This paper presents the design of an antenna array utilizing stacked multiloop structures to produce the electromagnetic radiations with dual circular polarizations and multiple directional beams. In particular, with a proper design of isolations, each combination of these polarizations and beams can be operated in a relatively independent fashion to increase the system capacity by using a single antenna set. The antenna structure was validated by numerical simulations and experimental measurements, which are presented to exhibit the characteristics of radiation.
The development of antenna technologies to enhance the capacity of communication system is increasingly important because of the limited available frequency bandwidths. The current trend is to fully explore the potentials of diversities [Winters et al., 1994; Goddard and Cherry, 1984] in a way that the system capacity can be increased by using the same set of frequencies. The well-known MIMO (multiple inputs and multiple outputs) [Gesbert, 2003] technologies appear to a successful one that utilizes the space diversity [Goldhirsh, 1982] of antennas at both the transmitting and receiving sides. The transmitting and receiving signals are uncorrelated to increase the system capacity.
Based on this logic, similar scenarios have been developed and attempted to extend the applicable scopes of communication enhancement with increased capacity. Two potential candidates are the polarization and angular beam diversities, where dual polarizations and multiple beams are created. The application scenario is described as follows: When the antenna is properly designed with low cross-polarization level of radiation, the two orthogonal polarizations (either the two linear polarizations or the two circular polarizations) can be used independently to double the system capacity. Similarly, the angular beam diversity can be possibly used when the far field patterns of antenna radiation are properly formed; the orthogonal property of beams will exist as described in the conventional MIMO technology. For example, the beam peak points to the direction near either the null directions of the other beams or the directions with very low sidelobe levels. When these beam patterns are properly overlapped, the orthogonality between beams can be possible to provide a class of applications. It noted that the applications of switching beams can be viewed as a format of angular beam diversity.
A successful application of both polarization and angular beam diversities to increase the system capacity is the application of digital television (DTV) signal receptions from multiple satellites [Sakakibara et al., 1999; Arapoglou et al., 2010]. The antenna system based on reflector antenna configurations produces several beams pointing to different directions of satellites [Ng Mou Kehn and Shafai, 2009]. Both polarizations are utilized to receive signals from different channels of communications and double the signal capacity. The implementation of conventional reflector antennas is relatively simple by using several feeds offset with different distances from the focus point of the reflector, each beam per feed. Each feed, realized by the horn antennas with waveguides as the input ports, produces radiations with two polarizations excited by two independent orthogonal ports feeding the waveguides. These two ports can then be operated independently to increase the capacity. However, the reality is that they have to use different antennas (different feeds in this case to feed the reflector as a reflecting structure) to produce multiple beams.
Similar efforts have been attempted in the design of phase array antenna because of their possibility of keeping low profiles. A typical effort is shown by Guinvarc'h and Haupt , where an array of spiral antennas are designed for dual polarization operations. In this work, the dual polarization radiations are created from two subgroups of interleaved array elements selected from the original array according a selection criterion or optimization algorithm. The antenna elements of each subgroup are designed to radiate electromagnetic fields of one polarization. This strategy also uses two different types of antennas, one per polarization. As a result, the antenna design will become sophisticated because of the complicated beam forming networks. It also has disadvantages of increasing the physical size of antenna array and may result in irregularly high grating or sidelobe levels, since the interelement spacing will be larger than a half wavelength.
In this paper, we attempt to develop an antenna structure that may simultaneously provide the dual polarization and angular beam diversities of multiple directional beams by using a single-antenna set. The application scenario can be the point-to-multipoint communications such as using a single antenna for multiple-satellite DTV receptions [Sakakibara et al., 1999; Arapoglou et al., 2010]. It can be also used for the switching beam applications.
To demonstrate the concept, we present the design of a 2 × 4 antenna array to produce a radiation with dual circular polarizations and three-directional beams, whose combinations will provide six modes of independent operations. For simplification of antenna prototyping and performance measurements, we consider a frequency band at 2.45 GHz, which is popularly used in wireless communications. This frequency is sufficiently low and may reduce unexpected errors during the prototyping. Both antenna structure and characteristic examination are presented to validate the design concept in the following sections. Furthermore, the applications of multibeams and switching beams are both examined in terms of numerical simulations and experimental measurements. This paper is presented in the following format: Section 2 describes the antenna design. Results obtained by numerical simulation and experimental measurement are presented in section 3. A short conclusion is given in section 4.
2 Antenna Design
The design example considers eight antenna elements to form a fundamental 2 × 4 antenna array as shown in Figure 1, which further forms two subgroups to demonstrate the design concept. The commercial software HFSS developed by Ansoft was employed to design the antenna. The frequency band is at 2.4–2.5 GHz, with the wavelength by λ = 120 ~ 25 mm. The period of the array elements is selected by approximately λ / 2 or 60 mm to avoid the occurrence of grating lobes. This array is accommodated within a space of 240 × 120 × 30 mm3 over an FR4 dielectric substrate of 1 mm in thickness. The top surface of FR4 substrate is a ground plane. The feeding networks are designed under this substrate to reduce their radiations and coupling with the antenna elements. The goal is to design the antenna array to provide a radiation with simultaneously existing three-directional beams and dual circular polarizations (both right- and left-handed circular polarizations, (RHCP and LHCP, respectively)). Thus, the design will provide six combinations (each combination is referred as an operational mode, hereafter) of radiations that will exist simultaneously to provide independent operations of six ports.
2.1 Description of the Elemental Antenna Structure
The elemental antenna consists of seven stacked metal loop wires [Harrington and Mautz, 1968; An and Smith, 1982] with an equal vertical interval, Δh, as illustrated by the “0”–“6” indexed wires in Figure 2. In this case, Δh = 1 mm, and the diameter of wires is 0.5 mm. The 0 wire is not excited and serves as a coupling wire to provide equal coupling effects for the other wires. This wire has an effect of enhancing the gains of antenna radiation by 1.5–2 dB from our numerical simulation experience. The other six wires are used to excite the six modes of operations with each excited independently per mode. These loop wires are horizontally oriented and are placed at approximately λ / 4 above the ground plane, because the horizontal current flow on the loop wires will result in an image current with an opposite sign in the strength. A λ / 4 separation distance will create an equivalent separation distance of λ / 2 between the real loop and the imaged one and thus results in a positive contribution to the antenna radiation in the boresite direction. The actual value is determined by numerical simulation. The radius of these loops have an identical value, which is adjusted to provide a good behavior of reflection coefficients and directional radiations pointing upward. Using the criterion of a helix antenna design to produce an axial mode radiation in Kraus  and Vaughan and Andersen , the length of wires is approximately between 3 λ / 4 and 4 λ / 3. In this case, the diameter of loop is selected as 43.5 mm and makes the length of wires by 136.59 mm. The design procedure can be found in Balanis . The details of design and the characteristics of helical beam antennas are referred to Kraus [1949, 1977], Kraus and Williamson , and Storer  and will not be repeated for brevity.
Starting from the feeding for a RHCP at wire “1” (or the first mode), the excitations for RHCP and LHCP are fed alternatively from wire 1 to 6 (i.e., modes 1–6). Each pair of loop wires, consisting of RHCP and LHCP modes, will be used as the equivalent antenna element for a directional beam. Thus, three pairs can be used to produce three-directional beams with dual circular polarizations as mentioned in section 1 and in the assignment of six modes in Table 1. Each mode will be in charge of providing the radiation of a particular directional beam with a polarization. The RHCP and LHCP radiations may coexist for a particular directional beam with respect to two different ports. These six modes may be operated simultaneously and independently when their corresponding components of cross polarization are sufficiently low. The gap widths, gi(i = 1 ~ 6), between adjacent vertical feeding wire segments as shown in Figure 2 are 0.5 mm, while the vertical lengths of the feeding wire segments, hi(i = 0 ~ 6), are 29, 28, 27, 26, 25, 24, and 23 mm, respectively.
Table 1. The Array Excitation Phases (Unit: Degrees)
The excitation of circularly polarized radiation considers the current flowing direction along the loop wire. In particular, one end of the wire loop is excited, while the other end remains open. Figure 3 shows the different wire structures exciting the circularly polarized radiations, in which the structures to excite both LHCP and RHCP are shown. Due to the limited space, these structures do not have sufficient lengths for current flows to radiate fields with good circular polarizations in comparison with the conventional helix antenna in its axial mode. A compensation of radiation to cancel their cross-polarized components is described in the following subsection illustrating the radiations of LHCP and RHCP fields.
2.2 Arrangement of Antenna Orientation for Circular Polarization Improvement
To improve the axial ratio (AR) or reduce the cross-polarized component of radiation, every four adjacent antenna elements were selected to form a subgroup as illustrated in Figure 1. A sequential rotation approach [Hall et al., 1989; Kraft, 1996] has been applied to rotate the orientations of each element by 90° with respect to the corresponding adjacent element. This sequential rotation approach improves the AR for both the cases of RHCP and LHCP without the needs to specially consider their polarization orientations, where both cases will be operated simultaneously in our antenna design. This assurance can be observed from their relative angular shifts in the clockwise and counterclockwise directions for LHCP and RHCP, respectively.
2.3 Antenna Feeding Network
The antenna feeding networks are designed in an independent fashion for each mode of operation. Thus, six feeding networks are designed independently and are operated simultaneously. The coexistence of these feeding networks can be realized by using the low-temperature cofired ceramics technology [Carchon et al., 2000; Uhlig et al., 2005] and will be reported in the future phase of this work. For brevity, only the separated feeding networks for modes 1–3 are shown in Figures 4a–4c, respectively, where Figures 4a and 4b show the feeding networks for the antenna radiations with RHCP and LHCP, respectively, in the boresite direction (θ = 0°), while Figure 4c shows that for RHCP of an offset beam pointing to θ = 30°.
In the design of each feeding network, Wilkinson's equal power dividers [Wilkinson, 1960; Carchon et al., 2000] are used to assure an equal power division for the eight ports to feed the array elements. The required phases of each element's excitations are obtained by the conjugated field matching technique [Ling et al., 1986] at the beam directions, which automatically includes the 90° phase difference between each pair of angularly adjacent elements [Teshirogi et al., 1985]. These phases are created by using different lengths of transmission lines. This technique is applied for the feeding network of each mode individually, where the phases are shown in Table 1 for the entire six modes. In particular, ports 1–4 and 5–8 are associated with the two different subgroups in Figure 1. Based on these simulated phases in Table 1, the feeding networks in Figure 4 were designed. The required phases are implemented by varying the lengths of transmission lines beyond the third stage of Wilkinson's equal power dividers. This structure of feeding network relatively simplifies the overall efforts, because the RF circuits prior to the third stage of Wilkinson's power dividers can be identical for all modes. This design strategy will assure an equal power division and a phase reference for all ports because of the symmetry of the structure. In our simulation of design, the errors in the phases created by the feeding networks are less than 2° and can be ignored. The measurement results over the prototype show similar errors and verify the validity of phase error ignorance.
3 Simulation and Experimental Results
3.1 Results of Numerical Simulations
Numerical simulations using HFSS have been performed. Table 2 summarizes the antenna performances using the ideal excitation phases in Table 1. The simulations were performed on a personal computer with a dual-core AMD 2.2 GHz CPU and 16 Gb RAM, which requires 2 h to complete the simulation of eight antenna elements with a single input port. The reflection coefficients shown in Table 2 are selected from the eight ports to provide a rough idea for brevity without losing the generality. It is first observed that the overall energy efficiency is better than 90%. It drops for the modes with wires located at the lower part of the stacked structure, which can be explained by the reduction of separation distance between the wire loop and the ground plane to become less than λ / 4. The axial ratios are less than mostly 2 dB and exhibit good characteristics in the radiation. The gains in this case are high than 10.73 dB. Table 3 shows the radiation performances of the array when the feeding networks are integrated into the simulation. It is observed that the efficiencies drop by roughly 40–50%, which is caused by the energy loss in the feeding networks implemented by a FR4 substrate (2–3 dB in our examination). However, the reflection coefficients and axial ratios are generally improved in most of the cases. The radiation patterns for RHCP components of modes 1, 3, and 5 at 2.45 GHz are shown in Figure 5. The cross-polarization levels can be observed from the axial ratios summarized in Tables 2 and 3 and will not be repeated for brevity. However, it can be observed from the overlapped patterns in Figure 5 that the peaks of the offset beams are located in the vicinity of first null region of the boresite beam. It validates the design concept of multibeam operations, which can be used for point-to-multipoint communications. Also, the peaks of these two offset beams are located in the sidelobe regions of each other. Thus, the control of sidelobe levels is very important in this application. One may further employ an optimization technique to synthesize the radiation patterns to produce null fields in the beam directions. It will be investigated in the future phase and does not affect the design philosophy presented in this work to realize the antenna structure.
Table 2. Radiation Performance Obtained by Using Ideal Excitation Phases in the Numerical Simulation
Table 3. Radiation Performance Obtained by Using Feeding Networks in Figure 3 in the Numerical Simulation
3.2 Results of Experimental Measurements
Experimental measurements have been performed to validate the antenna design. The measurements were performed in a 3-D far-field range (AMS-8500 3-D antenna measurement system by ETS-Lindgren) [ETS-Lindgren, 2013] with a valid frequency range of 0.7–6 GHz band and size of 7.32 × 3.66 × 3.66 m3, which satisfies the Cellular Telecommunications Industry Association (CTIA) Over-the-Air (OTA) test standard. The prototype of the antenna array is shown in Figure 6a with an element shown in Figure 6b, where the loop wires are fixed on a Styrofoam material. The prototype of feeding network for mode 1 is shown in Figure 6c. For brevity, we show the results of mode 1 operation and, in addition, also attempt to investigate the characteristics of radiation when a beam is offset for the applications of switching beams. Thus, a feeding network for an offset beam pointing to 10° was also prototyped, which was fed to mode 4 and radiates LHCP fields. The results of the other modes have similar characteristics and will not be repeated for brevity. Figure 7 shows the reflection coefficients of array in these two cases. It can be observed that the valid frequency band is within the desired band of 2.4–2.5 GHz. At 2.45 GHz, the reflection coefficients are −13.41 and −13.45 dB, respectively. Also, the radiation characteristics are shown in Figure 8 for these two cases, where the peak gains, radiation efficiencies, and axial ratio are shown. In particular, at mode 1, the results agree with those shown in Table 3 with slightly higher gains. The measured transmission coefficients (S21) for the output ports of feeding beam forming networks used to create the boresite and offset beams, with the port order as labeled in Figure 2, are shown in Table 4 for reference, which are roughly 11.2–11.6 dB and representing 2.2–2.6 dB power loss in the beam forming network. This agrees with the simulation results. Thus, the efficiency will be between 40 and 50% as expected in the numerical simulation. Also, when the beam direction is offset from the antenna boresite, the radiation characteristics are not significantly affected as one has observed in Figures 7 and 8.
Table 4. The Transmission Coefficients Between the Input Port (Port 0) and the Output Ports Labeled in Figure 4b (Unit: dB and Degrees for the Amplitude Phase, Respectively)
Boresite beam S0j
Offset beam S0j
Finally, the radiation patterns are shown in Figure 9 for these cases. In Figure 9a, the radiation patterns obtained by the simulation and experiment for the offset beam are shown as a comparison. The main beam is accurate as expected, while the sidelobes have some increases on the sidelobe levels. These increases can be caused by the handmade prototyping in our experiments, and in theory, they may be improved by imposing an amplitude taper. The beam overlapping is shown in Figure 9b for the offset and boresite radiation beams for reference, where a roughly 10° beam separation is observed as interested in the examination. The prototype of feeding network for the offset beam is shown in Figure 9c for the reference purpose.
This paper presents an effective antenna design for the operations of dual circular polarization and multibeam radiations. The design is demonstrated and validated at 2.45 GHz band for six operation modes in the combinations of two polarizations and three-directional beams. The proposed structure allows the feeding of antenna elements and their subsequently beam-forming networks to be designed and implemented independently and provides the flexibility of system operation. Both numerical simulations and experimental measurements have been performed and presented to exhibit the radiation characteristics.