High Gain and Wideband Fabry-Perot Resonator Antenna Based on a Compact Single PRS Layer

In this paper, a wideband and high gain Fabry Perot Resonator Antenna (FPRA) is proposed. It is based on a synthesized compact single partially reflective surface (PRS) layer acting as a superstrate to a slot-coupled feed antenna, which acts as a radiating source element for the proposed design. The PRS is based on a 2-D printed unit cell, where the lower part is a simple circular ring Frequency Selective Surface (FSS), incorporated with a synthesized FSS unit cell, using an automated system through a VBA based interface established between CST Microwave studio and Matlab, and optimized using a binary genetic algorithm. This new FSS layer acts as the upper part of the proposed PRS layer and provided a positive phase gradient. It almost perfectly resembles that of the optimum PRS over the desired frequency range, with a relatively high reflection magnitude, which makes it a promising superstrate candidate for wideband and high gain FP resonator antennas. The fabricated prototype achieved an important performance in terms of impedance bandwidth with 42%, ranging from 11.32 to 17.35 GHz. In addition, the 3-dB gain bandwidth is 36 % from 11.68 to 16.78 GHz, with a maximum peak gain of 14.72 dB achieved at 16 GHz. Consistent and almost invariant radiation patterns are achieved over the Ku-band frequency band of interest. The experimental and simulated results are in good agreement, justifying the feasibility of the proposed design as a high gain and wideband FP resonator antenna.

Therefore, the main objective of this study is to design a 90 wideband, compact, and low-profile FPRA with high gain. (BPSO) algorithm [25] can also be used instead of the genetic 100 algorithm to achieve the same objective of a compact with 101 positive reflection phase gradient PRS unit cell. 102 The proposed PRS design's main distinguishing features 103 are its compactness and small footprint in comparison to 104 dielectric-based designs, which have a complicated fabri-105 cation process (dielectric slab with gradient permittivity) 106 [20], [21], and a larger foot print. It is low in profile, easy-107 to-fabricate, and cost-effective design. In addition, using 108 3D printing techniques could reduce fabrication costs, 109 in combination with the proposed synthesizing system, and 110 could achieve a compact, low profile, yet effective design. 111 However, current 3D printing processes are far from achiev-112 ing the same performances as the PCB printing techniques 113 using small footprint structure as the one proposed here, and 114 not yet explored to the extent where it could replace it. For 115 example, in [26], a 3-D printed FPRA, was designed with 116 a paraboloid-shape superstrate for wider gain bandwidth in 117 comparison with the commonly adopted planar superstrate, 118 and achieved 22.2% gain bandwidth with a footprint of 4 λ 0 119 x 4 λ 0 x 0.66 λ 0 (λ 0 at 6GHz), in comparison to the proposed 120 design with a footprint of only 1.72 λ 0 x 1.96 λ 0 x 0.62 λ 0 121 and 3-dB gain bandwidth of 36%. 122 The simulation results yield an important improvement in 123 terms of impedance and 3-dB gain bandwidth, while keeping 124 the compactness of the final design footprint. The proposed 125 prototype was fabricated, and the results were verified exper-126 imentally, and they are in good agreement with the full-wave 127 simulation results. With these features, this antenna can be 128 a protentional candidate for the Ku-band applications with 129 higher gain and wideband requirements. 130 The proposed FPRA antenna as per our knowledge, is the 131 first of its kind in the literature that uses such Partially Reflec-132 tive Surface, that has been artificially synthesized, in a way 133 where it's reflection phase response, resembles that of the 134 optimum phase, which satisfies the cavity resonance condi-135 tion according to the ray-tracing analysis, to operate in wider 136 frequency band with relatively high gain.

137
The proposed designing process also, allows easy fre-138 quency tuning, using the proposed AI driven synthesizing 139 system. Hence, this antenna can be easily scaled to different 140 frequencies, which makes the operation more flexible. 142 Fabry-Perot resonator antenna (FPRA) is considered a highly 143 directive one [27]. It consists of a simple radiating source 144 backed by a ground-plane, and a partially reflecting surface 145 (PRS) (see Fig. 1).

146
Gain and directivity improvements are achieved, when 147 the spacing h c between the ground and the PRS, satisfies 148 the cavity resonance condition according to the ray-tracing 149 analysis, where the reflected waves from the ground and the 150 reflection phase response of the PRS are in phase, leading to 151 the maximum peak gain in the broadside direction [28], and 152 the spacing h c equals to the following: Assuming that the dielectric slab is lossless, the aperture 177 surface of the FPRA for a required directivity can be approx-178 imated using the following formula [29]: where λ is the operating wavelength of the FPRA.

181
By taking a PEC ground plane (ϕ GND = π), the PRS 182 reflection phase is expressed by rearranging (1) as the 183 following: where, Z d and Z 0 are the characteristics impedances of the 197 dielectric substrate and the air, respectively, k d is the dielec-198 tric phase constant, with a thickness of h PRS .

199
When the dielectric substrate uses Rogers RT/duroid 200 5880 dielectric slab (ε r1 = 2.2, tanδ = 0.0009), with a thick-201 ness h PRS of 1.575 mm, and according to (6), its reflection 202 coefficient phase response is depicted in Fig. 2, Where it 203 is clearly seen that the phase of the reflection coefficient 204 decreases as the frequency increases, when the height is 205 kept constant. Consequently, if the PRS shows positive phase 206 gradient as per (6) and from the results in Fig. 2, the FPRA 207 is able to operate in a wideband of frequencies. Moreover, 208 since the ground plane of the FPRA is a PEC(ϕ GND = π), 209 the Fabry-Perot antenna can operate in wide frequency band 210 when the PRS exhibits even smaller positive reflection phase 211 response, and not necessary covers the whole band of interest. 212

213
In the previous section, a PRS with a positive phase gradient 214 is discussed to be the main key point for designing a wideband 215 FP resonator antenna. Therefore, first of all, we should care-216 fully design one that meets the aforementioned requirement. 217 First, a simple circular ring is etched on one side of a 218 Rogers RT/duroid 5880 dielectric slab (ε r1 = 2.2, tanδ = 219 0.0009), with a thickness h PRS of 1.575 mm, creating the 220 initial PRS unit cell having the following design parameters: 221 w PRS =6 mm, r o =2.9 mm, r i =1.4 mm and g = 0.2 mm. 222 Fig. 3(a) shows the geometry of the initial PRS unit cell.

223
The PRS unit cell is designed and simulated using CST 224 Microwave studio, and the reflection phase characteristics 225 analyzing setup is depicted in Fig. 3   The well-established VBA-based connection between CST 283 Microwave studio and MATLAB's embedded genetic algo-284 rithm (GA) is used here, to create the automated pattern syn-285 thesizing system, where the upper side of the PRS structure 286 is the targeted area for this procedure. Each PRS unit cell 287 is pixelated into n x m pixels, and these latter are defined 288 using binary encoding, where having the value of either 1 or 0 289 indicates the presence or the absence of copper on the pixe-290 lated sub-cell (on top of the dielectric slab). Increasing the number of pixelated cells, increases also the 292 number of total possible structures, making it an impractical 293 way in solving such a problem, so, a confined search space 294 global optimization solution is needed, for finding the optimal 295 candidate, according to a predefined fitness function, making 296 the genetic algorithm an effective way in doing so.

297
In the proposed designing scenario, the fitness function is 298 defined as the root-mean-square error (RMSE) between the 299 reflection coefficient phase response of the generated PRS 300 unit cell under investigation, and the optimum phase, and is 301 defined as: where RMSE is defined as: The reflection coefficient phase of the PRS unit cell 306 is taken at equal frequencies fi, with the total number 307 VOLUME 10, 2022   a, where it is clearly shown that the final design doesn't 345 suffer from the infinitesimal connection problem, between 346 two sub-cells, when these constellations 10 01 or 01 10 are 347 present, which could lead to a malfunctioning PRS structure, 348 due to fabrication tolerances. Also, since the design imposes 349 the four-folded symmetry, the PRS unit cell exhibits polariza-350 tion insensitivity, which makes it also applicable to circular 351 polarized antennas.

354
The resulted reflection coefficient phase of the optimized 355 PRS has a positive gradient and is almost perfectly resem-356 bles that of the optimum PRS over a wide bandwidth 357 spectrum range, with a relatively high reflection magnitude 358 (See Fig. 7. b), and this can be a vital key point in designing a 359 wideband with high gain FPRA antenna, since higher reflective PRS leads to higher gain as per (2) & (3) but narrower 361 −3 dB bandwidth, and to broaden this latter a PRS with pos-362 itive phase gradient is a must to achieve that goal, as per (5).  The proposed feed antenna design is depicted in Fig. 8. 386 it consists of a parasitic patch, coupled to a feeding line 387 through a slotted ground plane (bottom layer), and spaced by 388 an airgap h air = 2 mm, for surface waves suppression since 389 this latter can contribute to performance degradation of the 390 FRPA cavity [37]. An impedance matching network is added 391 to the feedline to improve the bandwidth. The Rogers RT/ 392 duroid 5880 material having a thickness (h s ) of 0.787 mm, 393 permittivity ε r2 = 2.2, and a loss tangent of 0.0009, is used 394 as a substrate for both the parasitic patch and the bottom 395 layer (ground plane and the feed line). The antenna design 396 parameters are summarized in Table 1.

397
Both the simulated reflection coefficient (S 11 ) and the peak 398 gain of the feed antenna are illustrated in Fig. 9. a, where it 399 is noticed that the impedance bandwidth almost covers the 400 whole Ku-band spectrum, ranging from 12.6 GHz to 18 GHz. 401 Within this band, the peak gain increased from 7 to 8.8 dBi. 402 In addition, the co-and cross-polarization patterns at 16 GHz,403 in both the E-and H-planes are illustrated also in Fig. 9. b & c. 404 respectively, with low cross-polarization in both cuts, making 405 it a good candidate for being a feeding source for the proposed 406 FRPA antenna to be designed in the next section. The optimized PRS structure (Fig. 7) is applied as a super-409 strate layer to the proposed slot-coupled feed antenna in the 410 previous section (Fig.8), creating the proposed wideband and 411 high gain FP resonator antenna as shown in Fig. 10.

412
The PRS superstrate is suspended over the shared ground 413 plane at a distance h c using M2 nylon screws (2 mm in 414 diameter) at the four corners, and they are also taken into 415 account during the simulation process, alongside a 50 Ohm 416 VOLUME 10, 2022   Fig. 7 (b) and (1), where f r is the 420 FPRA's center frequency and ϕ PRS is taken also at that same 421 center frequency.

422
The reflection coefficient phase of the PRS at the center 423 frequency of 15GHz is about −158.6 • (see Fig. 7(b)), and by Thus, a PRS structure of an aperture size of 36mm×36mm 435 is used as the superstrate for the proposed FP resonator 436 antenna, which consists of an array of 6 × 6 PRS unit cells 437 since this latter's effective size is 6 mm.

438
Using the previously estimated parameters, namely the 439 aperture size (36mm x 36mm), and the cavity air gap h c = 440 10.5 mm, the resulted reflection coefficient S 11 of the pro-441 posed FPRA is illustrated in Fig. 11, where the impedance 442 bandwidth (|S11| <−10dB) is about 3.24 GHz divided into 443 a dual-band, ranging from 14.18 to 16.52 GHz, and from 444 17.1 to 18 GHz. The peak gain is 14.4 dBi at 14.5 GHz.

445
The air gaped cavity of the FP resonator antenna is an 446 important factor in designing a wideband antenna with high 447 gain; therefore, the influence of changing such a parame-448 ter on the overall performance of the proposed antenna is 449 investigated. All other parameters are fixed, while the cavity height is 451 changed in a step of 0.5 mm from 10.5 to 13 mm, to see the 452 effect of the cavity height variation on the S 11 impedance 453 bandwidth, the peak gain, and the 3-dB gain bandwidth of 454 the FPRA. The results are depicted in Fig. 11 & 12 and 455 summarized in Table 2. From there, it is noticed that when 456 the cavity thickness increases, both the impedance and 3-dB 457 gain bandwidths are gradually expanded, while the peak gain 458 is reduced, accordingly, as expected from an FP resonator 459 antenna.

460
For a high gain FP resonator antenna with a wider oper-461 ating bandwidth, and based on the results illustrated in both 462 Fig. 11 & 12, and Table 2, it can be concluded that the optimal 463 cavity height is 13 mm, where the proposed FPRA exhibited 464 an outstanding impedance bandwidth performance of 38.64% 465 ranging from 11.9 to 17.6 GHz (Fig. 11), and -a 3-dB gain 466 bandwidth of 39.11% (11.6 to 17.24 GHz), with a maximum 467 peak gain of 14.21 dBi at 16 GHz as shown in Fig. 12.

468
Another important parameter that considerably influences 469 the FPRA peak gain and bandwidth is its PRS superstrate size, 470 therefore, the influence of changing such a parameter on the 471 performance of the proposed antenna is investigated.

472
All other parameters are kept fixed, while the size of the 473 superstrate is changed by increasing the number of PRS unit 474 cells, to see the effect of this variation on the peak gain,  The best-achieved performance in terms of peak and gain 482 bandwidth is when a 6 × 6 PRS superstrate is used.

505
In this section, the prototype validation process is carried out,   The optimal cavity spacing is fixed to 13 mm for further 514 results validation and analysis. It is fed by a 50 Ohm SMA 515 connector soldered to the feed line with an offset of 5 mm 516 from the edge of the feed antenna.  Therefore, both simulation and measurement results are in 528 good agreement, despite the slight discrepancies between 529 them, which are probably due to assembly and fabrication 530 errors. Thus, the proposed antenna performed outstandingly 531 in terms of wide impedance bandwidth, covering almost the 532 whole Ku-band spectrum.

533
In addition, the peak gain and radiation patterns for the 534 proposed prototype were measured in an anechoic chamber, 535 and the setup is depicted in Fig. 14(b), by using the following where G horn is the gain of the standard horn antenna. 540 P horn and P AUT are the received power of the horn antenna 541 (receiver mode), and the antenna under testing, respectively. 542 The results are depicted in Fig. 15(b), confirming that by 543 using a PRS-based superstrate with a positively gradient 544 phase, the peak gain over a wideband of a frequency spectrum 545 is improved drastically.

546
From Fig. 15 (b), the measured maximum peak gain of 547 the proposed FPRA is 14.72 dBi at 16 GHz, with a 3-dB 548 gain bandwidth of 36% (5.1 GHz) ranging from 11.68 to 549 16.78 GHz, which is in good agreement with the simulation 550 results.

551
For further analysis, the normalized simulated and mea-552 sured radiation patterns of the proposed FP resonator antenna 553 are illustrated in Fig. 16 & 17 at three different frequencies, 554 12.5, 14, and 16 GHz, in both H-and E-planes. 555 The obtained radiation patterns (simulated and measured) 556 of the antenna, are directional in the broadside of both the 557 E-and H-planes, with lower side-lobe and cross-polarization 558 levels, and a narrower beam-width. Both simulation and 559 VOLUME 10, 2022 measurement results are in good agreement, with a slight 560 difference due to fabrication errors. With these features, the 561 proposed design could be a potential candidate, allowing a 562 high and consistent gain over a larger operating bandwidth, 563 and low profile, for applications in the Ku-band spectrum. 564 Finally, the proposed FPRA design is compared to others 565 in the literature, taking into account the bandwidth, the maxi-566 mum peak gain, the overall size of the design, and the number 567 of PRS layers. The comparison is summarized in Table 3.  The proposed prototype has been fabricated and measured.