] and dual-layer stacked patches for L- and S-band were utilized to realize dual-CP operation. In addition to a lower value of gain, 3-dB axial ratio bandwidth does not cover the operational bandwidth of B
op = 100 MHz. Moreover, the antenna structure operating at the K- and Ka-band in [
23] is complex in addition to having reduced gain. The proposed dual-band UE antenna terminal is verified through measurements after embedding into the rear spoiler of a passenger car compared to the above antennas. High-gain dynamic beam tracking antennas were developed for 5G satellite communications [
5,
6] but they are not feasible for automotive mass-market applications due to their significant complexity, size, and power consumption. In summary, the proposed work realizes a very compact, low-profile, and easy-to-manufacture circularly polarized dual-band UE nested antenna array.