Low-Profile Wideband Solar-Cell-Integrated Circularly Polarized CubeSat Antenna for the Internet of Space Things

This paper proposes a low-profile wideband circularly polarized (CP) antenna using solar cell patches as radiation elements and a sequentially rotated feeding network for CubeSat applications. To realize a wide axial ratio (AR) bandwidth with a compact size, a sequentially rotated feeding network was designed by modifying a quadrature hybrid coupler and a rat-race coupler that has a small change in phase difference even when the frequency changes. A wideband CP patch array antenna was designed by combining a C-shaped slot-coupled solar cell patch in conjunction with a novel feeding network. The overall size of the proposed CP CubeSat antenna is 100 <inline-formula> <tex-math notation="LaTeX">$\times 100\times7.2$ </tex-math></inline-formula> mm<sup>3</sup> (<inline-formula> <tex-math notation="LaTeX">$0.83\,\,\lambda _{\mathrm {o}} \times 0.83\,\,\lambda _{\mathrm {o}} \times 0.06\,\,\lambda _{\mathrm {o}}$ </tex-math></inline-formula> at 2.5 GHz). Solar cells occupy 79% of the antenna area, enabling efficient energy harvesting. The −10 dB impedance bandwidth is 1.98–3.0 GHz, which is a fractional bandwidth of approximately 41.0%. The 3-dB AR and 3-dB gain bandwidths are 1.98–3.0 GHz (41.0%) and 1.82–2.98 GHz (46.6%), respectively. The proposed CP solar patch array antenna demonstrates a constant radiation pattern within the −10 dB impedance bandwidth. The proposed CubeSat antenna is suitable for use in an Internet of Space Things (IoST) autonomous communication system.


I. INTRODUCTION
The Internet of Things (IoT) refers to things or people connected through a network, so they can effectively exchange information through an embedded communication system. The IoT is recognized as a key driving force for 5G/6G wireless communication due to its ubiquitous characteristics, which can operate anytime and anywhere, as well as its application-oriented operation, which can connect numerous physical points [1]. It is expected that more than 70 billion devices will be connected by 2025, which poses many challenges to the practical realization of the IoT. Therefore, countries around the world are seeking the evolution of a hyper-connected society based on IoT technology [1]- [3].
The associate editor coordinating the review of this manuscript and approving it for publication was Tutku Karacolak .
In line with this trend, various types of IoT devices are being developed. With the rapid increase of wide-area IoT and short-range IoT, the number of devices connected to the network is expected to increase explosively in the future. Currently, connectivity for IoT solutions is realized through a variety of terrestrial networks, including but not limited to wireless personal area networks and low-power wide-area networks. However, there are still many areas where it is difficult to provide coverage due to financial problems, complex environments, and rugged terrain. To this end, the concept of the Internet of Space, which utilizes Low Earth Orbit satellites as a possible solution, has been proposed. The Internet of Space Things (IoST) is a system that enables mobile communication anywhere in the world using low-orbit satellites located at altitudes of 160 to 2,000 km, as shown in Fig. 1 [4]. The rapidly changing IoT environment makes it difficult to VOLUME 10, 2022 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ implement global satellite communication based on existing medium and large satellites requiring a long development schedule and high development costs. Considering the shortcomings of existing medium and large satellites, a new type of small satellite called CubeSat is actively being developed to implement a new IoST [5], [6].
A CubeSat is a cubic-shaped standardized small satellite weighing less than 1.33 kg with dimensions of 10 cm × 10 cm × 10 cm, defined as 1 U [7], [8]. A CubeSat can be used alone or in groups of multiple units. CubeSats have the advantage of greatly reducing costs because commercial off-theshelf components are extensively used, and the development and distribution cycle is very short because standardized devices are implemented. In addition, CubeSat orbits can respond much more actively to satellite disturbances because the number of CubeSats used in orbit is large [7]. A CubeSat is very small, and many components must share limited space. Consequently, the antenna of CubeSat must be efficiently utilized within a small area. A solar cell is used as the power source for CubeSat; many studies on solarcell-integrated antennas are being conducted to efficiently utilize the restricted surface area of CubeSat [9]. A solar cell-integrated antenna using a transparent electrode was proposed [10]- [12]. An antenna using a transparent electrode, such as indium tin oxide (ITO), has a simple design. However, ITO is expensive and results in low antenna efficiency due to the low conductivity of thin conductors. An antenna structure combining a slot with a solar cell was proposed [13]- [15]. The structure in which the slots are arranged between the gaps of the solar cell array is also simple in design. However, it is difficult to change the antenna's structure for optimal performance [13]. An antenna with a slot inserted by cutting the solar cell's structure can be designed into various shapes. Therefore, it is easy to obtain the desired antenna characteristics. However, the amount of output current collected by the solar cell decreases because the area of the solar cell is reduced by the presence of the slot [14]. Antennas using solar cells as metasurfaces have been proposed [16], [17]. A metasurface antenna can realize high gains and wide bandwidths. However, as the operating frequency increases, the size of the solar cell used as the metasurface unit cell becomes very small. Therefore, connecting each solar cell for direct current collection is complicated. An antenna using solar cells as a patch, proposed in [18]- [20], is simple in design and can achieve high gains and wide bandwidths. In addition, the efficiency of the solar cell is not reduced, and the solar cell's energy harvest is easy. However, to design a solar cell patch antenna with a wide bandwidth, the overall size of the antenna must be increased.
Circularly polarized (CP) antennas are widely used as satellite antennas because they are less affected by the installation direction of the transmitting and receiving antennas, the multipath effect, and the Faraday effect in the ionosphere [21]- [26]. Many kinds of research have investigated the design of CP antennas [27]- [39]. The sequentially rotated feeding network is widely used as a feeding structure for CP antennas because of its small size and simple structure [27]- [33]. However, the conventional sequentially rotated feeding network has low isolation between output ports. This narrows the axial ratio (AR) bandwidth of the antenna. The phase difference between the output ports changes as the frequency varies, thus changing the antenna's radiation pattern. To realize a wide AR bandwidth and a constant radiation pattern, a sequentially rotated feeding network using a phase circuit, such as a Schiffman phase shifter, has been presented [34]- [38]. The sequentially rotated feeding network using the Schiffman phase-shifter displays a small phase difference between the output ports over a wide frequency range, and the AR bandwidth is wide due to the high isolation between output ports. However, the Schiffman phase-shifter requires a Wilkinson power divider to improve the isolation at the expense of power loss. In addition, the Schiffman phase shifter consists of long transmission lines, which causes the Schiffman phase shifter to be very large. This, in turn, limits its application to a small antenna feeding structure. In [39], a sequentially rotated feeding network using a branch-line coupler was put forward. The branchline coupler has high isolation and little change in the phase difference between output ports over a wide frequency range. In addition, the branch-line coupler can be easily miniaturized by bending the transmission lines of the coupler. Thus, it is easy to use in the antenna's feeding structure.
In this paper, we propose a CP antenna with low-profile, high-gain, wideband characteristics using a solar cell patch as the radiating element and a sequentially rotated feeding network implemented with a modified branch-line coupler for use in CubeSat. The proposed antenna has a wide bandwidth and little variation in its radiation pattern. In addition,   by using a solar cell as the radiating element and placing it on CubeSat's surface, the efficiency of the solar cell is not reduced, and solar cell energy harvesting is simple because few solar cells are used.

II. WIDEBAND PATCH ANTENNA
An aperture-coupled patch antenna has a wider impedance bandwidth than a microstrip line-fed patch antenna. Energy harvesting is simple when solar cells are used as a patch because the patch and feeding structure are not directly connected. Due to these advantages, a wideband patch antenna was designed using the aperture-coupled method.

A. C-SHAPE SLOT-COUPLED SOLAR CELL INTEGRATED PATCH ANTENNA
A single patch antenna with low-profile and broadband characteristics was first designed. The substrate used for the antenna design was a ROGERS RO4003C (ε r = 3.38, tanδ = 0.0027). The thickness of the substrate was 0.508 mm. Fig. 2 shows the structure of the single-patch antenna. The antenna consists of substrates 1 and 2, a reflector to reduce back radiation, and foams 1 and 2 to support the antenna structure. A rectangular silicon solar cell patch was placed on top of substrate 1. Foam 1 is inserted between substrates 1 and 2 to support both substrates. The ground plane and slot are printed on the top side of substrate 2. Instead of the conventional straight, narrow slot, a C-shaped wide slot with good impedance matching performance was used [40].  The microstrip line and the tuning stub of the antenna are printed on the bottom side of substrate 2. A thin microstrip line with a width W fa is a transmission line connecting a single patch antenna and a feeding network, and a wide strip line with a width W st is a tuning stub for a single patch antenna. The tuning stub with a wide line width has less impedance change in a wide frequency range; thus, a wider impedance bandwidth can be realized [41]. The impedance bandwidth was further widened by generating modes with varying polarization directions by adjusting the positions of the slots and feed lines [42]. A reflector is placed at the bottom of the antenna to reduce back radiation, and foam 2 is inserted between substrate 2 and the reflector to support substrate 2. The design and analysis of the antenna were performed using the ANSYS High-Frequency Structure Simulator (HFSS). The design parameters of the single patch A 2 × 2 CP array antenna using the designed single patch antenna was implemented, and its characteristics were demonstrated. After arranging the four patch antennas in a given 10 cm × 10 cm space, a phase difference of 90 • was set for each antenna to achieve circular polarization. When the array antenna was designed using four single-patch antennas, the characteristics of the single antenna changed. ANSYS HFSS was used to optimize the array antenna. Fig. 6 depicts an optimized 2 × 2 CP array antenna. The design parameters of the optimized array antenna are as follows: W = 100 mm,   Fig. 7 shows the characteristics of the array antenna. Fig. 7(a) demonstrates the reflection coefficient of the antenna. The −10 dB impedance bandwidth of the array antenna is 1.89-2.89 GHz, which is a fractional bandwidth of 41.8%. The −10 dB impedance bandwidth of the array antenna is wider than that of a single antenna. Fig. 7(b) shows the antenna gain and AR. The single-patch antenna has a problem in which the cross-polarization level increases and the polarization direction changes as the frequency increases. However, the gain and AR are not affected when the single patch antennas are configured as a CP array antenna.

III. SEQUENTIALLY ROTATED FEEDING NETWORK
To realize a CP antenna with a wide AR bandwidth and little change in the radiation pattern even with frequency changes, a sequentially rotated feeding network consisting of modified branch-line couplers with a small size and wideband characteristics is used. A. MODIFIED QUADRATURE HYBRID COUPLER Fig. 8 shows a modified quadrature hybrid coupler designed with 100-input and output ports. By modifying the quadrature hybrid coupler, the area occupied by the transmission line is greatly reduced [39]. Fig. 8(a) is a diagram of a conventional quadrature hybrid coupler with a large circuit area, limiting its use as an antenna-feeding structure. Fig. 8(b) is a diagram of a quadrature hybrid coupler in which the empty space inside the circuit is reduced by folding the transmission line with 0.707 Z o characteristic impedance. Fig. 8(c) is a schematic of the modified quadrature hybrid coupler with reduced spacing achieved by bending the conventional quadrature hybrid coupler 90 • . The proposed modified quadrature hybrid coupler has a much smaller circuit area than the conventional coupler. Fig. 8(d) illustrates a feeding structure with two input ports and four output ports designed using two modified quadrature hybrid couplers. Load resistors are connected to the isolated port of each coupler.
When power with the same phase is supplied to two input ports, the phase of the output power of output ports 1 through 4 is 0 • , 90 • , 0 • , and 90 • , respectively. The design parameters of the modified quadrature hybrid coupler are as follows: L hc = 19.2 mm, W c1 = 1.15 mm, W c2 = 0.75 mm, g c = 0.25 mm, W cf = 0.3 mm, and L c = 3.0 mm. Fig. 9 demonstrates the characteristics of the couplers described in Fig. 8. Fig. 9(a) shows the reflection coefficient of each coupler. The −10 dB impedance bandwidth of the proposed coupler is slightly wider than that of the conventional coupler; the impedance matching is very good. Fig. 9(b) depicts the difference in output power between the output ports of the couplers. The frequency range where the power difference between the output ports of the proposed coupler is less than 2 dB is wider than that of the conventional coupler. Fig. 9(c) shows the phase difference between the output powers of the couplers. The proposed coupler shows little phase difference change over a wide frequency range. By modifying the coupler, the area that the circuit occupied is greatly reduced, and the performance is still good.

B. MODIFIED RAT-RACE COUPLER
An additional phase difference of 180 • is required to implement circular polarization using the feeding structure in Fig. 8(d). A 180 • coupler with a modified rat-race coupler was designed to implement an additional phase difference. Fig. 10(a) shows a conventional rat-race coupler. The ratrace coupler can be used as a power divider with a phase difference of 180 • over a wide frequency range. However, the area occupied by the circuit is very large, and it is not suitable for use as a feeding structure for miniaturized antennas. Fig. 10(b) displays the structure in which the isolated port is removed from the conventional rat-race coupler. In the rat-race coupler, the power input to port 1 is not transmitted to port 4. Therefore, even if port 4 is removed, the rat-race coupler can be used as a power divider with a phase difference of 180 • . In the proposed coupler, port 4 was removed, and the transmission line width was changed so the coupler had a 50-input port and 100-output ports. The coupler consists of a pair of transmission lines with a characteristic impedance of 100 with a length of λ/4 and a pair of transmission lines with a characteristic impedance of 70.7 with a length of λ/2. Fig. 10(c) is a modified rat-race coupler in which the size of the rat-race coupler depicted in Fig. 10(b) is reduced by folding the transmission line. In the proposed coupler, the circuit area is greatly reduced to 1/6 of the size compared with the conventional rat-race coupler. The design parameters of the modified rat-race coupler are as follows: L r1 = 14.5 mm, L r2 = 12.0 mm, W r1 = 0.6 mm, g r1 = 0.6 mm, g r2 = 0.9 mm, L r3 = 6.0 mm, L r4 = 4.4 mm, L r5 = 3.05 mm, W r2 = 0.3 mm, g r3 = 0.4 mm, and d r = 2.45 mm.   Fig. 11(a) shows the reflection coefficient of the couplers depicted in Fig. 10, and Fig. 11(b) shows the output power difference of the output ports of each coupler. Fig. 11(c) demonstrates the phase difference between the output ports of each coupler.
The proposed coupler has a slightly narrower −10 dB impedance bandwidth compared to the conventional rat-race coupler. However, the output power and phase difference between the output ports are smaller than the conventional rat-race coupler within the impedance bandwidth.

C. PROPOSED SEQUENTIALLY ROTATED FEEDING NETWORK
By combining the previously designed modified quadrature hybrid coupler and the modified rat-race coupler, a novel sequentially rotated feeding network with one input port and four output ports was designed. Fig. 12 reveals the conventional and proposed sequentially rotated feeding networks. The area occupied by the proposed sequentially rotated feeding network is 19.9 × 19.9 mm 2 , which is smaller than the area of the conventional sequentially rotated feeding network of 21 × 21 mm 2 . Fig. 13 shows the amplitude characteristics of the S-parameters of the conventional and proposed VOLUME 10, 2022 feeding networks. Fig. 13(a) portrays the S-parameter of the conventional feeding network, and Fig. 13(b) shows the S-parameter of the proposed feeding network. The conventional sequentially rotated feeding network has a low reflection coefficient over a very wide frequency range, and the transmission coefficient of each output port is constant. The proposed sequentially rotated feeding network has good performance characteristics within 2.00-3.00 GHz. Fig. 14 shows the phase characteristics of the S-parameters of the conventional and proposed feeding networks. Fig. 14(a) shows the phase difference between the output ports of the conventional feeding network, and Fig. 14(b) demonstrates the phase difference between the output ports of the proposed feeding network. Because the conventional feeding network implements the phase difference based on the length of the transmission line, the electrical length of each transmission line changes at a frequency outside the center frequency, thereby changing the phase difference between the output ports. Conversely, because the proposed feeding network generates a constant phase difference over a wide frequency range, the phase difference between the output ports is stable, even at a frequency outside the center frequency.

IV. FABRICATION AND MEASUREMENT
We designed the 2 × 2 solar cell patch-integrated CP antenna by combining the proposed feeding network with the antenna designed in Section 2. Fig. 15 presents the geometry of this antenna. On the top surface of the antenna, there are four solar cell patches and a substrate supporting the solar cell, and the performance of the solar cells determines the DC performance of the antenna. Four RF decouplers, each consisting of a pair of inductors, were added for solar cell energy harvesting. One part of the RF decoupler is connected to the bottom contact of the solar cell, and the other part is connected to the grid. To implement DC energy harvesting, a pair of metal wires were connected to the end of the RF decoupler, and an inductor was used as the RF choke to  prevent the RF signal from leaking along the wire. The middle substrate has a ground plane, including a C-shaped slot on  top, and the feeding network is printed on the bottom. The reflector is located at the bottom of the antenna. A coaxial cable was used to feed the antenna. The inner conductor of the coaxial cable is connected to the input port of the feeding network, and the outer conductor is connected to the reflector. Metal pins connect the ground plane with the reflector surrounding the feeding network, so the reflector has the same electric potential as the ground plane. The designed antenna was fabricated, and its performance was measured. Fig. 16 shows the fabricated antenna. A Rohde & Schwarz ZVA 67 vector network analyzer was used to measure the antenna reflection coefficient, and the antenna radiation pattern was measured in the MTG anechoic chamber. Fig. 17 displays the simulated and measured antenna characteristics. The measured result of the −10 dB impedance bandwidth of the antenna is 1.98-3.0 GHz, a fractional bandwidth of 41%. The 3-dB gain bandwidth and AR bandwidth are 1.82-2.98 GHz and 1.98-3.0 GHz, respectively, which are fractional bandwidths of 46.6% and 41.0%, respectively. The measured results are very similar to the simulation results. Fig. 18 shows the antenna radiation patterns at 2.2 GHz, 2.5 GHz, and 2.8 GHz. The simulated and measured antenna gains at 2.2 GHz are 8.5 dBic and 8.7 dBic, respectively. The simulated and measured gains at 2.5 GHz are 8.4 dBic and 8.1 dBic, respectively, and the simulated and measured gains at 2.8 GHz are the same at 8.8 dBic. In the proposed antenna, the radiation patterns of the antenna do not change significantly, even when the frequency changes, and the radiation patterns in the xz-and yz-planes are very symmetrical. The simulation and measurement results of the antenna are summarized in Table 1.

V. EFFECTS OF THE SOLARCELL ON ANTENNA PERFORMANCE
The electrical properties of a solar cell are greatly affected by the amount of incident light. When a solar cell is exposed to light, the electrical conductivity of the solar cell increases significantly [43]. A change in the electrical conductivity of a solar cell may affect the characteristics of the antenna. Fig. 19 shows the measured reflection coefficient of the antenna with different light intensity values. To validate the effect of light intensity, the reflection coefficient was measured under light intensities of 0.0, 250.4, 499.5, 749.0, and 999.4 W/m 2 (with a 100 W halogen lamp). As shown in Fig. 19, the measured reflection coefficient differs slightly depending on the light intensity values. However, the shape of the curves in the figure is almost identical, indicating that the light intensity has little effect on the reflection coefficient of the antenna.

VI. COMPARISON
The proposed antenna was compared with other solar cellintegrated antennas. The solar cell-integrated CP antenna presented in [10] consists of a transparent electrode radiation element and a conventional sequentially rotated feed network. The electrical size of the antenna is 3.33 × 3.33 × 0.66 λ 3 o , and the −10 dB impedance bandwidth is over 30%. The 3-dB gain bandwidth is approximately 16.3%, and the AR bandwidth is approximately 20%. The peak gain is 17 dBic. However, the gain and AR bandwidths are narrow, and the antenna is very large. The solar cell-integrated microstrip slot array antenna presented in [13] used a slot antenna as a radiation element and implemented CP using a sequentially rotated feed network. The electrical size of the antenna is 2.71 × 2.71 × 0.001 λ 3 o . The −10 dB impedance, 3-dB gain, and 3-dB AR bandwidths are over 17.3%. However, the peak gain is only 6.6 dBic, a very low gain compared to its large size. In [17], a CP patch antenna with a solar cell metasurface is presented. The electrical size of the antenna is 0.87 × 0.87 × 0.076 λ 3 o , and the −10 dB impedance bandwidth is 19.1%. The 3-dB gain bandwidth is approximately 33%, and the AR bandwidth is about 16.1%. The peak gain is 8.8 dBic.
Although the gain was high compared to the size, the metasurface composed of solar cells requires many inductors for energy harvesting and is difficult to implement. A CP antenna using a solar cell patch antenna and a quadrature hybrid coupler is presented in [18]. The electrical size of the antenna is 0.67 × 1.14 × 0.016 λ 3 o , and the −10 dB impedance bandwidth is 12.5%. The peak gain of this antenna is almost 6 dBic. However, the 3-dB gain and AR bandwidths are very narrow. A low-profile solar cell-integrated patch antenna is proposed in [18]. The electrical size of the antenna is 1.6 × 1.2 × 0.024 λ 3 o . The −10 dB impedance bandwidth and 3-dB gain bandwidths are 15.5% and 21.3%, respectively. This antenna has a peak gain of 9.4 dBi, which is low compared to the size of the antenna. It also has a low form factor of 55.4%. A solar cell patch antenna with two aperture-coupled feeds is presented in [20]. The electrical size of the antenna is 1.31 × 1.31 × 0.060 λ 3 o . The −10 dB impedance bandwidth and 3-dB gain bandwidth are very narrow, at 6.8% and 8.4%, respectively. This antenna has a high peak gain of 10.8 dBi, with a low form factor of 45%.
The antenna proposed in this paper has an electrical size of 0.83 × 0.83 × 0.06 λ 3 o , a −10 dB impedance bandwidth of 41.0%, and a 3-dB gain and AR bandwidths of 46.6% and 41.0%, respectively. In addition, the solar cell occupies 79% of the antenna area. Thus, it has a higher form factor compared with that of conventional CP solar cell-integrated antennas ( [13]: 73.5%, [17]: 56.3%, [18]: 79.0%, [19]: 55.4%, [20]: 45.0%). Here, the form factor is defined as the ratio of the total area utilized by the solar cell for energy harvesting to the given surface area of the antenna. Hence, it has superior performance characteristics compared to conventional solar cell-integrated antennas. In addition, there is no problem with the beam tilt due to changes in frequency. The performance characteristics of the conventional antennas and the proposed antenna are summarized in Table 2.

VII. CONCLUSION
In this paper, we propose a 2 × 2 sequentially rotated CP array antenna that combines a solar cell patch antenna having wideband characteristics with a novel sequentially rotated feeding network that boasts a stable phase difference over a wide frequency range. The proposed antenna has wide gain and AR bandwidths, and the radiation pattern is stable even when the frequency changes. The electrical size of the proposed antenna at the center frequency of 2.5 GHz is 0.83 × 0.83 × 0.06 λ 3 o . The −10 dB impedance bandwidth is 1.98-3.00 GHz. The 3-dB gain bandwidth and AR bandwidth are 1.82-2.98 GHz and 1.98-3.0 GHz, respectively. Moreover, the proposed antenna has a high form factor of 79%. Due to these advantages, the proposed antenna is suitable not only for CubeSats but also for other satellite applications. Therefore, it is useful for implementing an IoST autonomous communication system.