Printed Dipole Antenna Array With Reconfigurable Feeding Network for Wide Elevation Angle of U2X Communications

In this paper, a printed dipole antenna array with reconfigurable feeding network for wide elevation angle of U2X communications is presented. The proposed array consists of a printed 8-element dipole antenna array in the top substrate, two supports with a tapered balun and two reflectors with the reconfigurable feeding network in the bottom substrate. To achieve high gain and wide beam coverage, the proposed reconfigurable feeding network consists of single-pole double-throw (SPDT) switches, single-pole three-throw (SP3T) switches, Wilkinson power dividers, and hybrid couplers. The proposed array antenna forms a 3×3 beam consisting of broadside and end-fire beams based on the x- and y-axis using a reconfigurable feeding network. The measured impedance bandwidth, peak gain, and half-power beamwidths (HPBWs) at ϕ=0° and ϕ=9° of the proposed array are approximately 1250 MHz (25%), 7.9 dBi, 197° and 100°, respectively. The overall volume size of the proposed array is 75×96×21 mm3 (1.3λ0×1.6λ0×0.4λ0 at 5.1 GHz).

motion of aerial vehicles and handsets [11]- [12]. Research has been conducted on an antenna for UAV communications that satisfies these conditions [13]- [17]. In [13], an omnidirectional antenna with a low height using a buried unipolar patch antenna was implemented by making like the radiation characteristics of a monopole. [14] is a helical antenna, and the antenna size is effectively reduced by using hexaferriteglass composite as the antenna core. In [15], a low-profile antenna was implemented using a magnetic dipole antenna. [16] is a conical patch antenna array using the flexible substrate and has wide elevation angles and high gain. [17] is a CP patch antenna with small size and uniform elevation angle. The antennas of [14]- [17] all have low height and operate at 2.45 GHz. The reason why these antennas are designed to operate at 2.45 GHz is that the conventional UAVs have been communicated in the S-band with 2.45 GHz as the center frequency. However, since this frequency band is the same frequency band as Wi-Fi, there is a possibility of crosstalk. If the signal of the UAV and Wi-Fi signal are confused, the UAV may become out of control or malfunction, leading to an accident or damage. Therefore, it is necessary to allocate a dedicated frequency band for UAVs. In 2012, through the World Radio Communication Conference (WRC-12) under the International Telecommunication Union (ITU), the frequency band for wireless communication of UAVs (5.03 to 5.091 GHz) and the frequency band for missions (5.091 to 5.15 GHz) has been assigned. Accordingly, studies have been conducted on high-gain antennas with low air resistance operating in the 5 GHz band [18]- [20].
In [18], an antenna with very high gain is implemented using a patch antenna array, phase shifters and amplifiers. [19] produced an antenna with a high gain and wide elevation angle using Yagi-Uda dipole antenna array. [20] achieved a high gain and wide elevation angles by implementing a relatively low-profile antenna by using a meander monopole and a circular patch antenna. Previous studies on antennas for UAV communications operating in the 5 GHz band were also conducted. One of our previous works was a wide beam coverage dipole antenna array with parasitic elements [21]. The proposed array has a wide elevation angle due to the influence of the parasitic element, but there is a limitation to cover the whole hemispherical region due to single beam. Another previous work was a UAV communication antenna array with a 3 × 2 switched beamforming network that forms three beams [22]. The proposed array formed three beams to improve the elevation angle and the gain. The main beam directions(φ) of the three beams are -90 • , 0 • , and 90 • respectively. The proposed array has a wide elevation angle with respect to the x-axis and can transmit and receive signals of vertical and horizontal polarization. In [23], the limitations of the previous two studies were supplemented by forming a quadruple- mode beam. The proposed array forms not only vertical polarization and horizontal polarization, but also polarization inclined at 45 • and polarization inclined at 135 • , so it has the advantage of being able to transmit and receive at various angles. In addition, it has the advantage of having a uniform beam elevation angle in the elevation plane by forming a quadruple mode beam. However, for stable communication, an array antenna with higher gain, wider beam elevation, and various polarization directions than the antennas of [18]- [23] is required. Therefore, in this paper, to improve the performance of previously proposed antennas, we propose an 8-array dipole antenna with a wide elevation angle and high gain various polarization directions using a linear array antenna theory, parasitic element, and reconfigurable feeding network. Simulations have been conducted using CST Microwave Studio [24] which show good agreement with experiments. Fig. 1 describes the proposed array with widecoverage beam patterns in the U2X communications.

II. ANTENNA ARRAY DESIGN THEORY
A typical half-wave dipole antenna has an omni-directional radiation pattern and a low gain of 2.15 dBi. In order to use a

Broadside (Center)
End-fire ( axis) y- x y z (d)  half-wave dipole antenna for UAV communications, it must be designed to improve gain and have a wide radiation pattern for θ. To satisfy the design conditions, by applying the linear array antenna theory to the dipole array antenna, a radiation pattern having a high gain and wide elevation angles can be formed. VOLUME

A. N-ELEMENT LINEAR ANTENNA ARRAY THEORY
In order to obtain the radiation pattern of the N-elements linear antenna array, as shown in Fig. 2, N-elements with current applied horizontally on the y-axis are arranged with a distance of d on the z-axis [25]. The array factor for this can be expressed by (1).
In (2), k, θ, and β are defined as the wave number, angle between the observation point and the antenna array axis, and phase of each antenna, respectively. (3) can be obtained by subtracting the expression (1) multiplied by e jΨ from (1).
The maximum radiation of an array directed normal to the axis of the array [broadside; θ = 90 • of Fig. 2] can be formed through the arrayed antennas and the array factor of (4). To optimize the design, the maxima of the single element and of the array factor should both be directed toward θ=90 • . Referring to (4), the first maximum of the array factor occurs when Ψ = kdcosθ + β = 0 Since θ = 90 • , then β = 0. Thus to have the maximum of the array factor of a uniform linear array directed broadside to the axis of the array, it is necessary that all the elements have the same phase excitation (in addition to the same amplitude excitation). To avoid any grating lobe and occur only maximum at broadside, the spacing between the elements should be λ/4. Instead of having the maximum radiation broadside to the axis of the array, it may be desirable to direct it along the axis of the array (end-fire). As a matter of fact, it may be necessary that it radiates toward only one direction (either θ = 0 • or θ = 180 • of Fig. 2). Substituting θ = 0 • (or 180 • ) for (2), then Ψ = -kd (or kd). Thus end-fire radiation is accomplished when β = -kd (for θ = 0 • ) or kd (θ = 180 • ). If the element separation is d = λ/2, end-fire radiation exists simultaneously in both directions (θ = 0 • and θ = 180 • ). To have only one end-fire maximum and to avoid any grating lobe, the spacing between the elements should be λ/4.
When all the elements have the same amplitude excitation and the phase difference between unit antennas is 90 • and -90 • , respectively, the array factor of the end-fire beam is defined by the equation (6). According to (6), the main beam direction is θ = 0 • (or θ = 180 • ) when the phase difference between the unit antennas is 90 • (or -90 • ). The radiation patterns of the maximum broadside and end-fire beams are shown in Fig. 3.

B. OPERATIONAL PRINCIPLE OF THE PROPOSED ARRAY
The dipole array antenna can form a broadside beam and two end-fire beams using the linear array antenna theory [22]. Theoretically, when there is no phase difference of each unit antenna, the radiation pattern of the array antenna radiates perpendicular to the array axis, but the rear radiation is reduced due to effect of the metal reflector placed behind the antenna, so Fig. 4(a). As in, it radiates only along the z-axis. When the phase difference between unit antennas is -90 • or +90 • , the end-fire beam radiates to the +x and -x axes, respectively, but contrary to the theory, the radiation to thez axis is reduced by installing a metal reflector behind the antenna. As shown in Figs. 4(b)-(c), the main beam angle of each radiation pattern is inclined by about 45 • along the z-axis. In the previous study, one broadside beam and two end-fire beams were formed with a 1×4 array antenna in the x-axis direction by applying the antenna theory. However, in this case, there is a limitation that the beam width of the yz-plane is relatively narrow. To overcome the limit of the previous antenna, in this paper, the 2×4 dipole array antenna was designed by placing the 1×4 dipole array antenna once more at a distance of about λ/2 in the y-axis direction. The 2×4 dipole array antenna can form a broadside beam and an end-fire beam in the yz-plane by controlling the phase difference between four antennas in the first row and four antennas in the second row. (Fig. 5). When forming the beams of modes 1, 4, and 7 in Fig. 5(d), four antennas located in the x-axis direction are used (A1, A2, A5, A6). In this case, the reason why eight antennas are not used is that the beam width becomes sharper as the number of array elements increased, and the elevation angles in the x-axis direction becomes narrow. Therefore, the elevation angles were widened using only four antennas (A3, A4, A7, A8) located in the x-axis direction. By the same principle, beams of mode3, 6, and 9 were formed with 4 antennas located in the -x-axis direction. And when implementing mode 2, 5 and 8 radiating from the center of the x-axis, all 8 antennas were used to increase the gain. In addition, using the experiment in [21], parasitic elements shorter than the length of the antenna were placed between each antenna. The parasitic element can adjust the directivity of the beam like the waveguide of the Yagi-Uda antenna. Therefore, in this paper, the parasitic element was used to widen the beam width by dispersing the beam of the array antenna ( Fig. 5(c)). Therefore, the proposed array antenna has a wide elevation angle and high gain by forming a 9-mode multiple beam using array antenna theories and parasitic elements. Table 1 shows each distribution of amplitudes and phase variations with regard to the input modes. A block diagram of 1×8 reconfigurable feeding network used in the proposed array is shown in Fig. 6. The proposed reconfigurable feeding network consists of four single-pole-doublethrow (SPDT) switches, two single-pole-three-throw (SP3T) switches, three Wilkinson power dividers, and four hybrid couplers. First, in the reconfigurable feeding network-1, there are a transmission line having the shorter electrical length and a transmission line having the longer electrical length by λ/4 (the transmission line with a phase delayed by 90 • ). One of two transmission lines type is selected through SPDT-1 and SPDT-3 (SPDT-2 and SPDT-4). Accordingly, the phase difference (0 • and ±90 • ) between the dipole antennas in the first row (A1-A4) and the dipole antennas (A5-A8) in the second row is determined through the reconfigurable feeding network-1. As a result, broadside beams radiating in a direction of 0 • and end-fire beams radiating in a direction of ±20 • with respect to the y-axis (φ=90 • ) are formed through the reconfigurable feeding network-1. Reconfigurable feeding network-2 is designed to form a broadside beam radiating in the direction of 0 • with respect to the x-axis like [3] and an end-fire beam radiating in a direction of ±45 • . When the P6 mode is selected in SP3T-1, only the outputs A1(0 • ) and A2(90 • ) are operated through the coupler and the transmission line with a phase delayed by 180 • to form an end-fire beam radiating to the x-axis. Next, when the P7 mode is selected in SP3T-1, power is distributed to two hybrid couplers by the Wilkinson power divider, so that the four outputs (A1-A4) have a phase difference of 90 • , 0 • , 0 • , and 90 • . This results in a centrally radiating broadside beam. Finally, when the P8 mode is selected in SP3T-1, only the outputs A3(90 • ) and A4(0 • ) are operated through the hybrid coupler and the transmission line with a phase delayed by 180 • to form an end-fire beam radiating to the x-axis. The phase and amplitude of A4-A8 according to the state of SP3T-2 are determined in the same way as the control method of A1-A4 by the state of SP3T-1. Fig. 7 represents the overall configuration of the proposed array. The proposed array structure consists of an upper substrate, a middle substrate, lower substrate, and supports as shown in Fig. 7(a). As shown in Fig. 7(b), the horizontal spacing between each antenna element is 1/4λ 0 and the vertical spacing between each antenna element is 1/2λ 0 (D r = 15.5 mm, D 1 = 36 mm), parasitic elements are placed between radiators at 1/12λ 0 (D p = 5.2 mm) spacing and used as an important parameter to widen the elevation angle. In addition, the metal reflector is the ground plane of the proposed reconfigurable feeding network and takes part in an important role in reducing backside radiation and improving the radiation gain. antennas, as shown in Fig. 7(c). The reconfigurable feeding network-1 and the reconfigurable feeding network-2 are connected through Semi-Regid cables and SMA connectors. All substrates used TACONIC RF-35 (ε r = 3.5 and tan δ = 0.0018). Their thicknesses are 0.76 mm and the copper thicknesses are 18 µm. The total size of the proposed array is 75×96×21 mm 3 (1.3λ 0 × 1.6λ 0 × 0.4λ 0 at 5.1 GHz).

IV. EXPERIMENTAL VERIFICATIONS
To verify the proposed design, the fully functional prototype of the proposed array in Fig. 8(a) was fabricated. In Figs. 8(b)-(c), the SPDT ( [29]) and SP3T ( [30]) switches for selecting and making nine linear polarization modes were CG2430X3 (absolute maximum input power = 33.5 dBm, 1 dB input compression point = 32 dBm) made by CEL, SKY13408-465LF (absolute maximum input power = 40 dBm, 1 dB input compression point = 34 dBm) made by Skyworks, respectively. The 90 • hybrid coupler is CMX55Q03 by RN2 Technologies ( [31]). Fig. 8(d) shows the topveiw of the fabricated radiators. Fig. 9 shows the simulated and measured reflection coefficients of the proposed reconfigurable feeding network with regard to different modes (mode 1 to 9). The S 11 to S 99 are the reflection coefficients of the RF input port (P1) and the output ports (A1, A2, A3, A4, A5, A6, A7, and A8), respectively. The measured results have a good agreement with the simulated results. In addition, the reflection coefficients of the proposed reconfigurable feeding network has a measured results of -10dB or less in all modes and within 3.2-6.0 GHz. Therefore, it can be seen that the proposed reconfigurable feeding network operates normally in the 5 GHz band. Fig. 10 shows the simulated and measured transmission coefficients of the proposed reconfigurable feeding network with regard to different modes (mode 1 to 9). The S 21 to S 91 are the transmission coefficients between the RF input port and the output ports, respectively. In modes 1, 4, 7 (Figs. 10(a), (d), (g)) with outputs of A1, A2, A5, and A6, the transmission coefficients is approximately minimum - . Simulated and measured phase differences of the proposed reconfigurable feeding network with regard to different modes: (a)-(i) simulated and measured phase differences.
11.2 dB and maximum -7.3 dB. It is included power loss of two SPDT switches, one SP3T switch, one power divider, and one hybrid coupler, respectively. In modes 2,5,8 (Figs. 10(b), (e), (h)) with outputs of A3, A4, A7, and A8, the transmission coefficients is approximately minimum -9.5 dB and maximum -7.3 dB. It is included power loss of two SPDT switches, one SP3T switch, one power divider, and one hybrid coupler, respectively. In modes 3,6,9 (Figs. 10(c), (f), (i)) with outputs of A1 to A8, the transmission coefficients is approximately minimum -13.8 dB and maximum -10.5 dB. It is included power loss of two SPDT switches, one SP3T switch, two power dividers, and one hybrid coupler, respectively. Since the measured results have a good agreement with the simulated results, it can be confirmed that the proposed reconfigurable feeding network operates normally without cross-talk or coupling. Fig. 11 shows the measured phase differences of the proposed reconfigurable feeding network with regard to different modes (Mode 1 to 9). Referring to Fig. 6, the maximum error of the measured phase difference for each mode is -8.3 • , so it can be seen that the intended phase difference was wellformed in the manufactured reconfigurable feeding network. In addition, it can be seen that the proposed reconfigurable feeding network is well designed because the simulated phase difference and the measured phase difference are almost identical.   The measured total efficiencies at 5.1 GHz from a minimum of 38% to a maximum of 49%. The reason that the total efficiencies do not exceed 50% is that the total efficiencies are lowered due to the loss caused by using the RF switches and the half power loss caused by using the Wilkinson power dividers and hybrid couplers. Radiation patterns for each mode in the yzand zxplane are shown in Figs. 14(a)-(d). Figs. 14(a)-(b) show the simulated and measured radiation patterns for each mode in the yz-plane. When comparing Figs. 14(a) and (b), it can be seen that the simulated radiation patterns and the measured radiation patterns are almost good agreement. Figs. 14(c)-(d)   show the simulated and measured radiation patterns for each mode in the zx-plane. When comparing Figs. 14(c) and (d), it can be seen that the simulated radiation patterns and the measured radiation patterns are almost identical. The total HPBW of the beam pattern for each mode of the proposed array is 197 • . Table 2 shows a comparison table of the antenna performance of the proposed array antenna and the previous study case. First, the antennas of [13]- [17] cannot all be used in the 5 GHz band, and they have a narrow 10-dB bandwidth, so it seems that there will be limitations in using them for various UAV applications. [18] achieves a high gain by using phase shifters and amplifiers, but has limitations in that it is expensive and has a narrow elevation angle. [19] implements a wide elevation angle using 16 antennas and MCU. However, it has a complex structure and a large size. Previous studies ( [21]- [22]) have the disadvantage of having a wide elevation angle only with respect to the x-axis. The antenna [23], which was proposed to improve [21]- [22], has a uniform elevation angle and four multi-beams, thereby eliminating the dead zone. However, it appears that the gain and elevation angle need to be improved. An antenna that can secure the limitations of [13]- [23] is proposed in this paper. The proposed array has a relatively small size, high gain, wide HPBW, and simple structure. In addition, the phase shifter, amplifier, and MCU are not used, which greatly reduces the cost. Therefore, the proposed array has advantages of high gain, wide elevation angle, wide impedance bandwidth, and low cost compared to [13]- [23]. Also, compared with the previously proposed antenna [3], the beam width and gain in the zx-plane and yz-plane were 17 • (9.4%), respectively. 40 • (66.7%) and 1.2 dBi(17.9%) were improved.

V. CONCLUSION
In this paper, we propose printed dipole antenna array with reconfigurable feeding network for wide elevation angle of U2X communications. The proposed array forms a 3×3 beam using broadside beams and end-fire beams based on the x-axis and y-axis using a reconfigurable feeding network.
And the proposed array uses parasitic elements to improve the elevation angle. In addition, in order to improve the gain by reducing backside radiation and to prevent the antenna from being affected by the UAV and the reconfigurable feeding network, the back side of the reconfigurable feeding network was used as a grounded metal plate. Therefore, the proposed array has high gain and wide elevation angle as well as low height and symmetrical structure, so it is expected to have high utility in applications using UAVs.