Abstract

In this article, a broadband semicircular antenna array geometry was examined, along with its design and configuration in the submillimeter wave range (9.35-42.89 GHz). The work has been divided into two prime phases; in the first phase of work, an innovative angular-phased 1 : 4 broadband power divider has been presented in which its isolation, insertion, and return losses are investigated. In the next phase, the proposed power divider was used to configure a 4-element antenna array, and its theoretical analysis was carried out. The angular path difference in the array was used to introduce phase difference in between the antenna elements to minimize the mutual coupling and to optimize the narrow beamwidth. Using geometry and fundamental array theory, the array factor of a novel broadband array configuration has been derived and analytically investigated.

1. Introduction

Recently, much attention has been drawn to submillimeter- (mm-) wave antenna arrays for future radar, satellites, and wireless communication applications. The feeding network is always a crucial component in the design of radiating structure for an array configuration [1]. For this purpose, series or corporate feeding is frequently employed, and for corporate feeding, the feeding network requires power dividers. Various configurations of power dividers have been introduced to support planar circuit applications [2]. The design of a power divider is affected by important factors such as reflection coefficient, insertion loss, isolation, and bandwidth to measure the overall performance of significant communication systems. The most commonly used Wilkinson’s power divider (WPD) is a great substitute for the designing of feed networks in antenna array systems due to its simple structure and minimum insertion loss [36]. However, WPD has drawbacks, such as narrow bandwidth and larger size due to the use of transmission line sections at a single frequency.

To overcome these issues that are present in classical Wilkinson power splitter, several techniques have been suggested to increase the bandwidth of Wilkinson’s power dividers using the methods such as open stubs [7], port extensions [8, 9], complementary conducting-strip transmission lines [10, 11], multisection techniques, and step impedance matching networks [12]. Rigorous analysis and investigation of the bandwidth broadening mechanism in the power divider with physical port isolation were also carried out to achieve the fractional bandwidth of 43.2% [13]. These techniques also report the use of lumped passive components instead of the transmission line section to reduce the circuit size and isolation loss between output ports. In an another approach, a wideband multisection Wilkinson power divider (WPD) circuit has been designed with the Euler method to calculate the characteristic impedance of WPD [14]. The percent bandwidth of the WPD was measured as 124.5% with minimum insertion loss. Another novel structure of the Wilkinson power divider with modified matching stub length has been discussed for the improvement in smaller size, low cost, equal power division, minimum insertion loss, better isolation, and a satisfactory return loss of all ports incorporated into the antenna array feeding network [15].

An innovative concept was developed for the power splitter to use additional transmission lines between quarter wave transmission lines that improve physical and electrical isolation where the elements are placed between quarter wave transmission lines at an adjustable phase angle [16]. In order to meet the demanding standards of telecommunications technology, researchers have also made efforts to address the requirements of the broadband and multiband power dividers to achieve high data rates. These techniques include the dual-band Wilkinson power divider [11], compact power divider with narrow notch band [17], broadband Gysel power divider [18], miniature broadband four-way power divider [19], and compact wideband power divider [20, 21]. These broadband PDs improve return loss, isolation, and impedance matching at high frequencies. But such PDs become very complex and require a very large area to bend the output ports. Various analysis and design procedures on compact and simple wideband power dividers have been proposed in literatures [2225]. A small, equal phase and equal power 1 : 4 wideband Wilkinson power divider was optimized in the range of 3.1 GHz-10.6 GHz [26]. A compact and easy wideband power divider was studied using linear tapered transmission lines in the frequency range of 3 GHz to 27 GHz to get the operational bandwidth of 24 GHz [27]. A conformal high-gain UWB 4-element array antenna with band rejection characteristics for WLAN applications was presented for the frequency from 3.6 to 12 GHz [28].

In the context of the above discussion on power dividers, the classical WPD has a narrow bandwidth mainly due to two reasons: one is the abrupt discontinuity between the feeder and the radiator, and the other is the mutual coupling between the radiators [29]. In the proposed design, abrupt discontinuity is removed by using stepped microstrip line feeding, and the junctions are made circular to overcome the losses due to mismatching. As a result, current travels longer path which causes an increase in electrical length. On the other hand, mutual coupling among the radiators are minimized by optimizing the angular distance among the receiver ports. By this novel configuration, a proposed design is able to get a good trade-off between strong isolation and high operational bandwidth in the design of power divider for wideband antenna array in sub-mm wave range. The work is divided into two prime phases: phase-I consists of an innovative angular-phased 1 : 4 broadband power divider in which its isolation, insertion, and return losses are investigated. In phase-II, the proposed power divider is used to design an antenna array, and its theoretical analysis is carried out.

The Wilkinson power divider is frequently used in traditional linear arrays to distribute power across the array elements that are linearly arranged. Nonlinear phased arrays had been used in certain previously documented approaches to steer the radiating beam of many antenna elements in a specific direction. Using bow-tie radiators, a 4-element subarray with a central feeding was created in [30] in which subarray was considered as a single element; then, there were 32 elements total scattered around the surface of the octal prism, which had 4 subarrays on each of its four sides. In order to determine the direction of arrival, six sides and the top of the truncated hexagonal pyramid were dispersed with seven coaxially fed microstrip patch antennas [31]. To create the directional array antennas, a few various feeding approaches, including closely linked multifeed, SIW series-parallel differential feeding network, and centrally fed by substrate integrated coaxial line, were presented [3240].

Even though numerous types of array structures have been discussed in recent years, broadband array antennas in the sub-mm wave region, which are highly desired in particular for future mobile technologies and short-range communication applications, have not yet been extensively examined. The suggested array’s innovative design offers angular path difference rather than linear path difference, as in other traditional arrays, increasing the structure’s compactness and enabling effective impedance matching across a broad frequency range in sub-mm wave. In the following subsections, the proposed antenna array is explored extensively from the perspective of antenna design, configuration, and various result curves.

2. Configuration of Power Divider

In the proposed configuration, abrupt discontinuity is removed by increasing the electrical length of the feeding by utilizing stepped microstrip line feeding to overcome the losses due to mismatch as shown in Figure 1. The dimensions of the proposed power divider are mentioned in Table 1, and the electrical length of these parameters is calculated at a center frequency of 26.12 GHz. Each subsection of the main feedline is having a size of and , respectively. These two subsections together were considered as the main microstrip line for the excitation. Two similar branches having the width of and length of were extracted out from the primary microstrip line at an angle of , either side from its axis for the power distribution. By rotating these sublines, angular path difference was created among the subbranches. As a result, mutual coupling among these subbranches can be minimized by optimizing the angular distance among the receiver ports.

These feed lines were further extended by microstrip line and were considered as the feed lines and . For current smoothening, a circle of radius of was drawn, considering the rotation point of the branches as the center of the circle. It was the first junction , from which the power was distributed from to and . The circle at the junction smoothens the flow of current, and in another way, it allows the travelling wave to overcome the losses so far and to retrieve its strength.

Similarly, other junctions were formed by extracting out the two more branches with the width of and the length of from the and at an angle of , either side from the axis of the lines, for the power distribution. Again the circle of radius of was drawn at the junction point for current smoothening. All the four extracted branches were further extended with the dimensions same as and but rotated at an angle of from the horizontal axis. To remove the discontinuity and for current smoothening, again, circles having the radius of were drawn as shown in the green (left) inset in Figure 1. Thus, two more junctions ( and ) were created to distribute the power from to and and from to and .

Considering the port of the main branch as port 1 and ports of four subbranches in a counter-clockwise direction as port 2-port 5, respectively, parameters of all the ports were calculated by using a 2.92 mm connector at each port for considering the losses due to connectors under the real-time practical conditions in the CST MWS simulator. Figure 2(a) shows the isolation among the ports, and it is to be noticed from the said figure that isolation curves for , , and are below -13 dB, whereas it is far below -16 dB for S24, S25, and S34 throughout the frequency range. Insertion and reflection coefficient curves for input port 1 are shown in Figure 2(b), in which plots of , , , , and are shown. As expected, reflection coefficient S11 is far below -10 dB, and insertion from port 1, representing the S-parameters , , , and , has approximated range of -6 dB to -4 dB in the entire range. Self-reflection at the other ports is presented in Figure 2(c), and as observed, reflection coefficients at ports 2, 3, 4, and 5 are meeting the expectations of being below -10 dB in the entire frequency range. It can also be analyzed in Figure 1 that there is a grouping of ports 2 and 3 and ports 4 and 5 together; in the same fashion, their return loss curves shown in Figure 2(c) are also matched throughout the operating range. It can be concluded that this novel configuration of a proposed power divider is able to attain a good trade-off between strong isolation and high operational bandwidth in sub-mm wave range. Once the power divider is ready, it can now be utilized for configuring the broadband antenna array which is discussed in the subsequent section.

3. Analysis and Design of the Array

At present, researchers and communication scientists are collaborating on design of beam-forming broadband networks using massive MIMO antenna systems for future short-range communication pursposes in sub-mm wave range. For this purspose, highly directive antenna arrays are required to be embedded in massive MIMO system so that multibeam forming networks may be implemented. Aiming this, a semicircular array (SCA) was implemented, using the proposed power divider for 50 Ω input impedance feedline in which a single antenna element was used that is a broadband (10-150 GHz) antipodal end-fire antenna, proposed by the authors recently [41]. To further improve the gain and directivity of the single element, this antenna was configured in an array configuration, as can be seen in Figure 3, which depicts the top perspective of the proposed array structure. According to Figure 1, four single element antennas were placed at the center points of the terminating edges of lines L31, L32, L33, and L34, respectively. When calculating the values of and , as shown in Figure 1 by the dotted line, it was considered that all four antenna elements were positioned orthogonally at the outermost edges of the semi-four octagon. The initial power junction node, P1, is located in the center of the semioctagon. These feed lines were rotated at an angle of from either side of their axes in order to produce the appropriate angular path difference between antenna elements to achieve desired highly directional power pattern and operational range. This is due to the fact that it has a direct impact on the antenna components’ mutual coupling.

Since each branch of an octagon makes an internal angle of 45° from the origin according to geometry, ° and °, an antenna element might be placed at an angle of 90° from the edge of the semioctagon. Consequently, the angular phase shift between the antenna components is . This whole array configuration was created on a semicircular Rogers RO4232 substrate with a thickness of 1.52 mm and a radius of . The geometry of the semicircular array (SCA) is presented in Figure 3, and it can be observed that point O is the center of the semicircle of radius which lies in the - plane. This semicircle was divided into six sections, and each section has an arc angle of with respect to the center of the semicircle. The derivation of array factor of semicircular array was carried out with the following assumptions [42]: (1)Single element antennas of the array were treated as the individual point sources, located at the center of their respective radiation patterns (points , , , and )(2)Current supplied to each antenna element was of same magnitude and phase(3)There was no mutual coupling among the antenna elements(4)The radiation pattern of the antenna array was observed in the far field, and hence, field vectors from all sources , , and were parallel to each other and were at an arbitrary angle of Φ from the -axis

The element was assumed as the reference point for the semicircular array. Angles between line segments and field vectors , , and are , , , respectively. Using geometry, angles were calculated in degree and can be expressed as where , 3, 4 and is the angle made by the antenna field vectors , , and from the -axis.

The line segments from the reference element to the antenna elements , , and , are 1, 3, and 4, respectively. The i was calculated by geometry, in terms of radius and angle and can be expressed as where . The path differences were calculated from the angles , , and and line segments , , and as shown in Figure 3. The relative path differences from the for the antenna elements , , and are the line segments and which are notated as , , and , respectively, and can be expressed as where . By array theory, array factor of the semicircular array can be computed as where and is the phase constant. By putting the values of , , and from eqs. (1)–(3), respectively, eq. (4) can be rewritten as where , , and

The magnitude of array factor was derived from eq. (5) and can be expressed as where .

To determine the point of maxima in eq. (6), the magnitude of array factor at centre center frequency 21.6 GHz, with respect to the angle phi (degree) was plotted as shown in Figure 4(a). It can be observed that magnitude of array factor is maximum at Φ = 90° and Φ = 270° proving that the direction of the main lobe of SCA is Φ = 90° and Φ = 270° which could have been the same in the linear phased array. As a result, the semicircular array obtained the same relative path difference among antenna components as the linear phased array. The main difference between linear and proposed phased array is the calculation of path difference. The required path difference among the antenna elements is achieved by linear displacement in the linear phased array and by angular displacement in the proposed semicircular phased array. The main advantages of the proposed array arrangement over the conventional array is size reduction, which is a crucial factor for space-borne antennas; minimal mutual coupling, and broadband characteristics.

Normalized directivity plot of a single antenna element predicted by CST MWS and a 4-element semicircular antenna array, calculated on MATLAB using pattern multiplication principle [42], at 21.6 GHz is shown in Figure 4(b). Particular reason of choosing 21.6 GHz was that it is the only resonant frequency of the SCA which is not only closest to the center frequency of the band (10 GHz to 42 GHz) but also nearest to the resonance frequency (22.5 GHz) of the single element antenna at which normalized directivity plot in H-plane was analyzed in [41]. It can be observed in Figure 4(b) that by modifying the single-element antenna into SCA, the direction of the maximum radiation remains the same for both the antenna and the array that is Φ = 90°, while the gain was significantly improved and array becomes highly direction as 3 dB angular beamwidth was decreased to 32° from 65.5°. There was tremendous change in the sidelobe level as well those were reduced to -14.1 dB from -0.5 dB for the earlier case. The same has been reflected in Figure 5 in which updations in polar patterns for ϕ = 90° from single element to array is analyzed. Thus, it can be concluded that radiation properties of the antenna array got highly improved compared to single element antenna, from the perspective of the gain as well as the directivity.

Measured reflection coefficient comparison between the single antenna element and proposed SCA is represented in Figure 6 along with the lab-fabricated prototypes, and it ensures the -10 dB impedance matching in both cases. According to the said figure, both single-element and antenna arrays comprise wide impedance bandwidth in the sub-mm wave range. The current density distribution of the proposed array at 21.6 GHz is shown in Figure 7; from the rainbow scale, it is analyzed that power is equally distributed among the antenna elements, and the mutual coupling among the antenna elements is all blue that shows minimal coupling, that results to broadband operation. The junctions are all red that shows the perfect distribution of the power at the circular junction. Figure 8(a) comprises the improvement of realized gain from the single-element to the 4-element antenna array. From the said figure, it is concluded that the average gain in array conversion goes up from 6.42 dB to 8.55 dB in sub-mm wave range 9 to 45 GHz. The effect in radiation efficiency is shown in Figure 8(b), which concludes that in both cases, average radiation efficiency is approximately 69.17%, which is fine at such a high range, meaning both the structures are radiating well in this range. Thus, from all discussed result analysis, it can be concluded that the radiation properties of the antenna array got highly improved compared to single element antenna, from the perspective of the gain as well as the directivity.

The performance of the proposed work is compared to that of other analogous low-profile and broadband end-fire antenna arrays in Table 2. As far as any of us are aware, the proposed SCA performs far better than any other known array structure in the suggested operating frequency range in terms of impedance bandwidth and gain.

4. Conclusion

An overview of the proposed semicircular antenna array architecture, as well as its design and configuration, was accomplished both theoretically and empirically. The designed antenna array was arranged with an angular phased 1 : 4 power divider, in which abrupt discontinuity was avoided using stepped microstrip line feeding, and the circular junctions were provided to prevent mismatching losses. Due to the increment of electrical length, current takes longer path to travel that causes compactness in the design as it allowed the antenna to radiate at lower frequencies as well. On the other hand, the sublines’ rotation created an angular path difference in between its subbranches. As a result, by optimizing the angular distance between the receiver ports, mutual coupling among the radiators in the array feed by these subbranches was minimized that allows wideband operation of the array. The array factor of a novel array arrangement was computed using geometry and fundamental array theory, and it has been mathematically shown that for the proposed structure, the direction of maximum radiation is Φ = 90° that is the end-fire H-plane. While comparing normalized directivity plots of the proposed 4-element array configuration to the single element, it was found that the antenna gain, directivity, angular beamwidth, and side lobe levels were all much improved when utilising the proposed array design. It can be concluded that in the sub-mm wave region, the proposed power divider’s innovative structure achieves a favourable trade-off between robust isolation and large operating bandwidth. The aforementioned array can be used in the design of multibeam-forming massive MIMO systems for long- and short-range communication networks in the future [43].

Data Availability

Raw data were generated at the research lab of the National Institute of Technology, Uttarakhand. Derived data supporting the findings of this study are available from the corresponding author, Dr. A. K Gautam, on request.

Conflicts of Interest

No potential competing interest was reported by the authors.