3-D Coverage Beam-Scanning Antenna Using Feed Array and Active Frequency-Selective Surface

This paper presents the design of a smart antenna that can achieve 3-D beam-scanning coverage. The antenna consists of a novel planar feed array and a cylindrical active frequency-selective surface (AFSS). First, an array fed metallic reflector is studied as a reference antenna to validate the beam-scanning characteristics in the elevation plane. Then, the AFSS is assessed through simulation and measurement results. Finally, the complete structure containing the planar collinear array and the AFSS is analyzed. A prototype at S-band has been designed, manufactured, and measured. The resulting antenna is shown to be able to operate at the 2.4–2.5 GHz frequency band and switch beams in both the azimuth and elevation planes. In the azimuth plane, the proposed antenna is capable of sweeping beams toward different directions to cover a full range of 360°. In the elevation plane, it can achieve beam steering within an angle range of +16°/ − 15°. The measured maximum gain of the antenna is 9.2 dBi. This is the first report of a low-cost 3-D coverage beam-scanning antenna based on AFSS.

example, active electronically steered array radars. The cost and complexity are mainly proportional to the number of array elements that connect to RF/microwave phase shifters, transmit/receive (T/R) modules, and power distribution networks. To facilitate the future 5G applications, it is expected that smart antennas would have a compact size, low power consumption, and are affordable to a larger number of users in contrast to conventional phased arrays. Another design consideration is the beam-scanning coverage. For example, in interbuilding wireless link scenario, it is desirable that smart antennas offer a full 360°angular coverage in the horizontal plane and steerable beams in the elevation plane to provide sufficient signal strength to dynamic users in different levels of the buildings.
As an alternative solution, reconfigurable antennas featuring switchable/tunable frequency/polarization/patterns can be used to increase system diversity level and achievable channel capacity while reducing prohibitive system costs [2], [3]. Similar to conventional smart antennas, pattern reconfigurable (switched beam) antennas have attracted extensive research interest to improve the antenna performance by applying new reconfiguring techniques. For instance, ESPAR antennas proposed in [4], [5] offer a cost-effective implementation, where p-i-n diodes/varactors are loaded at the parasitic antenna array elements instead of using phase shifters to steer radiation patterns in the azimuth plane. In [6], a circular Yagi-Uda array incorporates liquid metal mercury into the director and reflector elements to reconfigure the antenna radiation patterns. Recently, electromagnetic structures such as metamaterials and frequency-selective surfaces (FSSs) have also been engineered to achieve antenna beam switching/steering [7]- [10] thanks to their fabrication simplicity and enhanced reconfigurability using various tuning technologies. Ji et al. [11] present a beam steering partially reflective surface antenna fed by a two-element patch element array. It is shown that beam scanning from −15°to +15°off the broadside direction can be achieved in the frequency range from 5.5 to 5.7 GHz with a maximum gain of 12 dBi. To the best of the authors' knowledge, few solutions can be found in the literature which can realize 3-D beam coverage using reconfigurable FSSs, covering both the azimuth and elevation planes.
Electronic beam tilting has been used in multisector base station antennas to lower intersymbol interference in the network. Much effort has been devoted to developing tilted beam antennas [12]- [15]. Recently, Dadgarpour et al. [16] report This work is licensed under a Creative Commons Attribution 3.0 License. For more information, see http://creativecommons.org/licenses/by/3.0/ a tilted beam technique for planar endfire antennas. Double G-shaped resonator arrays are mounted in the antenna H-plane to create a 35°beam tilt in the elevation plane and a maximum gain enhancement of 5 dB at 3.6 GHz. However, the reported antennas cannot steer the main beam to cover the entire horizontal plane while maintaining the beam tilting in the elevation plane. In this paper, for the first time, we propose a 3-D beam coverage antenna using a reconfigurable cylindrical FSS illuminated by a novel planar feed array. The compact planar array can be easily fabricated and connected to external feed networks to achieve beam scanning in the elevation plane. Based on the study and optimization of the antenna structure that combines the feed array and the active FSS (AFSS) cylinder, the resulting antenna can steer the main beam in both the elevation plane and azimuth plane simultaneously.

II. 3-D COVERAGE BEAM-SCANNING MECHANISM
The concept of switching the antenna main beam only in the azimuth plane has been demonstrated in [17], where a conventional monopole antenna is explored as the feed. To fulfill beam steering in the elevation plane, we propose a novel collinear feed array excited by progressive phase shifts between elements. Fig. 1 compares different feed schemes inside a reconfigurable FSS cylinder. Increasing the number of elements increase the directivity of the array, and from the beam optimization viewpoint, a larger number of feed elements leads to more freedom to apply sidelobe reduction/beam-shaping techniques to the array.

A. Metallic Reflector Validation
A semicylindrical metallic reflector fed by a dipole is investigated to validate the feasibility of the concept using AFSS [8]. In our design, by exciting each element of the array with specific amplitude and phase, beam scanning in the elevation plane can be achieved. Indeed, the design concept is inspired by a conventional base station antenna design, which consists of incorporate feed dipole array over a planar metallic reflector [18]. Fig. 2 shows a two-element metallic half-wavelength dipole linear array placed along the axis of the semicylindrical reflector whose radius R = 60 mm. The length of the reflector is H = 250 mm, and the interelement spacing D e is set to be 70 mm. The simplest method to excite a uniform linear array is to assign the array ports with equal amplitude and a progressive phase shift ϕ. For base station applications, the maximum phase shift is given by the N-element array and the scan angle θ   where k = 2π/λ, d = 0.5λ. The structure in Fig. 2 is simulated in HFSS with lumped ports. Fig. 3 shows that the attainable beam scan range is up to ±25°with a maximum sidelobe level (SLL) below −10 dB. The simulated directivity difference across the scan range is less than 2 dB. To optimize the beam tilting performance, three parameters: H , R, and D e are examined. It is clear from Fig. 4 that the length of the reflector has almost no effect on the antenna directivity, while for the shorter reflectors the SLL is increased by 5 dB. Fig. 5 depicts the effects of the reflector radius. It is noted that larger R leads to extended beam tilt angle at the cost of raised SLL. Finally, Fig. 6 shows that enlarging the spacing between elements just slightly degrades the SLL. Based on the above analysis, the size of the reflector and the element spacing will be used for the following active FSS reflector design. Next, the scan performance of the four-element array is also carried out with the same R but H = 320 mm, as shown   in Fig. 7. Fig. 8 shows that by applying identical magnitude and phase shift excitation to the four ports, the radiation pattern scans over the range of ±35°with SLLs below −10 dB when the phase difference between the ports is ±120°. Also, the maximum directivity is 14 dBi which is 3 dB larger than the two-element case. Moreover, array synthesis methods can be adopted to provide the desired beam patterns. The most common technique is to introduce an amplitude taper across the array to reduce the SLL, and thus expanding the useful scan angle.

B. AFSS Unit Cell Design
As elaborated in [17], incorporating p-i-n diodes into slottype FSS elements enables electronic switching of the FSS between reflective or transparent mode. An H-shaped AFSS unit cell based on a switchable slot is designed to operate at 2.45 GHz, as shown in Fig. 9. Similar to a dipole antenna, a slot FSS resonates when the slot length of each element is half wavelength. By Babinet's principle, a slot FSS with slot length S x embedded in a homogeneous medium characterized by (ε and μ) has a resonance frequency around (2)  More specifically, with slot width S y , the resonance frequency can be calculated by In practice, it is difficult to incorporate a p-i-n diode into the single slot FSS copper layer. However, this issue can be addressed by placing the entire p-i-n diode bias network at the other side of the substrate [19]- [21]. Mylar polyester film with ε r = 2.7, loss tangent of 0.0023, and h = 0.005 mm is used as the substrate because: 1) it is flexible, strong, and suitable for conformal antenna design and 2) it has almost no effect on the AFSS performance while the p-i-n diode is located extremely close to the slot. The dimensions of the unit cell are given in Table I. The whole structure is simulated in CST using Floquet mode boundary setup. Specifically, the p-i-n diode ON/OFF states are modeled as series or parallel RLC lumped elements. For the ON state, the diode is simplified as a forward resistor R s = 2.1 , and as an equivalent capacitor, C s = 0.17 pF in the OFF state. Note that the H-shaped slot can reduce the FSS array periodicity P x , enabling accommodation of more cells for a fixed cylinder perimeter. Compared with the dimension in [17], the diameter of the AFSS cylinder consisting of H-shaped unit cells can reduce by one effective wavelength size. The simulated transmission coefficients of the AFSS are shown in Fig. 10. It is found that by inserting an OFF-state diode into the gap of the bias line, a capacitance is introduced which shifts the resonance frequency from 2.75 to 2.45 GHz. The simulated ON-state frequency response exhibits about −14 dB transmission magnitude, which indicates about 96% EM wave energy has been reflected. Thus, by switching the p-i-n diodes, the AFSS can be reconfigured as an EM transparent window or a reflector.

C. Array Fed AFSS Cylinder Validation
To imitate the radiating mechanism of a corner reflector antenna in Fig. 2, the planar AFSS was rolled into a cylinder which has the same dimensions of the metallic reflector. First, the numbers of the unit cells along the circumference and axis: N x and N y can be calculated from N x = 2 × π R/P x and N y = L/P y , where R = 60 mm and L = 190 mm. Next, the same linear arrays feeding the metallic reflector cases in Section II-A are employed to illuminate the AFSS reflector. Note that the more unit cells along the cylinder circumference, the more beams that can be generated in the horizontal plane. The whole antenna configuration was simulated using CST frequency-domain solver. Fig. 11 shows the radiation patterns of the AFSS cylinder fed by two-and four-element arrays. In Fig. 11(a), the cylindrical AFSS consists of eight columns and eight rows of unit cells whose dimensions are given in Table I. The p-i-n diodes in the left-half cylinder are switched ON to activate the reflecting mode of the left-half AFSS, whereas the other half diodes are all OFF, which allows the EM wave to pass through the right-half cylinder (half of the AFSS has been made transparent for clarity). By varying the phase shifts applied to the array element ports, similar beam tilting trends can be obtained. The maximum beam tilt angle with an SLL of −10 dB is reduced to ±21°, as shown in Fig. 12. Because the AFSS cannot absolutely replace the metallic reflector, the directivities of the scanned beams are slightly lower than the ones in Fig. 3. Fig. 13 shows the steered E-plane radiation patterns of the four-element array fed AFSS cylinder which consists of 8 columns and 12 rows shown in Fig. 11(b). It can be seen that the maximum beam scan angle is ±30°, which is 5°less than the one in Fig. 8.

III. FEED ARRAY DESIGN
In this section, a multilayer omnidirectional antenna array with corresponding power distribution networks is proposed to replace the ideal collinear dipole array in Section II.

A. Antenna Feed Array Element
It is desired that omnidirectional antenna elements should be fed collinearly and placed axially inside the AFSS cylinder. In [22], a simple stripline structure is introduced to serially feed loop slots etched on each metal ground. As described in [23] and [24], significantly reducing the size of the ground plane makes the H-plane pattern of the antenna nearly omnidirectional. Thus by arranging two identical rectangular patches back-to-back along the broadside direction, the radiated far field of the two elements is added to form an omnidirectional pattern in the azimuth plane. As shown in Fig. 14, the multilayered element consists of two identical patches, which are fed by a stripline through the two slots in the ground plane. The rectangular patches are etched on Arlon Cuclad 217 (ε r = 2.17 and loss tangent = 0.0009) with a thickness of 0.508 mm. Because the substrate is quite flexible, two pieces of ROHACELL foams are used as the spacer to support the patches. The substrate used for the stripline is 0.762 mm thick Arlon AD255A whose ε r is 2.55 and loss tangent is 0.0015.      Table II. It should be noted that if a harder substrate is used, this design does not require any foam layer as nylon screws can be used to fix the patch layers.

B. Antenna Feed Array Design
Using the above element, a two-element parallel fed array is designed and experimentally tested. Fig. 16 depicts the configuration of the array antenna which consists of two elements with a center-to-center spacing d e . The feeding points at the bottom can be connected to phase shifters and feed networks through stripline to coaxial/microstrip line transition. To avoid grating lobes, the element spacing d e is kept around 0.5-0.7λ in the substrate at the desired frequency. Fig. 17 shows the detailed feedline layout of the stripline. In order to maintain a good isolation between the two elements, the feedline of the upper element is offset some distance f 2 from the coupling slot of the lower element as far as possible. The stripline is designed to have a 50    Table III presents the parameters of the array design. One profound advantage of this design over the conventional metallic collinear dipole array is the feedlines hidden between the grounds, which allow one of the feedlines to pass through the lower element without affecting its performance. The simulated S-parameters are presented in Fig. 18. It can be observed that with respect to below −10 dB S 11 , the antenna array covers the frequency bands from 2.51 to 2.62 GHz (S 11 will shift toward lower frequency when integrating with AFSS). Moreover, the isolation between two ports is larger than 25 dB.
IV. AFSS ANTENNA DESIGN, FABRICATION, AND MEASUREMENT To integrate the AFSS and the feed array, measurements were carried out to ensure the performance of the final structure. First, the transmission coefficients of the proposed planar AFSS in Section II were measured in a plane wave chamber, where two wideband log-periodic antennas are used as the transmitting and receiving antennas. The flat AFSS sheet is comprised of 8 × 8 unit cells with dimensions in Fig. 9 and Table I and fixed in a support between the two antennas. Fig. 19 shows the fabricated FSS and its bias network. Eight parallel lines with p-i-n diodes in series are etched on one side of the substrate whereas, on the other side is the copper layer with FSS slots which is used as the common ground for the lines through a metallic via. Thus, it is convenient to control each column with a 5 V dc voltage and there is only one line connected to the ground. RF inductors are selected as chokes to isolate the RF signal   and a 1 k resistor is added to the ground line to limit the current. Fig. 20 compares the simulated and measured values of S 21 when switching ON/OFF all the p-i-n diodes. It can be noted that the performance of the fabricated planar AFSS agrees well with the simulated one except that there is 1 dB transmission loss at the resonant frequency.
Next, the feed array with the dimensions in Tables II and III was fabricated, as shown in Fig. 21. Fig. 22 shows the measured S-parameters. It is clear that the S 11 and S 21 follow the same trend as the simulated results in Fig. 18. However, the measured port isolation is less than the predicted results but still larger than 20 dB which is sufficient for most cases. Two coaxial cables are connected to the feed ports of the two antenna elements, which are also for attaching external power combining networks with required phase shifts. Fig. 23 illustrates the measured radiation patterns of the two array ports. As can been seen from the H-plane results, omnidirectional patterns (deviation from omni ≤ 1.5 dBi) can be achieved using the presented antenna elements.
The final stage is to combine the cylindrical AFSS and the feed array. To fully investigate the antenna characteristics,  the entire structure in Fig. 24 is simulated in CST using the frequency-domain solver. Note that all the p-i-n diodes in the gaps of the slots are omitted for clarity. The feed array is placed in the center of the cylinder parallel to the xoz plane. The bottom two feed points are extended beyond the cylinder to connect the feed network circuits. Within this configuration, where the solid sector (yellow) acts as a reflector by switching ON all the diodes in it, the phase shifts required are obtained by optimizing the progressive phase difference and element spacing in the simulation. Then, the attained phase shifts are applied to the microstrip lines of the feed network, allowing the array pattern tilted to the desired angle. Fig. 25 compares the simulated and measured S-parameters of the array fed AFSS antenna. It is clear that the resulting antenna has a below −10 dB S 11 bandwidth covering 2.4-2.5 GHz. Meanwhile, the isolation of the two ports is found to be reduced to around 14 dB due to the presence of the AFSS cylinder. To compensate this adverse effect, the coupling between the antenna elements and the single-element reflection  coefficient has to be kept as low as possible in order to allow an independent design of the array feed distribution and the feed network. The array element spacing is a key parameter to tune the coupling, but it also affects the beam scan angle and SLL. According to the simulated results, the required phase shifts for the up and down tilt beams are 45°and 245°, respectively, when the beams are pointed at ±15°, with maximum −10 dB SLLs. Thus two individual feed network circuits including the required phase shift values are simulated with the structure in Fig. 24. In Fig. 26, the simulated and measured reflection coefficients of the final structure with external feed networks are compared. For the two tilt angles, both designs can satisfy the 2.45 GHz WIFI frequency band. The fabricated antenna under test is shown in Fig. 27. The feed array is inserted into a hollowed cylindrical foam, which supports the AFSS. The whole structure is supported in turn by a vertical section of foam, to minimize scattering.
The beam tilting performance of the proposed antenna was experimentally validated in an anechoic chamber. Figs. 28 and 29 show the comparison between the simulated  radiation patterns and the measured ones in the E-plane and H-plane. In Fig. 28(a), the simulated up and down tilt angles are 73°and 105°, respectively, with below −10 dB SLLs. Fig. 29(a) shows the measured up tilt angle has a 2°decrease over the operating frequencies from 2.4 to 2.5 GHz but a 3 dB lower SLL at 2.5 GHz. The measured down tilt scan angles are 2°larger than the simulated ones at 2.45 and 2.5 GHz. Also, there is a slight rise in SLL for the down tilt beam at 2.5 GHz. Note that for the measurements, a half sector is biased to act as a reflector. It is expected that eight identical up/down beams can be generated if we switch ON/OFF half of the AFSS cylinder in turn due to the symmetry of the antenna structure. Compared with the simulated results of the array fed metallic reflector, the scan angle for the same SLL (−10 dB) is reduced mainly because the AFSS is not ideal as a conventional reflector. Furthermore, unwanted scattering happens when EM waves pass through the OFF-state AFSS, which degrades the beam tilting performance as well. Fig. 30 shows the measured eight beams in the horizontal plane for the up and down tilt angles at 2.45 GHz.  Table IV compares the characteristics between several beam-switching antennas recently reported and our design. It can be seen that the proposed antenna has the advantages of being able to achieve beam scanning in both the horizontal and elevation planes with a moderate gain. In addition, the dimensions of our antenna are smaller than the ones in [17] which also uses a slot-type active FSS cylinder.

V. CONCLUSION
Low-cost smart antennas are important for a wide range of applications in wireless communications. This paper presents the first report of a 3-D coverage beam-scanning antenna utilizing an active FSS. A prototype at 2.45 GHz has been designed, fabricated, and measured to verify the design principles. In the azimuth plane, the proposed antenna is shown to be able to achieve electronically beam switching covering a full 360°range. In the elevation plane, the antenna achieves a scan range of +16°/ − 15°. A measured maximum gain of 9.2 dBi is obtained and the antenna is scalable to achieve a higher gain. Compared with traditional smart antennas, this antenna does not require a large number of microwave phase shifters, has a simple structure and low cost, and thus is promising for base station applications in wireless communications.