**Preface to "Antennas and Propagation"**

Antennas are essentially transducers, as they convert electromagnetic fields into signals and vice versa. Moreover, remote sensing or sensor networks cannot be imagined without antennas, and radiowave propagation in complex environments is a crucial aspect for the operation of such systems. New technologies, such as 5G, generate further perspectives for sensor networks and raise additional challenges for antenna design.

This Special Issue gathers topics of utmost interest in the field of antennas and propagation, such as:


We originally invited the authors who contributed to the 2020 International Workshop on Antenna Technology, held in Bucharest (Romania) on 25–28 March, to submit thoroughly extended versions of their work. Surprisingly, half of the submissions to the Special Issue through the deadline were not conference paper extensions, but completely new contributions.

> **Razvan D. Tamas, Alina Badescu, Tudor Palade, Florin Alexa, Ioan Nicolaescu** *Editors*

### *Editorial* **Antennas and Propagation: A Sensor Approach**

**Razvan D. Tamas**

Department of Electronics and Telecommunications, Constanta Maritime University, Str. Mircea cel Batran nr. 104, 900663 Constanta, Romania; tamas@ieee.org

Antennas are essentially transducers, as they convert electromagnetic fields into signals and vice versa. Moreover, remote sensing or sensor networks cannot be imagined without antennas, and radiowave propagation in complex environments is a crucial aspect for the operation of such systems. New technologies such as 5G generate further perspectives for sensor networks and raise additional challenges to antenna design.

This Special Issue gathers topics of utmost interest in the field of antennas and propagation, such as:


We originally invited the authors who contributed to the 2020 International Workshop on Antenna Technology, held in Bucharest (Romania) on 25–28 March, to submit thoroughly extended versions of their work. Surprisingly, half of the submissions to the Special Issue through the deadline were not conference paper extensions but completely new contributions.

#### **1. Summary of the Special Issue**

*1.1. New Directions and Challenges in Antenna Design and Propagation*

A slot-fed terahertz dielectric resonator antenna driven by an optimized photomixer is proposed in [1], and the interaction of the laser and photomixer is also studied. It is demonstrated that, in a continuous wave terahertz photomixing scheme, the generated THz power is proportional to the fourth power of the surface electric field on the photoconductive layer. The total efficiency was considerably improved due to enhancements in the laser-to-THz conversion as well as the radiation and matching efficiencies.

In Ref. [2], radio transmission and impedance matching in medical telemetry are investigated. Impedance matching inside a human body is studied both for electric and magnetic dipoles. The authors demonstrate that the implantation of a magnetic dipole is more beneficial than that of an electric dipole, as it provides impedance characteristics that are more appropriate to the human body as an antenna environment.

The simultaneous influence of the substrate anisotropy and substrate bending are numerically and experimentally investigated in [3] for planar resonators on flexible textile and polymer substrates. The effects are studied on various resonant structures with different types of slots and defected ground as well as on fractal resonators.

In Ref. [4], the authors present small, flexible, low-profile and light-weight all-textile antennas for wearable wireless sensor networks (W-WSNs) and investigate the impact of the textile materials on the antenna performance. A step-by-step procedure for design, fabrication and measurement of small wearable backed antennas for W-WSNs is also suggested.

The work in [5] proposes a random spreading of the scrambled timestamp sequence (STS) in ultra-wide-band (UWB) pulse communications. The timestamp estimation is

**Citation:** Tamas, R.D. Antennas and Propagation: A Sensor Approach. *Sensors* **2021**, *21*, 4920. https:// doi.org/10.3390/s21144920

Received: 7 July 2021 Accepted: 15 July 2021 Published: 20 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

improved by reducing the side lobes of the correlation. A transmission in a UWB channel with frequency selective fading shows low timestamp detection errors.

A modified, compact dipole antenna for energy harvesting applications is proposed in [6]. The design is based on a folded, short-circuited dipole radiator, and three feeding techniques are investigated. Such antenna configurations show good conversion efficiency when powering Bluetooth low-energy wireless sensors.

#### *1.2. Innovative Antenna Technologies for Space Applications*

In Ref. [7], multibeam antenna systems are proposed with the aim to provide multispot coverage for broadband satellite communications in the Ka-band. Two multibeam reflectarrays are used, making it possible to halve the number of required antennas onboard the satellite.

The work in [8] presents an antenna design that can be used as an array element for monitoring and detecting radio emissions resulting from cosmic particle interactions in the atmosphere. The proposed antenna provides a high gain over a large relative bandwidth, a narrow beamwidth, a small group delay variation and a very stable radiation pattern between 110 and 190 MHz.

#### *1.3. Metamaterial, Metasurface and Other Periodic Structures*

In [9], a new design method for a planar and compact dual-band dipole antenna is proposed. The antenna has a hybrid CRLH (composite right- and left-handed) structure with lumped elements for a dual-band operation. A design for 2.4 and 5.2 GHz mobile applications is presented.

A printed edge-fed counterpart of the Bruce wire antenna array for frequency scanning applications is presented in [10]. The unit cell of the proposed antenna consists of bowtie and semi-circular elements that cover a bandwidth between 22 GHz and 38 GHz.

The work presented in [11] focuses on design issues in the implementation of holey glide-symmetric periodic structures for waveguide-based components. An analysis of realistic hole structures is performed by using an effective hole depth method that can be used as a tool for designing electrically large waveguide-based components.

An integration of a microstrip slot antenna array for 5.8 GHz with dye-sensitized solar cells is proposed in [12]. It is shown that the antenna array has a slight influence on the solar cell performance, and the interference of the solar cells with the antenna feeding system is also negligible.

A new slot-based antenna system with horizontal polarization for unmanned aerial vehicle (UAV) ground control stations (GCS) is outlined in [13]. The proposed antenna system consists of coaxial cylinders and slots. The vertical slots are periodically placed around the outer cylinder and generate in-phase, horizontally polarized beams, resulting in an omnidirectional radiation pattern.

#### *1.4. Antennas for 5G*

In Ref. [14], a multiband antenna for microwave and millimeter-wave applications is presented. The proposed antenna consists of a slotted, conical patch connected to a small triangular patch. It can be used for wireless local area network (WLAN) applications and for fifth-generation (5G) communication devices.

In Ref. [15], the authors present a simple, compact and low-cost design method for low-profile, multi-band antennas. Such antennas can be employed in overcrowded, future generation networks in the K/Ka band. The proposed antenna structures consist of several monopoles, one for each operating frequency, along with a frequency selective feeding network. This concept leads to scalable structures suitable for 5G applications.

In Ref. [16], a small antenna for sub-6GHz 5G communications is proposed. The design consists of a wide-band antenna connected to a small multiplexer comprising three metamaterial channel filters. It can be used on channels within three frequency

bands; channel selection is experimentally demonstrated so as to prove the validity of the presented design.

#### *1.5. Electromagnetic Field Measurements and Remote Sensing Applications*

In Ref. [17], the authors propose an ultra-wide band (UWB) antenna system and a direction-finding (DF) approach based on using energy-based descriptors instead of classical frequency domain parameters. The method can be applied for locating electric discharges in high-voltage power distribution systems through their electromagnetic signature in the radio frequency range.

A calibration method for high-resolution, hybrid MIMO turntable radars is presented in [18]. A line of small metal spheres is employed as a test pattern in the calibration process in order to measure the position shift caused by undesired antenna effects. The unwanted effects resulting from the near-field antenna response are analyzed, modelled and significantly mitigated, by exploiting the symmetry and response of the MIMO configuration.

In Ref. [19], the authors show that the distance averaging technique can be applied to reduce the effect of the common mode currents for measuring the field radiated by symmetrical antennas. Two measuring configurations are proposed, depending on the number of symmetry degrees of the antenna under test. A differential approach for extracting the field created by the common mode currents is also developed.

In Ref. [20], the authors elaborate on how a sensor using modified split ring resonators (SRRs) can be designed, simulated, fabricated and used for advanced investigation and accurate measurements of the complex permittivity of solid dielectrics.

The work in [21] proposes a near-field to far-field transformation algorithm based on spherical wave expansion for near-field RCS measurements. Each weight in this expansion is calculated by using an iterative least squares QR factorization method. The proposed NFFFT is verified for several types of scatterers.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


### *Communication* **Compact Antenna in 3D Configuration for Rectenna Wireless Power Transmission Applications**

**Alassane Sidibe 1,2,\* , Alexandru Takacs <sup>1</sup> , Gaël Loubet <sup>1</sup> and Daniela Dragomirescu <sup>1</sup>**

	- <sup>2</sup> Uwinloc, 9 Rue Humbert Tomatis, 31200 Toulouse, France
	- **\*** Correspondence: alassane.sidibe@laas.fr

**Abstract:** This work presents methods for miniaturizing and characterizing a modified dipole antenna dedicated to the implementation of wireless power transmission systems. The antenna size should respect the planar dimensions of 60 mm × 30 mm to be integrated with small IoT devices such as a Bluetooth Lower Energy Sensing Node. The provided design is based on a folded short-circuited dipole antenna, also named a T-match antenna. Faced with the difficulty of reducing the physical dimensions of the antenna, we propose a 3D configuration by adding vertical metallic arms on the edges of the antenna. The adopted 3D design has an overall size of 56 mm × 32 mm × 10 mm at 868 MHz. Three antenna-feeding techniques were evaluated to characterize this antenna. They consist of soldering a U.FL connector on the input port; vertically connecting a tapered balun to the antenna; and integrating a microstrip transition to the layer of the antenna. The experimental results of the selected feeding techniques show good agreements and the antenna has a maximum gain of +1.54 dBi in the elevation plane (E-plane). In addition, a final modification was operated to the designed antenna to have a more compact structure with a size of 40 mm × 30 mm × 10 mm at 868 MHz. Such modification reduces the radiation surface of the antenna and so the antenna gain and bandwidth. This antenna can achieve a maximum gain of +1.1 dBi in the E-plane. The two antennas proposed in this paper were then associated with a rectifier to perform energy harvesting for powering Bluetooth Low Energy wireless sensors. The measured RF-DC (radiofrequency to direct current) conversion efficiency is 73.88% (first design) and 60.21% (second design) with an illuminating power density of 3.1 µW/cm<sup>2</sup> at 868 MHz with a 10 kΩ load resistor.

**Keywords:** compact antenna; wireless power transmission (WPT); energy harvesting; rectenna; wireless sensors

#### **1. Introduction**

Over the last decades, we have been faced with the miniaturization of electronic devices, especially in the field of wireless systems. The aim is to have multiple functionalities on an ever-smaller surface area. Recent IoT applications (Internet of Things) tend to employ tiny and low power electronic components [1]. However, batteries are still widely used for powering the devices despite their significant size and the frequent need for replacement. An alternative is using a battery-free system powered by energy harvesting (EH) or wireless power transmission (WPT), for instance, based on a rectenna circuit.

A rectenna is a combination of a rectifier and an antenna used to scavenge ambient or specially generated far-field radiofrequency (RF) waves [2]. The implementation of a rectenna in a battery-free system can allow increasing its lifetime, reducing its manufacturing costs, while ensuring reliable performances for decades. Several kinds of application

**Citation:** Sidibe, A.; Takacs, A.; Loubet, G.; Dragomirescu, D. Compact Antenna in 3D Configuration for Rectenna Wireless Power Transmission Applications. *Sensors* **2021**, *21*, 3193. https:// doi.org/10.3390/s21093193

Academic Editor: Razvan D. Tamas

Received: 12 February 2021 Accepted: 28 April 2021 Published: 4 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

exist, such as, for instance, rectennas, which were developed for a biomedical device pasted on the human body [3] or used in the IoT domain [4–6].

The antenna is a ubiquitous element in IoT and other wireless applications. However, the required size stays important due to the important dependence of the geometrical elements with the targeted wavelength. Several miniaturization techniques are described in the literature. Structural modification on a Printed Circuit Board (PCB) antenna consists of acting on the geometry of the antenna or by adding another element on the antenna shape.

In this sense, a coupling element, such as a rectangular ring, can allow a reduction in size and an increase in bandwidth as presented in [7]. Traditionally, small size antennas are designed by using meander lines which reduce the resonant frequency [8]. Fractal geometry is also used to miniaturize antennas [9]. Another possibility is to add reactive loading elements. This technique is employed in [10] through an LC load (a combination of a lumped inductor and a distributed capacitor). Miniaturization can also be achieved by reducing the guided wavelength through the use of a higher permittivity material, such as a ceramic—polymer composite [11]. Recent research activities were focused on developing small antennas on metamaterial as presented in [12]. In this paper, we present a miniaturization technique which consists of shaping the antenna to form a three-dimensional (3D) structure tuned for the Industrial Scientific and Medical (ISM) 868 MHz frequency band. This configuration allows us to have an electrically small antenna as described in Section 2. A more compact antenna based on the previous one is designed to compare the trade-off between size and rectenna performances. Then, Section 3 presents the experimental results of the fabricated antennas with different feeding methods to validate the use of the U.FL connector for the next steps. Simulations were carried out on the Ansys HFSS software and verified with far-field measurements performed in an anechoic chamber. The final section of this paper describes an original concept of powering a Bluetooth Low Energy Sensing Node embedded in concrete element with a compact and efficient rectenna design.

#### **2. Design of a Compact 3D Dipole Antenna**

#### *2.1. Antenna Miniaturization Methods and the First Design of the Compact 3D Dipole*

In this study, we proposed a miniaturized antenna design for WPT applications. An antenna is a resonant structure with a proper frequency depending on its length. Therefore, there are size and performance limitations for small antennas [13,14]. The size reduction imposes a smaller radiation resistance, so a lower radiation efficiency and a bandwidth limitation. Wheeler defines a electrically small antenna as one defined by the formula given in Equation (1). It means that the antenna sphere is smaller than the radian sphere, also defined as Wheeler Cap. *λ* is the wavelength at the operating frequency, *k* represents the free-space wavenumber and a is the minimum antenna sphere radius. The choice of proposing an electrically small antenna should allow us to have an antenna design in the maximum planar size of 60 mm × 30 mm. Nevertheless, the antenna bandwidth reduction does not matter with the applications in the ISM 868 MHz frequency band. The targeted antenna gain is about +1 dBi less compared to a conventional dipole antenna gain (+2.15 dBi).

$$k \cdot a < 1 \text{ with } k = \frac{2\pi}{\lambda} \tag{1}$$

The proposed design is based on a dipole antenna according to its multiple advantages, such as the symmetry of its shape and radiation pattern, its easy and low-cost fabrication, and the possibility of receiving balanced signals. The required dimensions of 60 mm × 30 mm allow us to have a conventional half-wavelength dipole antenna at 2 GHz, as presented in Figure 1.

Starting from this antenna at 2 GHz, we had investigated a technique for reducing the physical dimensions of the antenna without significantly degrading its performances (+1 dBi less of the gain). The planar structure of the final antenna design on an FR4 substrate is presented in Figure 2. Overall, a folded dipole antenna designed after size optimization

(Antenna in Figure 3 stamped "None") presents a resonant frequency close to 1.12 GHz

−

**Figure 1.** Conventional dipole antenna simulated on the required size dimensions. (**a**) The input impedance representation on Smith chart; (**b**) Half-wavelength antenna dimensions; (**c**) Simulated 3D radiation pattern at 2 GHz.

λ Ω This antenna can be considered as a half-wavelength dipole antenna at 1.12 GHz. The total length of each folded arm (L<sup>1</sup> + W<sup>1</sup> + L<sup>2</sup> = 43.4 mm) is closed to the quarter wavelength (<0.25·λ<sup>g</sup> = 31.9 mm). Figure 3 presents the width tuning of the inductive shorting loop to make a T-match structure [15]. The resonance frequency is greatly influenced by the width and the length of the T-match structure. As seen in Figure 3, the absence of the T-match structure (None) shows a significant resonant frequency of the antenna higher than 1.12 GHz where the impedance is (17.35 + j·27.3) <sup>Ω</sup>. The shorting line (W = 0) connected to the two dipole arms downshifts the resonant frequency around 1.07 GHz. Thereafter, the geometric parameters (W and L) were adjusted to match the input impedance of the antenna (50 Ω) at lower frequency. λ Ω Ω

Ω

**Figure 2.** Detailed geometry of the compact, T-match dipole antenna on an FR4 substrate (*Lsub* = 56 mm, *Wsub* = 32 mm, *L*<sup>1</sup> = 20.65 mm, *W*<sup>1</sup> = 5.95 mm, *L*<sup>2</sup> = 16.8 mm, *W*<sup>2</sup> = 10.5 mm, *L*<sup>3</sup> = 14 mm, *W*<sup>3</sup> = 9.625 mm, *E*<sup>1</sup> = 0.35 mm, *E*<sup>2</sup> = 1.05 mm, g = 2.03 mm).

**Figure 3.** Evolution of the input impedance (Real and Imaginary parts) as function of the frequency with various shorting line width (W) values.

− Ω The simulated results of the optimization on HFSS software are displayed in Figures 3 and 4. The return loss (reflection coefficient S11) indicates how much the power is reflected on the input port when the antenna is excited. A practical criterion for the antenna (impedance) input matching is generally specified at −10 dBm at least for a reference impedance of 50 Ω. As represented in Figure 4, the greater the length (L) of the short-circuited line, the more the antenna resonates at higher frequency.

**Figure 4.** Simulated S11 (return loss) of the antenna for different configurations; L = 10 mm (black), L = 30 mm (blue) and L = 51.4 mm (red).

For the next miniaturization step, the T-match structure parameters were fixed to those which give the lower resonant frequency (W = 4 mm and L = 10 mm). To downshift the operating frequency obtained from the planar antenna from 1.12 GHz to 868 MHz, a 3D configuration was investigated. The operating/resonant frequency of the antenna can be downshifted by connecting two metal strands [16]. In our case, two capacitive metallic arms with a height of 10 mm are vertically connected to the planar antenna with a short transmission line (Figure 5a). This line's position at the edge of the planar antenna was tuned to obtain an operating frequency as close as possible to 868 MHz with a bandwidth of 30.7 MHz.

**Figure 5.** (**a**) Geometry of the 3D configuration antenna with the connected metallic arms; (**b**) Simulated 3D gain polar plot at 868 MHz.

In our application at the European ISM 868 MHz frequency band, a narrow band antenna is not critical. The radiation pattern looks like a doughnut shape and the maximum simulated gain is +1.5 dBi (Figure 5b).

#### *2.2. Second Miniaturized Antenna Design*

In terms of compactness, the previous antenna was modified and two different antennas (A1 and A2) were designed, respectively, the unconnected and connected antenna to the metallic arms. The horizontal arms of the planar dipole are folded in a spline shape to occupy the blue part display in Figure 6a and reduce the length (*Lsub*) of the planar antenna. By the way, the width of the arms was adjusted to the substrate width. This curving shape was obtained after several optimized simulations, and the size of A2 (represented in Figure 6b) is only 40 mm × 30 mm × 10 mm.

**Figure 6.** Designed antennas on HFSS: (**a**) The first miniaturized antenna and the second miniaturized antenna named A2: 3D FDA; (**b**) 3D polar plot of the radiation pattern of A2 antenna at the resonant frequency (HFSS results).

The simulations performed show an operating frequency at 1.15 GHz for A1 and 868 MHz for A2. However, there is a small gain reduction of 0.4 dBi compared to the previous designed antenna, so that we still respect the first condition of a maximum +1 dBi loss on the gain from a conventional dipole antenna. The radiation patterns in the E-plane and the H-plane are presented in Figure 7.

**Figure 7.** Radiation pattern on the E-plane and H-plane at 868 MHz. (**a**) The antenna first 3D dipole antenna in Section 2.1; (**b**) The modified 3D dipole antenna (A2).

#### **3. Characterization of the Designed Antennas**

The designed antenna was manufactured in an FR4 substrate with a thickness of 0.8 mm. Dipole antennas require a balanced-unbalanced (balun) circuit or a microstrip transition for a coaxial measurement. The balun allows canceling the flowing current on the outside surface of the outer coaxial conductor and then affects the measurement [17]. In the literature, several dipole antenna-feeding techniques were proposed. A microstrip tapered balun was used as a feeding line in [18] and a microstrip-to-coplanar (CPW) feed network balun for a flexible bowtie antenna in [19]. On the other hand, in many electronic devices especially in IoT, a surface mount coaxial U.FL connector is strongly used for characterization and RF connection for embedded antennas. Its advantages are its low cost, small size and light weight.

To well characterize our designed antenna, three feeding methods were selected. The first feeding technique carried in this paper is to use a U.FL connector [20]. The pads of the connector were soldered on the antenna feeding pads, as seen in Figure 8a. The second method consists of vertically connecting a designed tapered microstrip balun (Figure 8b). The last one is a microstrip transition: while one pad of the antenna is connected to a 50 Ω transmission line, the other balanced pad is connected to the ground plane (Figure 8c). For the return loss (S11) and the gain measurements, a compatible coaxial cable was connected to the vector network analyzer (VNA). The measured return loss is plotted in Figure 9 (by using three different connection methods). They all fit well to each other but present a 15 MHz frequency shift compared to the simulation. The same frequency shift can be observed in Figure 10 between simulation and measurement for the curved 3D antenna A2. Without the two connected arms, the simulated and measured return loss fit perfectly, but once they are connected, a 50 MHz frequency shift appears. The difference is mainly due to the soldering effect and the vertical capacitive arms whose dimensions are not perfectly respected as compared with the simulated ones. The A2 antenna was measured with a U.FL connector and resonates at 878 MHz, as shown in Figure 10. The measured and simulated gain in the E-plane are shown in Figure 11.

Ω

**Figure 8.** (**a**) Antenna with a U.FL connector named AC; (**b**) Antenna with connected tapered balun named ATB; (**c**) Antenna with integrated microstrip transition named AIT.

**Figure 9.** Comparison of the measured return loss (S11) of the D1 antenna with different feeding methods.

**Figure 11.** Simulated (dashed line) and measured (continuous line) radiation pattern (gain plot in the E-plane): (**a**) AC antenna; (**b**) A2 antenna at the resonant frequency (868 MHz).

Table 1 compares in terms of performances and size some state-of-the-art antennas and the presented solution operating in ISM 868 MHz or ISM 915 MHz frequency bands. The proposed compact 3D antenna was implemented on an FR4 substrate that is lossier as compared with the substrates used by other state-of-the-art designs.

**Table 1.** Comparison with the different compact antennas for rectenna of IoT devices in the state of the art.


BW: Bandwidth, PIFA: Planar Inverted F-Antenna, DB: Dual Band.

#### **4. Rectenna and Wireless Power Transmission Experimentations**

The antennas presented in this paper are more compact than the 3D antennas in [16,23], and exhibit the best tradeoff between size, gain and bandwidth and can be good candidates for WPT applications. They were integrated in rectennas used for battery-free wireless sensors embedded in concrete structures [24,25].

Ω − Ω Ω The rectifier part is designed using ADS/Momentum software from Keysight. The circuit is designed on a microstrip coupled transmission line allowing differential feeding by the dipole antenna (Figure 12). It is composed of a SMS7630-005LF Schottky diode, an LC matching network (L1-C1) ensuring 50 Ω input impedance at the input of the rectifier and a low pass filter formed by a shunt capacitor (C3) and a resistive load (R1). However, considering the nonlinear behavior of the Schottky diode, the rectifier was designed and optimized for a low input power of −15 dBm and a 10 k<sup>Ω</sup> resistor to emulate the load (that is the input impedance of the power management unit (PMU)). The simulation result was performed and described in [24]. Two rectennas, R1 and R2, in Figure 12 represent the rectifier with the antenna AC without any connector and A2, respectively. They were made by integrating the rectifier with the antennas on the same substrate. Their characterization was performed in an anechoic chamber at the distance of 1.5 m from the patch-transmitting antenna, and the harvested DC power was measured across a 10 kΩ load through a multimeter. The rectennas R1 and R2 allow a harvested DC voltage of 788 mV and 726 mV for 1 µW/cm<sup>2</sup> .

**Figure 12.** Manufactured rectenna with the rectifier schematic.

The conversion efficiency as a function of the power density is presented in Figure 13. It is obtained by computing the formula given in Equation (2), where: 


 <sup>௧</sup> The variation by decade shows a difference between R1 and R2 of 13.67%, 6.01% and 1.48%. The difference can be explained by the non-linearity of the used Schottky diode. By reducing the planar antenna size of 30%, we reduce the antenna gain of 0.44 dBi and then the RF-DC conversion efficiency of 13% with a power density of 3.1 µW/cm<sup>2</sup> at 868 MHz. The second rectenna R2 will be preferred in cases where the power density is very low (e.g., 31 nW/cm<sup>2</sup> ). For our application, in the aim of supplying the power management unit, which requires a DC voltage of 600 mV at least, R1 was chosen.

$$\eta = 100 \cdot \frac{P\_{D\overline{C}}}{P\_{RF}} = 100 \cdot \frac{4 \cdot \pi \cdot P\_{dc}}{\text{S} \cdot \text{G}\_{\overline{I}} \cdot \lambda^2} \text{ with } \text{ S} = \frac{E^2}{120 \cdot \pi} = \frac{30 \cdot P\_{\overline{I}} \cdot G\_{\overline{I}}}{d^2 \cdot 120 \cdot \pi} \tag{2}$$

ଶ

In the last set of tests, underrun in our laboratory, an innovative Bluetooth Low Energy (BLE) battery-free wireless sensor network (composed of battery-free sensing nodes and a communicating node [26]) was operated in a harsh environment (Figure 14). Once embedded, the main objectives were to periodically monitor the physical data (e.g., temperature and humidity) of the concrete and broadcast them (by BLE) to the communicating node.

**Figure 14.** Photograph of the experimental setup for the sensing node embedded and powered by using WPT system in the concrete structure.

The sensing node in Figure 15 is composed of a rectenna (including the compact 3D antenna), an ultra-low-power BLE System on Chip (SoC) (NXP QN9080) [27], a power management unit microcontroller (Texas Instruments bq25570) [28], a capacitor of 100 µF acting as an energy storage element (Panasonic EEEFK0J101P) [29] and low-power temperature and humidity digital sensors (Texas Instruments HDC2080) [30]. The capacitor was chosen after evaluating the power needed by the PMU (870 µJ) to startup from deep-sleep mode. Through a far field WPT system, the illuminating power over a distance of two meters can be harvested by the rectenna and allow powering the sensing node. Initially empty, the capacitor is charging through the PMU on deep-sleep mode up to 1.5 V and switches on fast charging until the stored voltage reaches 5.3 V. Once it has enough energy stored by the capacitor, it discharges and allows powering the active component on the board (sensors and BLE chip) for a measurement and data transmission.

**Figure 15.** Developed sensing node for BLE communication.

Preliminary tests, with an EIRP power of +33 dBm, allow a periodicity of charging, measuring and wireless communication equal to at most 190 s, representing the first charge duration (cold start + fast charging + data measurement and transmission).

#### **5. Conclusions**

An electrically small antenna operating in the ISM 868 MHz frequency band and dedicated to wireless power transmission applications is proposed in this paper. Its design is based on a folded dipole antenna with a shorting line to form a T-match antenna. A 3D configuration allows the exploitation of the Z-plane in the final implementation. Thus, the connection of two metallic arms was operated by the planar antenna to downshift the operating frequency from 1.09 GHz to 868 MHz. The design was modified to apply a spline shape to the planar antenna arms in order to reduce the overall dimensions (in the horizontal plane) from 56 mm × 32 mm to 40 mm × 30 mm.

For the characterization, three feeding methods were used to provide accurate experimental results. In this application, the U.FL connector can be used for measurement but may have limitations at higher frequencies. The experimental results confirm that the proposed 3D antenna exhibits a gain of +1.54 dBi and +1.1 dBi at the operating frequency, respectively, for AC and A2. As shown in Table 1, a tradeoff between the compactness, the maximal gain and the bandwidth should be performed before the design. Our solution is also the most compact and has a maximum gain higher than +1 dBi at the price of a reduced bandwidth. The combination of these antennas and an optimized rectifier performs a high efficiency of 73.88% and 60.31%, respectively, for R1 and R2 with an illuminating power density of 3.1 µW/cm<sup>2</sup> . Future works with a WPT system associated with a battery-free wireless sensor network using BLE are in progress.

**Author Contributions:** Conceptualization, A.S. and A.T.; methodology, A.S.; software, A.S.; validation, A.S. and G.L.; investigation, A.T.; writing—original draft preparation, A.S, A.T., G.L. and D.D.; writing—review and editing, A.S., A.T., G.L. and D.D.; supervision, D.D.; project administration, A.T. and D.D.; funding acquisition, A.T. and D.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** Region OCCITANIE, France in the frame of the OPTENLOC project, supported this research. The system tests were performed in the frame of the ANR McBIM project.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Communication* **Horizontal Polarized DC Grounded Omnidirectional Antenna for UAV Ground Control Station**

**Muhammad Shahzad Sadiq <sup>1</sup> , Cunjun Ruan 1,2,\* , Hamza Nawaz <sup>3</sup> , Shahid Ullah <sup>1</sup> and Wenlong He <sup>4</sup>**


**Abstract:** A new slot-based antenna design capable of producing horizontal polarization for unmanned aerial vehicle (UAV) ground control station (GCS) applications is outlined in this paper. The proposed antenna consists of oversize coaxial cylinders, slots, and slot-feed assembly. Each of the four vertical slots, arranged periodically around the antenna's outer cylinder, emits a horizontally polarized broad beam of radiation, in phase, to produce an omnidirectional pattern. The antenna possesses a low-ripple ±0.5 dB in azimuth gain (yaw) due to its symmetric axis shape and an enclosed feed within itself, which does not radiate and interfere with the main azimuth pattern. This is crucial for a UAV GCS to symmetrically extend its coverage range in all directions against yaw planes. Simulation and measurement results reveal that the antenna maintains stable gain in the omnidirectional pattern (+0.5 dB) over the entire operational frequency band (2.55 GHz to 2.80 GHz), where S11 is lower than −10 dB. A further advantage of this configuration is its enhanced polarization purity of −40 dB over the full frequency band. The direct-current (DC) grounding approach used in this antenna is beneficial due to its electrostatic discharge (ESD) and lightning protection. Furthermore, its aerodynamic, self-supporting, and surface-mount structural shape makes this antenna a good and worthy choice for a UAV GCS.

**Keywords:** horizontal polarization; UAV ground station; Omni-directional

#### **1. Introduction**

Since the discovery of the interdependency between electrical parameters and electromagnetic radiation, antennas have been developed that actively exploit this phenomenon. Antennas convert electrical parameters (current and voltages) into electromagnetic parameters (electric and magnetic fields) and vice versa. Hence, an antenna can be regarded as a transducer or a sensor as it converts electrical energy to electromagnetic energy, or the opposite [1]. Antennas are always considered essential parts of communication systems, and their radiation and polarization characteristics play a vital role in defining such systems' performance and efficiency [2,3].

The use of unmanned aerial vehicles (UAVs) is rapidly expanding to commercial, scientific, agricultural, and military applications [4]. To overcome the difficulty of finding the exact location of mobile UAVs from ground control stations (GCS), omnidirectional antennas are utilized to resolve acquisition and pointing complications [5–10]. It has been proven that using horizontally polarized antennas can achieve a 10 dB improvement in terms of system gain as compared to vertically polarized antennas [11]. For GCS deployment, where the antenna is intended to cover a wide range of angles at variant

**Citation:** Sadiq, M.S.; Ruan, C.; Nawaz, H.; Ullah, S.; He, W. Horizontal Polarized DC Grounded Omnidirectional Antenna for UAV Ground Control Station. *Sensors* **2021**, *21*, 2763. https://doi.org/10.3390/ s21082763

Academic Editors: Shuai Zhang and Razvan D. Tamas

Received: 16 January 2021 Accepted: 5 April 2021 Published: 14 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

distances, it is essential to utilize a low-gain ripple radiation pattern to ensure continuous coverage in the yaw plane (azimuth plane) [12–14] as gain ripple fluctuates and reduces the coverage range at variant horizontal plane angles [15].

The challenging aspect of designing horizontally polarized omnidirectional antennas is producing a uniform and in-phase current in the antenna's azimuth plane. That necessary condition can be fulfilled by utilizing a single loop or multi-element arrangements [16]. Three primary topological schemes are used with omnidirectional horizontally polarized antennas. In the first topology, a single radiating structure, such as a loop, is utilized to achieve horizontally polarized radiations [17–20], but it is inherently band-limited due to an open feed. The second group imitates the loop arrangement of first with dipole elements arranged in a ring or circular array format [21–26] at the expense of a complex open feeding arrangement. The third topology utilizes slots to complete the horizontally polarized antenna. There are a few slot-based omnidirectional antennas described in the literature. In [27], an omnidirectional antenna operating at X band used an array of slot doublets etched in the broadside wall of the rectangular waveguide. However, there was no mention of the azimuth gain, gain fluctuations, and operating band. A slotbased antenna capable of producing horizontal polarization was constructed by arranging alternate slots with opposite tilt angles along the axis with intervals of λg/2. To improve the antenna's performance, alternate axial slot arrays were shifted by λg/4 along the axis. Even then, it was not improved by more than −7 dB [28]. In [29], an omnidirectional antenna was proposed, but it was circularly polarized. Moreover, it was not direct-current (DC) grounded and had a built-in main beam frequency scanning problem. In [30], a slant polarized omnidirectional antenna was presented. All slot-based horizontal polarized topologies were arranged in a series of fed axial arrays to achieve the required polarization. The other two methodologies had open feeding networks that interfered with the radiating apertures and perturbed antenna radial symmetry causing an uneven azimuth gain pattern, which further reduced antenna coverage range.

This paper proposes an omnidirectional antenna capable of achieving low azimuthal gain variations of ±0.5 dB. This work is the first single-element design based on slots capable of horizontal polarization and stable gain without making a complex axial array to achieve the required polarization. The flaunted antenna comprises four slot apertures evenly spaced around the antenna's outer circumference. It also encloses the feeding topology, so antenna symmetry is not disturbed. The device's compactness, ruggedness, and direct-current grounding are further important features of this antenna design. The proposed technique has improved polarization stability since the cross-pols are very weak relative to the co-pols. The antenna structure is exhibited and explained in Section 2. Section 3 elaborates on the slot-feed mechanism. In Section 4, simulation verification is performed. Section 6 describes the manufacturing and measurements of the antenna prototype. Section 6 presents a comparison of the proposed work with those published. Finally, Section 7 details our conclusions.

#### **2. Antenna**

The structure of the suggested horizontally polarized omnidirectional antenna is shown in Figure 1. It must be shaped like a pole due to vehicle-mounting requirements. It is based on the coaxial line and is composed of inner and outer conductors. There are etched slots around the coaxial cylinder, and the internal and external coaxial cylinders are separated by air. The primary radiation is emitted via the slots (each slot is matched to a dipole with the magnetic current source), which are periodically positioned along the antenna's outer cylinder as depicted in Figure 1a. According to Babinet's principle, the slots are complementary to the dipole antenna. The far field of the linear dipole [31], is found using:

$$E\_{\theta} = \left(j60I\_m \cos\left(klcos\theta\right)e^{-jkr}\right)/r\sin\theta$$

= (60 ()

half wavelength slot, 2l = λ/2, and

= ( () − )/

= (2)/

*θ*

−

)/

**Figure 1.** The geometry of the horizontally polarized omnidirectional antenna: (**a**) 3D view, (**b**) crosssectional view.

In the equation, *θ* is the angle between the line direction and the dipole. This means the pattern function of the dipole is the same as the slot antenna.

 $F\_{θ} = (cos$  $\cos$  (kl $cosθ)$   $-cos kl$ )/ $sin θ$ 

For idea half wavelength slot, 2l = λ/2, and

$$F\_{\theta} = \cos(2\pi cos\theta) / \sin\theta$$

The pattern of the slot antenna is the same as the dipole with the same length, but their elevation plane (E-plane) and omnidirectional plane (H-plane) are exchanged according to the duality principle. Each slot aperture produces horizontally polarized radiation. Four apertures around the circumference complete the antenna, as illustrated in Figure 1a,b, and radiate in an omnidirectional pattern. The SMA connector smoothly converts the TEM modes from SMA to a large antenna assembly with a matching structure that is an optimized inner pin height, as given in Figure 1b.

The diameter of the outer cylinder, the diameter of the feed pin that connects the inner cylinder to the outer part of the antenna, and the length of the slots are what primarily impact the performance of the antenna. The optimal specifications are listed in Table 1.


**Table 1.** Optimal parametric values of the antenna.

At UAV GCSs, there are relaxed limitations with regard to size and weight compared to aerial platforms [32]. For military operations, the UAV operator at the GCS is located in a harmless, secured place while the desired information or strategic data from the battlefield is gathered remotely. For such applications, antennas must be capable of withstanding all terrain operational area requirements and should be able to function correctly under extreme weather scenarios. So the required antenna should be mechanically robust and sturdy without external supports as these supports would increase the antenna's size [4] and result in more drag, which might weaken the antenna's structure due to rigorous

terrain and weather conditions [4,33]. Thus, it is crucial to use a compact, aerodynamic design. There is often a chance that an instance of peak instantaneous power (PIP) happens inside the printed circuit board- (PCB) based feed network. Such an event would easily damage the PCB [34], so the feed must be capable of bearing sudden PIP. The antenna would also be the primary source to channel electrostatic discharge (ESD) and lightning into the electronic systems. An ESD incident would place the functionality and safety of these systems at risk, while a lightning bolt would annihilate them. Keeping the antenna DC grounded is the most feasible and efficient strategy used in combat [35]. This antenna design would circumvent all the problems described above. The axis-symmetrical, allmetal rugged antenna is primarily constructed of brass and is DC-grounded. The solid metal feed network is enclosed inside the antenna's conformal and compact shape.

#### **3. Feed Mechanism**

Horizontal slots induce vertical polarization as they can quickly interrupt the longitudinal surface current on the antenna's outer surface [36], as seen in Figure 2. Conversely, the longitudinal slots in the antenna's outer surface cannot be stimulated due to their orientations that are in line with the surface current, and even a short circuit would not modify the flow of the surface current [28]. So it is not easy to produce horizontal polarization using a slot configuration on a coaxial cylinder. In our design, feed pins are inserted to excite the vertical slots, which connect the outer conductor of the oversized coaxial cable with its inner conductor, as shown in Figure 1b. Thus, these slot apertures are energized sideways while the opposing sides are kept floating. The slot is regarded as a dipole having a magnetic current source [29], so the slot is λg/2 long. Normally, the external feed has a built-in problem where it radiates along with the main radiating elements and causes a significant gain ripple in the omnidirectional pattern. Here, we have designed an internal feed that runs inside the radiating part and does not interfere or radiate. As for the actual feeding of the antenna, a standard SMA connector is used for feeding. The SMA connector is a coaxial structure and the antenna designed in this section is also based on coaxial structure, so the matching structure is designed and inserted between the radiation part and the feed part according to impedance transformation of coaxial transmission line [37],

$$Z\_{\text{oversize}} = Z\_{\text{match}}(Z\_{\text{suma}} + jZ\_{\text{match}}\tan\mathfrak{B}\mathrm{T})/(Z\_{\text{match}} + jZ\_{\text{suma}}\tan\mathfrak{B}\mathrm{T})$$

where the *T* is the length of the matching pin and *Zsma*, *Zmatch* , and *Zoversize* are the characteristic impedances of the SMA connector, matching pin, and oversize antenna assembly, respectively. The SMA connector's inner pin's optimized height ensures a seamless transition from normal Coaxial TEM mode to oversized TEM cable mode. The slot excitation of the proposed antenna is simple and easy without involving baluns or impedance transformers. Four pins join the inner cylinder to the vertical slots in the antenna's outer cylinder. The feed and antenna can be conveniently integrated by arranging the manufactured parts together around the central axis.

**Figure 2.** Slot configurations: (**a**) vertical and horizontal slot, (**b**) vertical slot feed.

total number of slots along the antenna's circumference. Each slot radiates a directed

umber of slots along the antenna's circumferential axis.

ntenna's circumferential

reflection. The change of diameter changed the antenna's

#### **4. Simulation Verification**

CST Microwave Studio was been used to simulate and optimize the antenna design. Figure 3 demonstrates the mutual connection between the antenna azimuth gain and the total number of slots along the antenna's circumference. Each slot radiates a directed pattern. With each increment in the number of slots along the antenna's circumferential axis, these directional radiations widened, as shown in Figure 3. Four slots made the radiation patterns combine and generate a low-ripple horizontal polarized omnidirectional radiation pattern. total number of slots along the antenna's circumference. Each slot radiates a directed ntenna's circumferential

umber of slots along the antenna's circumferential axis. **Figure 3.** Azimuthal gain vs. the number of slots along the antenna's circumferential axis.

#### **A. Determination of Pin Diameter and Slot Size Effect**

reflection. The change of diameter changed the antenna's Figure 4 helps us to see the impact of the slot-feed pin diameter on the antenna input reflection. The change of diameter changed the antenna's matching, as shown in Figure 4a. Figure 4b depicts how the slot-length variations shifted the antenna resonance region.

**Figure 4.** Effect on S<sup>11</sup> (**a**) by changing the feed pin diameter and (**b**) by changing the slot length.

#### **B. Field Verification**

software. The electric and magnetic fields' cross The field simulations were performed with the help of CST Microwave Studio software. The electric and magnetic fields' cross-sectional views through the SMA connector are shown Figure 5a,c. The cross-sectional views of the electric and magnetic fields through

radially outward as that of the TEM mode. This demonstrates that the SMA connector's inner pin's adjusted length effectively converted the connector TEM mode into the

the pins that connect the inner coaxial cylinder to the slot apertures in the outer coaxial cylinder are shown in Figure 5b,d. At the input SMA connector of the antenna, the electric field is spread radially outward (TEM mode). At the edge of the oversized coaxial antenna assembly near the SMA connector, the electric field is again radially outward as that of the TEM mode. This demonstrates that the SMA connector's inner pin's adjusted length effectively converted the connector TEM mode into the oversized coaxial assembly TEM mode, as shown in Figure 5a. As this mode travels toward the pin connected to the slot, the field circulates the slot area. All four slots have the same circulation pattern, which indicates that they are all in phase, as seen in Figure 5b. The electric field steadily travels from the slot middle toward the slot end and thereby emits a horizontally polarized field, as seen in Figure 5a,b. The directional radiation slot patterns are large enough to converge to generate an omnidirectional, horizontally polarized outward wave. Correspondingly, the magnetic fields form closed loops (TEM mode) at the SMA feed and eventually transform to perpendicular loops corresponding to the E field outside the antenna, as clearly visible in the simulated field trajectories in Figure 5b,d.

**Figure 5.** The cross-sectional views of the fields at the SMA connector: (**a**) the electric field, (**b**) the magnetic field crosssectional views of the fields at the slot-feeding pins, (**c**) the electric field (**d**) the magnetic field.

less than −10 dB from 2.5 GHz to 2.8 GHz, which were in good harmony with sim

VNA's help. In Figure 6b, the simulated and measured S

#### **5. Antenna Fabrication and Measurement Result**

An antenna was manufactured using brass for the design validation. This antenna can be assembled using CNC-machined parts or expensive 3D printing. This antenna was built utilizing the first approach. The antenna had a reduced footprint and conformal shape to maintain low air resistance. The simulated and measured antenna test results are discussed in this section. Figure 6a displays an image of the prototype antenna. The input scattering parameter S<sup>11</sup> of the manufactured antenna was measured with Agilent N5242A VNA's help. In Figure 6b, the simulated and measured S<sup>11</sup> are plotted. Measurements were less than −10 dB from 2.5 GHz to 2.8 GHz, which were in good harmony with simulations. The antenna was a reasonably broadband structure (achieved bandwidth of 11.3%).

**Figure 6.** (**a**) The fabricated prototype antenna. (**b**) Simulated and measured S<sup>11</sup> of the antenna.

polarization levels in the azimuth plane are more than −40 dB down, which agrees Measured and simulated vertical elevation planes and horizontal azimuth planes of the antenna at 2.6 GHz and 2.7 GHz are plotted in Figure 7. Measurements were done in the the compact antenna test range (ATR) of March Microwave Systems B.V., which uses a source antenna that radiates a spherical wavefront and two secondary reflectors to collimate the radiated spherical wavefront into a planar wavefront within the desired test zone where the test antenna is placed and precalibrated standard gain antennas are used to determine the absolute gain of the AUT(antenna under test). The simulated and measured co-polarization (normalized) and cross-polarization (normalized) radiation patterns in the omnidirectional plane (H-plane) are shown in Figure 7a. The 360 ◦ radiation at the horizontal plane helps to maintain complete yaw plane operation. The measured crosspolarization levels in the azimuth plane are more than −40 dB down, which agrees with the simulation. Figure 7b depicts the simulated and measured co-polarization (normalized) radiation patterns in vertical elevation (E-plane). Figure 8a,b shows a measured azimuth gain ripple of ±0.5 dB, whereas the azimuth pattern phase ripple is only 10 ◦ peak-to-peak. These were measured at 2.6 GHz and 2.7 GHz, respectively. Both affirm the excellent stability of the antenna pattern.

In Figure 9a, the measured and simulated azimuth gain ripple are plotted for the entire frequency range for the clear visibility of the gain fluctuations. The maximum peakto-peak value is 1 dB in the azimuth plane, confirming a good omnidirectionality. Figure 9b illustrates the simulated and measured gain of our antenna. This DC-grounded antenna demonstrated reliable gain within the entire band. These results indicate promising and effective radiation characteristics in the yaw plane, making this antenna an appealing option for GCS applications.

(**a**) (**b**)

polarization levels in the azimuth plane are more than −40 dB down, which agrees

**Figure 7.** (**a**) Measured and simulated normalized co-polarization and cross-polarization in the omnidirectional plane, or H-plane; top 2.6 GHz; bottom 2.7 GHz. (**b**) Measured and simulated normalized co-polarization in the elevation plane, or E-plane; top 2.6 GHz; bottom 2.7 GHz.

**Figure 8.** (**a**) Measured gain ripple vs. azimuth angle, (**b**) measured phase angle ripple vs. azimuth angle, at 2.6 GHz and 2.7 GHz, respectively.

**Figure 9.** (**a**) Measured and simulated azimuth gain ripple vs. frequency. (**b**) Measured and simulated antenna gain vs. frequency.

#### **6. Comparison**

Table 2 compares this study to the prior works published in the literature that also featured horizontal polarization. All works tabulated were designed with external open feed and lack lighting protection capability. The polarization purity was also not so high. This work proposes for the first time an antenna that could produce horizontal polarization utilizing slot radiators and form a stable gain at the azimuth plane like traditional omnidirectional design topologies such as loop or printed dipole antenna arranged in the form of a circle. Due to vertical slot radiations, its cross-polarization levels are extremely low, as shown in the Table. The proposed antenna is novel because it has an internal axis-symmetric feeding system. Due to its enclosed nature, the feed does not radiate and interfere with the main radiating slots. So low-gain ripples in the azimuth plane are achieved as compared to the listed works. Moreover, it is DC-grounded, which is essential for any practical deployment.


**Table 2.** Performance comparison of the proposed antenna with the existing literature.

#### **7. Conclusions**

A novel horizontally polarized omnidirectional antenna that is built on the slot structures is introduced in this work. By positioning four vertical slots along the antenna circumference and energizing them with a robust central feeding mechanism, steady gain with improved polarization stability is realized in the antenna's azimuth axis. The internal axis-symmetric feed network itself radiates no power, hence it does not interfere with the radiating structure. Low ripples appear among the antenna azimuth gain. Steady yaw

plane gain with low gain fluctuations enhances coverage area or increases the link efficiency. So it is desirable in numerous ground station-based applications, such as UAV communication and direction finding, to have a low azimuth gain ripple antenna. This antenna also possesses requisite mechanical features, which are crucial for its smooth operation. It has a sturdy, DC-grounded construction that does not need any exterior framework or support. Its conformal and aerodynamic shape minimizes air drag and any corresponding degradation attributed to military operations' environmental and terrain conditions. Altogether, this may make this antenna a favorable candidate for ground stations.

**Author Contributions:** M.S.S., in collaboration with H.N., proposed the antenna's configuration and simulated it in CST. M.S.S. executed the fabrication and tuning of the antenna prototype. H.N. did the measurements. M.S.S. and S.U. did the plotting and manuscript preparation. C.R. and W.H. provided step-by-step supervision, reviewed the work, and did manuscript refinement. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work is supported by the National Natural Science Foundation of China (Grant No. 61831001) and the High-Level Talent Introduction Project of Beihang and the Youth-Top-Talent University (Grant No. ZG216S1878) and Support Project of Beihang University (Grant No. KG12060401).).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** Authors declare no conflict of interest.

#### **References**


**Ilkyu Kim <sup>1</sup> , Sun-Gyu Lee <sup>2</sup> , Yong-Hyun Nam <sup>2</sup> and Jeong-Hae Lee 2,\***


**Abstract:** The development of biomedical devices benefits patients by offering real-time healthcare. In particular, pacemakers have gained a great deal of attention because they offer opportunities for monitoring the patient's vitals and biological statics in real time. One of the important factors in realizing real-time body-centric sensing is to establish a robust wireless communication link among the medical devices. In this paper, radio transmission and the optimal characteristics for impedance matching the medical telemetry of an implant are investigated. For radio transmission, an integral coupling formula based on 3D vector far-field patterns was firstly applied to compute the antenna coupling between two antennas placed inside and outside of the body. The formula provides the capability for computing the antenna coupling in the near-field and far-field region. In order to include the effects of human implantation, the far-field pattern was characterized taking into account a sphere enclosing an antenna made of human tissue. Furthermore, the characteristics of impedance matching inside the human body were studied by means of inherent wave impedances of electrical and magnetic dipoles. Here, we demonstrate that the implantation of a magnetic dipole is advantageous because it provides similar impedance characteristics to those of the human body.

**Keywords:** biomedical devices; wireless communication link; near-field region; impedance matching characteristics

#### **1. Introduction**

Recent advancements have been made in fields related to the development of biomedical devices for the diagnosis and treatment of patients. In particular, pacemakers implanted in the chest offer real-time health care for patients who suffer from cardiac disease. Pacemakers provide the important capability of sensing intrinsic cardiac activity and transferring the pacemaker data using a wireless communication link. A class of antennas with different characteristics in terms of sizes and radiation patterns have been studied for viable in-body, on-body and off-body medical telemetries [1,2]. In addition to the antenna designs, studies on radio transmission have been performed with the aim to create reliable communication links between devices outside and inside the human body. Wireless power transfer (WPT) using electromagnetic waves can be classified into the following groups: (1) non-radiative, (2) radiative and non-radiative mid-field, and (3) radiative far-field [3]. In addition, WPT using an acoustic link can be classified as a non-EM (Electromagnetics) solution [4]. Active research has been taking place on the use of inductive coupling for non-radiative wireless links [5–7]. Inductive coupling has been studied with the aim to achieve a higher transfer efficiency by utilizing a hybrid microstrip and coil [5], in addition to resonators inside and outside body [6]. The design guidelines for inductive power transfer have been studied in terms of miniaturization, power consumption, and links of biomedical devices [7]. The antennas and coils used need to be interpreted as a circuit model, which requires an additional step to compute the wireless link. Radiative WPT has garnered a lot of attention due to its robustness against the misalignment of two antennas. Radio transmission of the

**Citation:** Kim, I.; Lee, S.-G.; Nam, Y.-H.; Lee, J.-H. Investigation on Wireless Link for Medical Telemetry Including Impedance Matching of Implanted Antennas. *Sensors* **2021**, *21*, 1431. https://doi.org/10.3390/ s21041431

Academic Editor: Razvan D. Tamas

Received: 31 January 2021 Accepted: 17 February 2021 Published: 18 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

different in-body, on-body and off-body telemetries has been investigated both by using numerical methods, such as the FDTD method, and via experimentation [8]. RFID (Radio Frequency Identification) technology for biomedical implants has been studied in terms of the wireless transfer of power and data [9]. The challenge is to calculate the wireless communication link with a reduced computational intricacy. In addition, although there have been a wide range of studies on radiative mid-field WPT [10,11], little attention has been directed towards the inclusion of accurate radiation characteristics in the near-field region. The antenna gain typically reduces as one antenna is located in the near-field region of the other antenna, which needs to be considered in the calculation of the wireless link. A simple quadratic correction term [12–14] to the Friis formula has been studied to increase the accuracy of the prediction in the Fresnel region. The transmission integral [15] with the plane–wave scattering matrix (PWSM) has facilitated the development of the near-field antenna measurement technique for various antennas. Based on the transmission integral, an integral form of coupling formula has been developed in order to provide the convenience of using an easily attainable far-field pattern [16]. The formula provides an enormous capacity to compute the near-field or far-field antenna coupling, which is not restricted to the type of antenna or the motion of the antenna [16–20]. Advancements in the formula have been made in terms of its widespread applicability to electromagnetic problems. It has been proven that the employment of a larger solid angle is beneficial for achieving converged results [19]. For microwave applications, it has been used to compute the coupling, including the dielectric structure between two antennas [19], and to provide near-field power densities of the array antenna [20]. The qualitative comparison among different techniques is summarized in Table 1. Another important aspect in realizing a reliable wireless link is the use of an optimal antenna to provide the best impedance matching in the environment of a human implant. In order to provide the best match, antennas must be electromagnetically characterized in the environment, for example, inside and outside the human body. In order to achieve improved matching characteristics, the design of a pyramidal horn antenna, including a composite material which is similar to the human skin, has been optimized [21]. The impedance characteristics of electrical and magnetic dipole antenna have been studied in detail [22,23]. The advantage of the magnetic dipole antenna is that it provides lower impedance characteristics, corresponding to those of human tissues. The use of a magnetic dipole is advantageous in terms of achieving characteristics that best match those inside the human body. In order to establish a stable wireless link, the wireless link needs to be estimated using an efficient method, including the selection of the optimal antenna to provide the best impedance characteristics inside the human body.


Coupling formula [16–20] Radiative near-field and far-field region High Low

**Table 1.** Comparison among different methods for estimating the wireless link.

In this paper, radio transmission between antennas that are placed inside and outside the human body is studied in the context of realizing a stable wireless communication link. The wireless link is used to transfer pacemaker data, which operates at 402–405 MHz for a medical implant communication system (MICS). Figure 1 depicts the evaluation scheme for radio transmission between antennas inside and outside the human body. The complete wireless communication link is evaluated in terms of the efficient estimation of the wireless link and the best matching characteristics inside the human body. The key milestones of this work are that: (1) The coupling formula is applied to estimate the wireless link for biomedical applications; (2) for use in biomedical applications, the formula is slightly modified in terms of including characterization of the far-field pattern inside

the body and adding a reflection coefficient term between the implanted antenna and the human body; (3) impedance matching in the environment of the human body was studied through representative examples, such as electrical and magnetic dipole antennas. The wireless link was computed using the coupling formula and the result was compared with full-wave simulation, FEKO, and measurements. The wireless link was measured using phantom fluid that provides relative permittivity *εr* = 46.4 and conductivity *σ* = 0.67, which is similar to the characteristics of the human skin. The measured result agrees well with the computed result and FEKO simulations. 

**Figure 1.** Overview of this research study for the evaluation of the communication link between two antennas.

#### **2. Computation of the Antenna Coupling in the Near-Field and Far-Field Region**

The power transmission is computed using the integral coupling formula based on the 3D vector far-field pattern. The formula computes the antenna coupling in the near-field and far-field region. The formula was modified slightly to provide an accurate link analysis for biomedical applications. The far-field pattern of the antenna placed inside human body was characterized using a piece of human skin, with the associated antenna. Furthermore, the impedance mismatch of the antenna in the human body environment was added in order to include the reflection coefficient between the antenna and the human body. The formula was used to evaluate the link between two biomedical antennas placed in various configurations.

#### *2.1. Revisiting the Integral Coupling Formula*

The transmission integral with the plane–wave scattering matrix (PWSM) theory has expanded its applicability to arbitrarily oriented antennas [15]. The coupling formula in [16,18], based on a normalized vector far-field pattern, was derived from the transmission integral. The formula utilizes the complex vector far-field pattern, which is easily attainable through numerical methods or measurements. In this section, the essentials of the coupling formula presented in [16,18] are revisited. The geometries of the two arbitrarily positioned antennas are shown in Figure 2. The coupling quotient can be determined based on the relationship between the amplitude of the input wave of the transmitting antenna, *a*0, and the output wave of the receiving antenna, *b* ′ 0

$$\frac{b\_0'}{a\_0}(\stackrel{\rightarrow}{R}) = -\frac{\text{C}}{4\pi\text{k}} \iint\limits\_{\sqrt{\left(k\_x^2 + k\_y^2\right)} < k} \frac{\stackrel{\rightarrow}{f}\_{TX}(\stackrel{\rightarrow}{k}) \cdot \stackrel{\rightarrow}{g}\_{RX}(-\stackrel{\rightarrow}{k})}{\stackrel{\rightarrow}{f}\_z \stackrel{\rightarrow}{\cdot}^\*} e^{i\vec{k}\cdot\vec{R}} \, dk\_x \, dk\_y \tag{1}$$

where → *k* = *k<sup>x</sup> x*ˆ + *kyy*ˆ + *kzz*ˆ is the wave vector in free space and → *f TX*( → *k* ) and <sup>→</sup> *g RX*(− → *k* ) are the complex vector far-field pattern of the transmitting and receiving antenna, respectively. A double integral of the scalar product between the two vector far-field patterns was used to calculate the coupling quotient. The relationship between the coupling quotient and the S<sup>21</sup> can be defined as S<sup>21</sup> = *b* ′ 0 /*a*<sup>0</sup> 2 . ሬ⃗ = ௫ො+௬ො+௭̂ ⃗ ்൫ሬ⃗൯ ⃗ோ൫−ሬ⃗൯ Sଶଵ = |′/ | ଶ

**Figure 2.** Geometries of the mutual coupling between the two antennas used for biomedical applications.

ට൫<sup>௫</sup> <sup>ଶ</sup> + <sup>௬</sup> ଶ ൯ < ିఠ௧ Reductions in the integration range, such as r *k* 2 *<sup>x</sup>* + *k* 2 *y* < *k*, only provides propagating waves, while it neglects evanescent waves. It is worth noting that Formula (1) depends on the *e* −*iωt* time convention and neglects multiple reflections. The solid angle *α* can be defined as the angle subtended by two spheres enclosing the transmitting and receiving antennas. The first step is the acquisition of the 3D vector far-field patterns of the two antennas. The 3D vector far-field pattern can be acquired through numerical methods, full-wave simulations, and measurements. One must obtain the far-field pattern at the phase center of the antenna. The phase center of the antenna is placed at the point where the uniform phase response of the far-field pattern is observed. This procedure will be helpful in terms of achieving sufficient convergence of the calculation. In this study, the far-field pattern characterized in the human body environment was employed. It was demonstrated in [8], that the use of a small piece of human skin creates similar EM characteristics to those of the entire human body. Therefore, the antenna was implanted inside a small piece of human skin to characterize the far-field pattern. The next step is to perform the coordinate transformation in order to create the geometries of the two antennas. Eulerian angles were used to transform the coordinate system of each antenna into the global coordinate system. The relative orientation and separation distance between two antennas can be created using coordinate transformation. An interpolation process was used to evaluate the far-field pattern at the sampling points on the *x*–*y* plane.

The impedance matching of the antenna placed inside and outside human body represents an important consideration in realizing a stable communication link. Therefore, it is advantageous to contain the reflection coefficient between an antenna and the human body. Figure 3 shows the two-port network for the impedance mismatch term *C*. It was assumed that the transmitting antenna is placed inside the human skin, as discussed in Figure 1. The complete impedance matching term, constant *C*, can be expressed as

$$\mathcal{C} = \frac{Z\_{\rm FL,RX}}{Z\_0} \frac{1}{(1 - \Gamma\_{\rm O, \, TX} \Gamma\_{\rm O, \, T \text{size}})} \frac{1}{(1 - \Gamma\_{\rm O, \, RX})} \tag{2}$$

where *ZFL*,*RX* is the impedance of the receiving antenna feedline and *Z*<sup>0</sup> is the intrinsic impedance in a free space. For the transmitter, Γ*0,TX*, and Γ*0,Tissue* indicate the reflection coefficient of the transmitting antenna and human tissue, respectively. For the receiver, Γ*0,RX*, represents the reflection coefficient of the receiving antenna. It was assumed that the transmitting and the receiving antenna are perfectly matched to the source and the load, respectively. The detailed procedure to derive the impedance matching term *C* is provided in Appendix A. ி,ோ *Г Г Г*

1 ൫1– ,்,்௦௦௨൯

1 ൫1– ,ோ൯

=

ி,ோ 

**Figure 3.** Description of the two-port network for the wireless communication link between the transmitting implanted antenna placed inside the human skin and the receiving antenna outside the human tissue.

The computer program in [18] was developed in order to compute the antenna coupling between two antennas, which allows for flexibility in terms of the relative axis, orientation, and arbitrary antenna movement. Furthermore, the sampling frequency has been introduced to adjust the size of the integration for sufficient convergence. The sampling frequency presented in [18] is

$$f\_s \underset{\overline{f}}{=} 2\kappa \times (D\_{TX} + D\_{RX}) \tag{3}$$

 where *DTx* and *DRx* are the diameters of the transmitting and the receiving antennas, respectively, and *κ* is termed the oversampling ratio. One can use the constant *κ* in order to change the sampling frequency. Spectrum integration is confined to the solid angle *α* subtended by the diameter *DTx* and *DRx*. This confinement is an important requirement in reducing the necessary computational resources. The recent advances in computer capabilities have allowed for the testing of the converged results using different solid angles. It was demonstrated in [19] that the utilization of a larger solid angle *α* is beneficial for offering a degree of improvement in terms of the converged results. The upper and lower bounds for the effective separation distance can be defined as

$$\frac{D\_{TX} + D\_{RX}}{2} < R < \frac{\left(D\_{TX} + D\_{RX}\right)^2}{\lambda} \tag{4}$$

 ሺ௫ <sup>௬</sup> The integral form of coupling formula was implemented using the double summation at the sampling points (*kx*-*ky*). The summation form of the formula makes it possible to implement through the computer program presented in [18]. The summation form of the formula is presented as

$$\frac{b\_0'}{a\_0}(\vec{\tilde{R}}) = -\frac{\mathbb{C}}{k}(\Delta k)^2 \times \sum\_m \sum\_n \frac{\stackrel{\rightarrow}{f}\int\_{TX}(k\_x^{mn}, k\_y^{mn}) \cdot \stackrel{\rightarrow}{g}\_{RX}(k\_x^{mn}, k\_y^{mn})}{k\_z^{mn}} \times e^{i\stackrel{\rightarrow}{k}\stackrel{\rightarrow}{\cdot}}\tag{5}$$

 = ෝ <sup>௫</sup> + ෝ <sup>௬</sup> + ෝ<sup>௭</sup> Δ = ଶగ ೞ ሺାೃሻ where *k* = *x k* ˆ *mn <sup>x</sup>* + *y k* ˆ *mn <sup>y</sup>* + *z k*ˆ *mn <sup>z</sup>* and ∆*k* = <sup>2</sup>*<sup>π</sup> fs* = *π κ*(*DTX*+*DRX*) . The index *m*, *n* for the double summation can be defined as

$$1 \le m\_\prime \; n \le \frac{2\left(D\_{TX} + D\_{RX}\right)^2}{\lambda R} \tag{6}$$

Note that *κ* = 4 is used for the index *m*, *n*. The upper bound of the effective range is restricted to compute the antenna coupling in the far-field region. The complete form of the Friis formula, including the reflection coefficient between the implanted antenna and the human skin, can be defined as

$$\left|\frac{\left|\boldsymbol{\theta}\_{0}\right|^{2}}{a\_{0}}\right|^{2} = \left(\frac{\lambda}{4\pi R}\right)^{2} G\_{TX}(\boldsymbol{\theta},\boldsymbol{\phi}) G\_{RX}(\boldsymbol{\theta},\boldsymbol{\phi}) \times \left|\boldsymbol{\uptheta}\_{TX}\boldsymbol{\uptheta}\_{RX}\right|^{2} \left(1 - \left|\boldsymbol{\upGamma}\_{0,TX}\boldsymbol{\upGamma}\_{0,T\text{size}}\right|^{2}\right) \left(1 - \left|\boldsymbol{\upGamma}\_{0,RX}\right|^{2}\right) \tag{7}$$

where *GTX*(*θ*, *φ*) and *GRX*(*θ*, *φ*) represent the far-field gain of the transmitting antenna and the receiving antenna, respectively, and *ρ*ˆ*TX* and *ρ*ˆ*RX* are the polarization vectors of the transmitting antenna and the receiving antenna, respectively. It was assumed that both the transmitting and receiving antennas are connected to the identical feed line, and that there was no reflection from the source and the load.

#### *2.2. Electromagnetic Characterization of Implanted Antennas*

In order to establish a reliable communication link, implanted antennas need to be investigated in terms of impedance matching the characteristics present inside the human body. The characteristics of the wave impedance vary according to the different kinds of antennas. The human body typically has a large dielectric constant, causing low impedance characteristics. Therefore, the use of an antenna with lower wave impedance is beneficial in order to provide the best matching characteristics.

The wave impedance of the electric dipole *ZE*(*r*) and the one of the magnetic dipole *ZH*(*r*) can be derived from the ratio between the electric and magnetic field of each antenna [22,23], given by

$$Z\_{\rm E}(r) = \eta\_0 \left| \frac{\left[1 + \frac{1}{j\beta r} + \frac{1}{(j\beta r)^2}\right]}{\left(1 + \frac{1}{j\beta r}\right)} \right| \tag{8}$$

$$Z\_H(r) = \eta\_0 \left| \frac{\left(1 + \frac{1}{j\beta r}\right)}{\left[1 + \frac{1}{j\beta r} + \frac{1}{\left(j\beta r\right)^2}\right]}\right|\tag{9}$$

where *η*<sup>0</sup> and *β* represent the characteristic impedance of the free space and the propagation constant of the free space, respectively. Figure 4 shows the comparison of the wave impedance of the electric dipole and the magnetic dipole. As previously discussed, the magnetic dipole provides low impedance characteristics, while the electric dipole shows an opposite trend in terms of impedance characteristics. The impedance inside the human body can be derived from taking into account the different kinds of electrical properties of the human body using the relationship

$$Z = \sqrt{\frac{\mu}{\varepsilon}} = \sqrt{\frac{1}{\varepsilon\_r}} \sqrt{\frac{\mu\_0}{\varepsilon\_0}} = \sqrt{\frac{1}{\varepsilon\_r}} \eta\_0 \tag{10}$$

where *µ*<sup>0</sup> and *ε*<sup>0</sup> are the free space permeability and permittivity, respectively, and *ε<sup>r</sup>* is the relative permittivity which can be determined by the electrical properties of the human tissue. The reflection coefficient between antenna and human tissue can be derived using

$$
\Gamma\_{0,\text{ Tissue}} = \frac{Z\_A - Z\_{\text{Tissue}}}{Z\_A + Z\_{\text{Tissue}}} \tag{11}
$$

where *Z<sup>A</sup>* and *ZTissue* are the impedance of the antenna and the human tissue, respectively.

**Figure 4.** Wave impedance of the electric dipole and the magnetic dipole antenna.

#### **3. Results**

In this section, the complete wireless communication link is investigated including calculation of the wireless link and analysis of the matching characteristics inside and outside the human body. It was assumed that the TX antenna is located inside the skin, while the RX antenna is placed outside the human body to establish communication links for the pacemaker system. All evaluations were performed at the center frequency of 400 MHz.

#### *3.1. Calculation of the Link between Two Antennas*

The coupling formula was used to evaluate the link between the two antennas. This section contains the evaluations for: (1) the free space link, and (2) the link between the two antennas placed inside and outside the human body. The computed results are compared with the results obtained using a full-wave simulation FEKO. The calculated results show an excellent level of agreement with the results from the full-wave simulation FEKO.

#### 3.1.1. Antenna Design Used for the Link Analysis

 tan tan − − The microstrip patch antennas are designed to operate inside and outside the human body. The size of the patch antenna outside the human body was (L<sup>1</sup> × W1) = (40 cm × 40 cm), while the patch antenna inside the human body was smaller at (L<sup>2</sup> × W2) = (20 cm × 20 cm) in order to provide the best matching characteristics inside the human body. The patch antenna outside the human body was designed on a 3.2 mm thick substrate with *ε<sup>r</sup>* = 2.2 tan *δ* (loss tangent) = 0.0009. For the patch antenna inside the human body, the thickness of the substrate was set as 6.4 mm and the microstrip antenna was printed in the middle of the substrate with *εr* = 10.2 and tan *δ* = 0.0025. The size of the skin tissue was set as (L<sup>3</sup> × W3) = (25 cm × 25 cm), with a thickness of 4 mm above the patch antenna. The characteristics of impedance matching for the patch antennas are shown in Figure 5a. The patch antennas placed inside and outside the human body provide reasonably good matching characteristics of less than −10 dB around 400 MHz, however, the characteristics of the implanted patch deteriorate when it operates in air. Figure 5b shows a comparison of the radiation patterns in diverse environments. The patch antenna placed outside the human body has a maximum gain of 6.8 dB, while the other one placed inside the human body has a maximum gain of −7.65 dB. The patch antennas were used to evaluate the free-space link and the link between the implanted antenna and the outside receiving antenna.

**Figure 5.** (**a**) Impedance matching characteristics of the patch antennas used for the link analysis and (**b**) the radiation patterns of the patch antennas at 400 MHz.

#### 3.1.2. Free-Space Link Analysis

Prior to estimating the link, including the implanted antenna, the link was evaluated using the two microstrip patch antennas placed in a free space. The coupling formula was used to compute the coupling versus the separation distance and the coupling versus the transverse displacement [16–18]. In this section, the coupling formula is applied to both of the antenna displacements.

 For the coupling versus separation distance, the antenna coupling was evaluated by moving the RX patch antenna in terms of the *R* direction. For the coupling versus transverse displacement, the receiving antenna was located at different transverse displacements, and the antenna coupling was evaluated. The antenna coupling for the first scheme is shown in Figure 6a. It is shown that the coupling formula provides the coupling level which is roughly 1.5 dB lower than the one of the Friis formula in the closest distance. This is attributed to the gain reduction effect in the near-field region. The computed results show a good agreement with the full-wave simulation FEKO with a deviation less than 0.6 dB. The second scheme was evaluated as shown in Figure 6b. The coupling was evaluated for the transverse displacements of 0.5 and 0.75 *λ*. There is a significant difference in the coupling level for different transverse displacements in close proximity, while all curves converge to a coupling level in the far-field region. This is because the offset geometry in close proximity is more influential in determining mutual coupling. There was a 1 dB deviation between the computed and the simulated results. 

**Figure 6.** Free-space link analysis for (**a**) the coupling in terms of separation distance normalized with the wavelength at 400 MHz and (**b**) the coupling in terms of transverse displacement normalized with the wavelength at 400 MHz.

#### 3.1.3. Link between the Two Antennas Inside and Outside the Human Body

The link between two patch antennas placed inside and outside the human body was evaluated. The patch antenna embedded in a piece of human tissue was used to represent the implanted antenna inside the human skin. It was reported in [8], that the simplified model provides similar characteristics to those of the antenna implanted inside the large human body model. This will help reduce the required computational resources. The attenuation due to the propagation inside the human body is simply characterized through utilizing the difference between the maximum antenna gain of the implanted transmitting patch and that of the receiving patch. The antenna coupling was evaluated for the first scheme with the inclusion of a rotational one, as shown in Figure 7a. For an accurate estimation, the rotation angles are restricted to the angles which are smaller than *θt* = 30◦ [17]. Therefore, the coupling curves were obtained at the rotation angles *θ<sup>t</sup>* = 0◦ and *θ<sup>t</sup>* = 30◦ . It can be observed that the deviation between the computed and simulated results is less than 0.8 dB for both scenarios. The difference between the two coupling curves agrees well with the one of the radiation pattern at the different rotation angles. Coupling curves for the second scheme are shown in Figure 7b. The deviation between the computed and simulated results is 1.2 dB, which is slightly higher than that of the free-space link. ௧ <sup>௧</sup> <sup>௧</sup>

**Figure 7.** Link between the two antennas inside and outside the human body for (**a**) the coupling in terms of separation distance, normalized with the wavelength at 400 MHz and (**b**) the coupling in terms of transverse displacement, normalized with the wavelength at 400 MHz.

#### *3.2. Matching Characteristics Inside and Outside the Human Body*

The impedance matching inside the human body was investigated based on the characteristics of the implanted antennas. Owing to the electrical properties of the human body, the antenna design needs to compensate for the differences between free space and the human body. To investigate the characteristics of the antenna inside the body, we evaluated the design of a helix antenna on PEC (Perfect Electric Conductor) ground and a dipole antenna mounted on an EBG (Electromagnetic Band Gap) structure, which are representative examples of magnetic and electric dipole antennas, respectively. In particular, a small helix antenna is favorable—due to its miniaturized design—for implantation inside the human body. The wave impedance of the representative antennas was obtained using a full-wave simulation FEKO. The simulated results were compared with the theoretical results discussed in Section 2.2. As expected, the curve of the helix antenna resembles that of the magnetic dipole, while the curve of the dipole antenna is similar to that of the electric dipole. In particular, the helix antenna provides low impedance characteristics in close proximity. This corresponds to the characteristics of the magnetic dipole, and this antenna will possess optimal values that are analogous to the human body. Using Equation (10), the impedances of different parts of the body were calculated. The computed

results are provided in Table 2. It can be seen that the impedance of the human body is lower than the characteristic impedance of the free space. Based on the impedance of the skin and the wave impedance of the antenna, the matching characteristics were acquired from Equation (11). Figure 8 shows the reflection coefficient in terms of the different wave impedance of the antenna. The best matching characteristics are obtained when the impedance of the antenna is similar to that of the tissue. The matching characteristics were evaluated through using representative antennas, such as magnetic dipole and electric dipole antennas. A small helical antenna on the PEC was selected as an example of the magnetic dipole antenna, while a dipole antenna on the EBG structure was chosen as an example of the electric dipole antenna. The configuration of the representative antennas is shown in Figure 9. For the helical antenna, the radius and the height were set as 2.1 and 9 mm, and the height from the PEC to the center of the helical antenna (H1) was designed as 3.4 mm. The size of the PEC was set as (L<sup>4</sup> × W4) = (5.4 cm × 5.4 cm). The dipole-EBG was designed and scaled based on a previous study [24]. One difference is that the miniaturized 6 × 6 EBG structure was used in this study. The size of the EBG structure was set as (L<sup>5</sup> × W5) = (9.7 cm × 9.7 cm). ×

**Table 2.** Impedances of the different parts of the human body.


**Figure 8.** Matching characteristics of an antenna inside the different parts of the human body.

**Figure 9.** Configuration of the representative antennas: (**a**) helical antenna mounted on PEC and (**b**) dipole antenna mounted on EBG structure.

The height from the EBG structure to dipole antenna (H2) was set as 2.2 mm. The helical antenna on the PEC corresponds to the magnetic dipole on the PEC, while the dipole antenna on the EBG structure represents the electric dipole on the PMC. The wave impedance of the two antennas was investigated in order to evaluate the matching characteristics inside the human body. Figure 10 shows the simulated results of the wave impedance and the calculated wave impedance of the electric and magnetic dipole antenna. It can be seen that the wave impedance of the helical antenna is similar to that of the magnetic dipole antenna, while the one of the dipole antenna resembles the trend of the electric dipole antenna. In particular, within close proximity, the magnetic dipole possesses low impedance, and the electric dipole has high impedance, when compared to the characteristics of impedance in air. The matching characteristics of the two kinds of antennas were investigated by placing the antennas inside the human body (the skin tissue). Figure 11 shows the simulated impedance characteristics inside and outside the skin tissue. As predicted, the helical antenna provides the best matching characteristics inside the skin tissue, while it shows the deteriorated one outside the skin tissue. In contrast, the electric dipole exhibits slightly degraded matching characteristics inside the skin tissue when compared to those outside the body. It was demonstrated that the magnetic dipole type antenna is advantageous for providing the best matching characteristics inside the human body.

**Figure 10.** Wave impedance of the dipole on EBG and helix antenna on PEC, and a comparison to the calculated wave impedances of the dipole and the helix antenna.

**Figure 11.** Simulated impedance matching characteristics inside and outside the body (the skin tissue).

*ε δ*

*ε δ*

− Ω

− Ω

Ω −

Ω −

−

−

#### **4. Measurements**

Experiments were conducted to verify the calculated and simulated results of the wireless link. After the experiment to verify the impedance matching performance of the patch antenna inside and outside the human body, the coupling between the antennas was measured by changing the separation distance. The MS46522B model vector network analyzer(VNA) from the company, Anritsu, was used for the experiment. Figure 12a shows the patch antenna outside the human body, fabricated for operating at a frequency of 400 MHz. The patch antenna was fabricated on a Taconic TLY-5 substrate (*ε<sup>r</sup>* = 2.2, tan*δ* = 0.0009) with a thickness of 3.2 mm, and the size of the ground and the conducting patch were 40 cm × 40 cm and 24.82 cm × 24.82 cm, respectively. The reader patch was fed from the SubMiniature version A(SMA) connector from the company, WithWave, and the feed position was located 4 cm from the patch center to improve the impedance matching. The fabricated implant patch antenna operates with the performance of <sup>Z</sup><sup>11</sup> = 55.7 − *<sup>j</sup>*7.07 <sup>Ω</sup> and S<sup>11</sup> = −22.3 dB at 401 MHz, which is the result of a 1 MHz up shift compared to the center frequency of the simulated result. Nevertheless, the patch antenna outside the human body operates with excellent performance with Z<sup>11</sup> = 53.5 + *<sup>j</sup>*28.7 <sup>Ω</sup> and <sup>S</sup><sup>11</sup> <sup>=</sup> −12 dB at 400 MHz, and the impedance matching characteristic matches well with the simulated result, as shown in Figure 12b,c.

**Figure 12.** Patch antenna outside the human body: (**a**) fabricated structure, (**b**) input impedance Z11, and (**c**) reflection coefficient S11.

*ε σ* λ The measurement of the implant patch antenna was conducted both in skin tissue liquid (corresponding to inside the human body) and in free-space (corresponding to outside the human body). For the human body, 10 L of Skin Tissue Simulating Liquid (SKIN350-500V2) from Schmid & Partner Engineering AG company (Zürich, Switzerland) was used as shown in Figure 13a. The relative permittivity and conductivity values are *ε<sup>r</sup>* = 46.4 and *σ* = 0.67 at room temperature (22 ◦C) and a frequency of 400 MHz, which are specified in the data sheet. Figure 13b shows the 10 L acrylic tank used to contain the skin tissue liquid. The size of the acrylic tank was made to be 31.6 cm × 31.6 cm × 10 cm considering the size of the implant patch antenna, and the wall thickness was set to 8 mm, considering the relative density of the skin tissue liquid (1.2–1.4 kg/L). It is worth noting that the 8mm of the wall thickness corresponds to about 0.01 λ, so its effect on the radiation performance of the implant patch antenna is negligible. The antenna was assumed to be located 4 mm from the surface of the acrylic tank, and M5 size polycarbonate (PC) screw hole structures were added to stably mount the implant patch.

*ε δ*

*σ*

**Input Impedance (Ohm)**

**360 380 400 420 440**

**Frequency (MHz)**

 **Measurement: Re(Z11) Measurement: Im(Z11) Simulation: Re(Z11) Simulation: Im(Z11)**

*ε*

λ

**360 380 400 420 440**

 **Measurement Simulation**

**Frequency (MHz)**

**-25 -20 -15 -10 -5**

**Reflection Coefficient (dB)**

**0**

**Figure 13.** Implementation of the human body (**a**) skin tissue liquid and (**b**) acrylic tank.

*ε δ* The patch antenna inside the human body consisted of a dielectric substrate (metal free) on the upper layer and a conducting patch with a grounded dielectric slab on the lower layer. Figure 14a shows the lower layer of the patch antenna fabricated for an operating frequency of 400 MHz inside the skin tissue. The patch antenna was designed on Taconic RF-10 substrate (*ε<sup>r</sup>* = 10.2, tan*δ* = 0.0025) with a thickness of 3.2 mm, and the sizes of the ground and the conducting patch were 20 cm × 20 cm and 10.8 cm × 10.8 cm, respectively. The feed position of the implant patch was located 5.2 cm from the patch center to improve the impedance matching. Figure 14b shows the structure in which the upper layer and the lower layer were assembled. Both layers were assembled with M5 size PC screws, and were mounted through screw holes in the acrylic tank. To prevent the occurrence of air gaps and leakage of skin tissue, waterproof tape was firmly attached to the four corners of the assembed structure. Finally, the assembled structure was fixed by screwing into an acrylic tank containing skin tissue liquid, as shown in Figure 14c. Figure 14d–g shows the measured results of the fabricated implant patch antenna. Similar to the simulation, the implant patch antenna operated at 400 MHz in the skin tissue, and operated near 420 MHz in the air, due to the decrease in relative permittivity. The measured input impedance and reflection coefficients were Z<sup>11</sup> = 44.4 + *<sup>j</sup>*11.2 <sup>Ω</sup> and S<sup>11</sup> <sup>=</sup> −17.5 dB at a 400 MHz frequency in the skin tissue, as shown in Figure 14d and e. The frequency downshift occurred in the measured results due to a small air gap by soldering the SMA connector.

The measurements were conducted to verify the coupling formula using the fabricated patch antennas. Figure 15a shows the measurement setup of the coupling between one patch antenna inside the phantom fluid tank and the other one in air. The two antennas were set to face each other in the broadside direction, and the coupling was measured from S<sup>21</sup> of the VNA as a separation distance R changes from 0.5–5λ (correspond to 37.5–375 cm). The measured coupling S<sup>21</sup> shows a similar tendency to the results calculated from the coupling formula and full-wave simulation, as shown in Figure 15b. The ground effect caused by the floor generated at the far separation distance was minimized by the installed microwave absorber. However, at a near separation distance, a measurement error of 0.4 dB occurred due to the reflected wave from the table used in the measurement.

**Figure 14.** Patch antenna inside the human body: (**a**) lower layer of fabricated structure, (**b**) assembly of the fabricated structure, (**c**) antenna measurement setup in skin tissue liquid, (**d**) input impedance Z<sup>11</sup> in skin tissue liquid, (**e**) reflection coefficient S<sup>11</sup> in skin tissue liquid, (**f**) input impedance Z<sup>11</sup> in free space, and (**g**) reflection coefficient S<sup>11</sup> in free space.

**Figure 15.** Measurement of coupling between the reader patch and the implant patch antennas (**a**) measurement setup and (**b**) measured result.

#### **5. Discussion**

The coupling formula is advantageous since it enables us to compute the near-field antenna coupling based on the far-field radiation pattern. The coupling formula was applied to compute the wireless link between two antennas inside and outside of the human body. The patch antenna inside the small part of the human tissue was used to characterize the far-field pattern of the implanted antenna. The computed results obtained from the coupling formula show good agreement with the full-wave simulation FEKO and measurements. The deviation between the computed results and full-wave simulation was less than 0.6–1.2 dB for both cases versus separation distance and transverse displacement. The matching characteristics inside the human body was investigated in terms of the electric and magnetic dipole antennas. The representative examples of the electric and magnetic dipole antennas were selected as the dipole antenna on the EBG structure and the helical antenna on the metal ground plane, respectively. It was found that the magnetic dipole provides low impedance characteristics which are similar to those of the human body, which is advantageous in terms of providing optimal impedance matching inside the human body. The indoor measurement was performed using one patch antenna inside the phantom fluid and the other one in air. The measured results show a good level of agreement with the simulated results in terms of matching characteristics and link performance. This study provides an important guideline for the creation of reliable wireless links based on an accurate numerical method and antenna design in terms of matching characteristics.

**Author Contributions:** Conceptualization, J.-H.L. and I.K.; methodology, I.K.; software, I.K.; validation, S.-G.L. and Y.-H.N.; formal analysis, I.K.; investigation, I.K.; resources, I.K.; data curation, I.K.; writing—original draft preparation, I.K. and S.-G.L.; writing—review and editing, S.-G.L. and Y.-H.N.; visualization, I.K.; supervision, J.-H.L.; project administration, J.-H.L.; funding acquisition, J.-H.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research of three of the authors, S.-G.L., Y.-H.N. and J.-H.L., were supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education (No. 2015R1A6A1A03031833).

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank to Ahmed Akgiray for his initial support in the development of the coupling program.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

The impedance mismatch of an antenna inside the human body was proposed in order to estimate an accurate coupling quotient. In this Appendix, the derivation of the antenna mismatch term is presented in detail. Figure A1 shows the two-port network, linked between the transmitting antenna and the receiving antenna. Except for the propagation part, the coupling quotient between the transmitting antenna and the receiving antenna is derived, which is identical to the impedance mismatching term. One can start with the mismatch term of transmitting antenna. The mismatch term between the TX antenna and the human tissue can be defined as

$$a\_{TX} = \Gamma\_{0, \text{Tissue}} b\_{TX} \tag{A1}$$

$$a'\_{TX} = \Gamma\_{0,TX} b'\_{TX} + a\_0 \tag{A2}$$

் ′்

் = ,்௦௦௨ ்

′் = ,் ′் +

**Figure A1.** Description of the two-port network used for the derivation of the impedance mismatching term.

் =

′் ் Substituting Equation (A1) into Equation (A2) based on the relationship *a* ′ *TX* = *bTX* and *aTX* = *b* ′ *TX* will produce

$$b\_{TX} = \frac{a\_0}{1 - \Gamma\_{0,\,\,TX} \,\Gamma\_{0,\,\,Tissue}}\tag{A3}$$

1−,் ,்௦௦௨ For the receiving antenna, the amplitude of the received wave can be derived using the following relationship

$$b\_{TX} - b\_{TX} \Gamma\_{0, \, RX} = b\_0' \tag{A4}$$

் − ்,ோ = ′ Substituting Equation (A3) into Equation (A4), the coupling quotient can be derived as

$$\frac{b\_0'}{a\_0} = \left(\frac{1}{1 - \Gamma\_{0,\text{ Tissue}}}\right) \frac{1}{1 - \Gamma\_{0,\text{ RX}}}\tag{A5}$$

′ = ቆ <sup>1</sup> 1−,்௦௦௨ <sup>ቇ</sup> 1 1−,ோ It was assumed that both the transmitting and receiving antennas are fed by an identical waveguide. Note that multiple reflection is ignored in the derivation. The Equation (A5) was applied to compute the coupling quotient presented in Equation (1).

#### **References**


**Xiaohang Li 1,\*, Wenfei Yin <sup>2</sup> and Salam Khamas <sup>1</sup>**


**Abstract:** A slot fed terahertz dielectric resonator antenna driven by an optimized photomixer is proposed, and the interaction of the laser and photomixer is studied. It is demonstrated that in a continuous wave terahertz photomixing scheme, the generated THz power is proportional to the 4th power of the surface electric field of photocondutive layer. Consequently, the optical to THz conversion efficiency of the proposed photomixer has an enhancement factor of 487. This is due to the fact that the surface electric field of the proposed photomixer with a 2D-Photonic Crystal (PhC) superstrate has been improved from 2.1 to 9.9 V/m, which represents a substantial improvement. Moreover, the electrically thick Gallium-Arsenide (GaAs) supporting substrate of the device has been truncated to create a dielectric resonator antenna (DRA) that offers a typical radiation efficiency of more than 90%. By employing a traditional coplanar strip (CPS) biasing network, the matching efficiency has been improved to 24.4%. Therefore, the total efficiency has been considerably improved due to the enhancements in the laser-to-THz conversion, as well as radiation and matching efficiencies. Further, the antenna gain has been improved to 9dBi at the presence of GaAs superstrate. Numerical comparisons show that the proposed antenna can achieve a high gain with relatively smaller dimensions compared with traditional THz antenna with Si lens.

**Keywords:** photomixer; terahertz source; two dimensional photonic crystal; frequency selective surface superstrate; terahertz antenna; dielectric resonator antenna

#### **1. Introduction**

Terahertz (THz) spectrum extends from 300 GHz to 10 THz, which covers the frequency range between mm-wave and infra-red bands. In addition, the corresponding wavelengths represent the transition between photonics and electronics. Higher attention has been paid to the development of THz technologies, owing to the variety of THz spectrum potential applications including monitoring and spectroscopy in pharmaceutical industry [1,2], imaging [3,4], material spectroscopy [5], security [6,7], biology and medicine [8,9], and high-speed communication [10].

However, the main limit to the development of THz technologies is the lack of available THz emitters and detectors [11]. To date, most of the THz systems that utilize time domain techniques employs bulky and expensive femtosecond lasers. In this case, the optical excitation from the lasers can generate and detect sub-picosecond electrical pulses. On the other hand, frequency domain techniques can achieve higher resolutions and high scanning speed in a low cost and portable devices [12]. So far, it has been demonstrated by numerous that continuous wave THz sources can either be generated directly or converted up and

**Citation:** Li, X.; Yin, W.; Khamas, S. An Efficient Photomixer Based Slot Fed Terahertz Dielectric Resonator Antenna . *Sensors* **2021**, *21*, 876. https://doi.org/10.3390/s21030876

Academic Editors: Razvan D. Tamas and Shuai Zhang Received: 24 November 2020 Accepted: 25 January 2021 Published: 28 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

down from microwave and optical frequencies, respectively [11]. Nevertheless, there is a lack of efficient room temperature THz sources without the need of cryogenic cooling system and external magnetic field [13].

One of the most promising continuous wave THz sources that utilizes Optical Heterodyne Generation (OHG) [14] is known as photomixer that is capable of generating tunable and coherent THz signals with low-cost and low power consumption in a compact devices [15–18]. A photomixer consists of two set of metal electrodes, a photoconductive layer and a bulky supporting dielectric substrate. In the photomixer, as two interfering laser beams incident on the non-linear photoconductive medium, electrons are excited from valence band to conduction band, hence, spatiotemporal electrons and holes are generated. Owing to the applied DC biasing voltage, induced photocurrent is driven at the beating frequency of two incident laser beams [19]. Typically, a conventional photomixer can only achieve 0.1% optical to THz power conversion efficiency [20,21]. It has been demonstrated earlier that an enhancement factor of 4 in the optical to THz conversion efficiency can be achieved by using plasmonic material as interdigital electrodes [22]. Additionally, more than 4 times of THz power has been achieved by using optical antenna array of ZnO nanorods [23]. Further, an enhancement factor of 25 in terahertz radiation has been demonstrated by utilizing transparent-conducting oxcides nanocylinders between photomixer electrodes [24]. Additionally, double the effective electric energy can be generated by utilizing embedded electrodes [25]. In addition, to date, a highest reported laser to THz conversion efficiency is 7.5%, where replaced the conventional photomixer electrodes have been replaced by three-dimensional plasmonic contact electrodes [26,27]. However, though many researchers have attempted to optimize the optical to THz conversion efficiency, it is still significantly less than tenth of the theoretical maximum of 100% [28].

As mentioned previously, a photomixer is implemented on a photoconductive layer that is supported by a bulky dielectric substrate. The THz antenna's input resistance is expected to be reduced by a factor of √ ((ε<sup>r</sup> + 1)/2) due to the presence of the bulky dielectric supporting layer that has a dielectric constant of εr. On the other hand, the output resistance of a photomixer is in the order of ~10 kΩ [29]. Therefore, the reduced input resistance of the THz antenna leads to a poorer matching efficiency. A full wavelength dipole has been used to drive a Yagi-Uda array to achieve an input resistance of 2.6 kΩ [30]. Moreover, a 3.3 kΩ input resistance has been achieved by implementing an isolating metallic ground plane with a dipole placed on a thin dielectric slab [31].

In addition, it has been reported that THz communications are more likely to be influenced by the atmosphere, especially, the humidity [32]. Therefore, the radiation power enhancement becomes another challenge. In this case, multiple types of Si lenses have been utilized to achieve a higher gain [33,34]. Beyond that, with the presence of a thick supporting dielectric substrate, Si lens can be used to collect and collimate the generated THz power to minimize the power dissipation in the substrate. However, the usage of Si lens makes the entire antenna configurations even larger on the top of the usage of a thick supporting dielectric substrate.

In this study, the optical to THz conversion efficiency of the photomixer has been optimized based on a numerical study and the utilization of a two-dimensional photonic crystal optical frequency selective surface (FSS) superstrate. Then, a dielectric resonator antenna (DRA) with the appealing features of low cost, small size, and high radiation efficiency, as well as gain [35], has been truncated from the bulky dielectric supporting GaAs substrate, in this case, the low temperature grown GaAs, LT-GaAs, to reduce the size of the antenna configuration and enhances the radiation efficiency. Coplanar stripline and THz dielectric superstrate have been implemented for the further optimization of matching efficiency and antenna gain, respectively. The simulations have been conducted using computer simulation technology (CST) microwave studio.

#### **2. Photomixer Design**

In this section, the generated THz power from a photomixer, that is biased using a DC voltage and is illuminated by two laser beams with a difference in their central frequencies in the THz range has been calculated analytically. According to the numerical analysis, the generated THz power can be enhanced substantially by optimizing the photomixer configuration.

#### *2.1. Derivation of the Generated THz Power from the Photomixer*

Figure 1a illustrates two linearly polarized continuous wave laser beams with beating frequency falling in the THz spectrum that are incident on the DC biased photoconductive layer. On the other hand, the structure of a typical photomixer electrodes is presented in Figure 1b. Since the applied biased voltage cannot change the absorption coefficient, mobility and recombination time of the low temperature grown GaAs, LT-GaAs photoconductive layer, the electrodes of the photomixer can be considered as an ohmic conductances while the time varying source conductance is electrically modeled as a photoconductance. Consequently, the photomixer based THz antenna can be modeled using the equivalent circuit illustrated in Figure 2 from which it can be noted that the photomixer consists of a photoconductance, *G<sup>s</sup>* −1 *(*Ω*,t)*, and a paralleled capacitance, *Celectrodes* . The capacitance depends on the structure of the photomixer and the dielectric constant of the photoconductive layer. The generated photocurrent is driven by the biased voltage excites and the THz radiating antenna, which is represented in Figure 2 by a resistance of *Rantenna* [36].

**Figure 1.** (**a**) Typical photoconductive photomxing scheme; (**b**) Top view of photomixer electrodes.

**Figure 2.** Equivalent circuit of photomixer based THz antenna.

The electric fields of the two incident laser beams on the LT-GaAs' surface can be expressed as:

$$E = |E\_i|e^{j\omega\_l t} \tag{1}$$

where *ω* is the lasers' angular frequency and *I* = 1,2, represents laser 1 and laser 2, respectively. The laser intensity been absorbed by the LT-GaAs is proportional to the square of the total incident electric field on the LT-GaAs' surface:

$$I(\Omega, t) = (1 - \Gamma) \sum\_{i} |E|^2 = I\_0 (1 - \Gamma) [1 + 2 \frac{\sqrt{m I\_1 I\_2}}{I\_0} \cos(\Omega t)] \tag{2}$$

where *I*<sup>0</sup> is the maximum optical intensity on the LT-GaAs' surface, Γ is the reflection coefficient at the LT-GaAs-air interface, *m* describes the overlap of the laser beams, which is known as the mixing efficiency and <sup>Ω</sup> is angular beat frequency, (*ω*1−*ω*2).

The induced photo-carriers generated from the incident laser beams as a function of time is:

$$\frac{dn(t)}{dt} = -\frac{n(t)}{\tau\_c} + \frac{n(T)}{hf\_l} I(\Omega, t) \tag{3}$$

in which *h* is the Plank's constant, *f<sup>l</sup>* is the mean frequency of the laser beams and *τ<sup>c</sup>* is the carrier lifetime. In addition, *α(T)* is the temperature-dependent absorption coefficient and *T* is the temperature in kelvin. For a GaAs layer with a direct band gap, *α(T)* can be expressed as [37]:

$$\mathfrak{a}(T) \approx \mathcal{K}\_{\text{abs}} \sqrt{\frac{hf\_l - E\_{\mathcal{S}}(T)}{q}} \tag{4}$$

where *<sup>K</sup>abs* is a certain frequency-independent constant which is approximately 9.7 <sup>×</sup> <sup>10</sup><sup>15</sup> for GaAs [37], and *Eg(T)* is the LT-GaAs' temperature dependent band gap energy defined as:

$$E\_{\mathcal{S}}(T) = E\_{\mathcal{S}}(0) - \frac{\mathfrak{a}\_E T^2}{T + \mathfrak{P}\_E} \tag{5}$$

in which *Eg*(0) is the GaAs' gap energy at 0 ◦K which is about 1.519 eV, *α<sup>E</sup>* and *β<sup>E</sup>* are material constants of GaAs which are approximately 5.41 <sup>×</sup> <sup>10</sup>−<sup>4</sup> eV/K and 204 K, respectively [38].

By assuming that *I*<sup>1</sup> = *I*<sup>2</sup> = *I*<sup>0</sup> and *t*/*τ<sup>c</sup>* >> 1, then substituting (2) into (3), the generated carrier density can be obtained as:

$$m(\Omega, t) = \frac{a(T)}{hf\_1} l\_0 (1 - \Gamma) \tau\_c (1 + \sqrt{m} \frac{\cos(\Omega t) + \Omega \tau\_c \sin(\Omega t)}{1 + \left(\Omega \tau\_c\right)^2}) \tag{6}$$

The conductance of the photomixer can be expressed as:

$$\mathcal{G}\_{\mathbf{s}}(t) = \int d\mathbf{G}\_{\mathbf{s}}(t) = \int\_{0}^{T\_{\text{sub}}} \sigma(t) e^{-a(T)\mathbf{z}} \frac{\mathcal{W}}{L} dz = \frac{\mathcal{W}}{a(T)L} \sigma(t) \left(1 - e^{-a(T)\mathbf{T}\_{\text{sub}}}\right) \tag{7}$$

in which *Tsub* is the depth of photoconductive region, *W* is the width of the electrode, *L* is the length of the electrode and *σ(t)* is the conductivity. The electrical conductivity is defined as:

$$\sigma(t) = e\mu\_{\varepsilon}n(t) = \frac{a(T)e\mu\_{\varepsilon}}{hf\_{l}}I\_{0}(1-\Gamma)\pi\_{\varepsilon}(1+\sqrt{m}\frac{\cos(\Omega t)+\Omega \pi\_{\varepsilon}\sin(\Omega t)}{1+\left(\Omega \pi\_{\varepsilon}\right)^{2}})\tag{8}$$

where *e* is the electron charge and *µ<sup>e</sup>* is the electron mobility. Therefore, the photomixer's conductance can be derived by substituting (8) into (7):

$$\mathcal{G}\_{\rm s}(\Omega, t) = \frac{\mathcal{W}e\mu\_{\rm c}I\_{0}\tau\_{\rm c}}{hLf\_{\rm l}}(1 - \Gamma)(1 - \mathbf{e}^{-\mathbf{a}(\Gamma)\mathbf{T}\_{\rm sub}})(1 + \sqrt{m}\frac{\cos(\Omega t) + \Omega \tau\_{\rm c}\sin(\Omega t)}{1 + \left(\Omega \tau\_{\rm c}\right)^{2}}) \tag{9}$$

The impedance of the system can be given by analyzing the equivalent circuit shown in the Figure 3:

$$Z\_t(\Omega, t) = \frac{1}{j\Omega \mathbb{C}\_{electroducts} + \mathbb{G}\_s(\Omega, t)} + R\_{antenna} \tag{10}$$

and the radiation power can be defined as:

$$P\_{THz}(\Omega, t) = R\_{antema}(\frac{V\_{biased}}{Z\_t(\Omega, t)})^2 \tag{11}$$

**Figure 3.** (**a**) Configurations of the 2D-PhC unit cell; (**b**) Top view of the photomixer based slot.

Therefore, as *RantennaG<sup>s</sup>* is much smaller than 1 and by replacing system impedance by (8), as well as neglecting the imaginary part, the radiation power can be expressed as:

$$P\_{THz}(\Omega, t) \approx R\_{antenna} \frac{V\_{biased}^2 G\_s^2(\Omega, t)}{1 + (\Omega R\_{antenna} C\_{electrodes})^2} \tag{12}$$

In addition, by employing (9) and averaging the power, the mean generated THz power can be expressed as:

$$P\_{\rm THz} \approx \left[\frac{\mathcal{W}e\mu\_{\rm \varepsilon}\tau\_{\rm c}}{hLf\_{\rm l}}(1-\Gamma)(1-e^{-\mathfrak{a}(\Gamma)\mathcal{T}\_{\rm sub}})\right]^2 \frac{mR\_{\rm antenna}V\_{\rm biased}^2}{[1+(\Omega R\_{\rm antenna}\mathcal{C}\_{\rm electrons})^2][1+(\Omega \tau\_{\rm c})^2]})I\_0^2 \tag{13}$$

It can be noted from (13) that the generated THz power depends on three main factors, (Ω *Rantenna Celectrode* ), (Ω*τc*) and *I<sup>0</sup>* 2 . As mentioned previously, *Rantenna* and *Celectrodes* depend on the photomixer's configuration and dielectric constant of the photoconductive layer. In addition the carrier lifetime is a function of the applied bias voltage [38–40]. Hence, it can be demonstrated that for the same photoconductive material and photomixer configuration that are used with the same biased voltage, the generated THz power is proportional to the square of the incident laser intensity on the surface of the LT-GaAs layer. Since the intensity is proportional to the square of electric field, the generated THz power is proportional to the 4th power of the electric field at the surface of LT-GaAs. Consequently, a design that optimizes the laser intensity is proposed instead of manipulating the photoconductive material and photomixer electrodes' configuration.

#### *2.2. Photomixer Modeling*

In order to optimize the electric field on the surface of LT-GaAs, two dimensional photonic crystal (2D-PhC) has been introduced by utilizing a periodic plane and a nonperiodic third dimension to provide a pass, or stop, band frequency response [41]. The unit cell of the 2D-PhC is illustrated in Figure 3a. The 2D-PhC has been used as an optical

frequency selective surface (FSS) superstrate that is placed at an optimum height above the photomixer. The electromagnetic wave bounces between the FSS and ground plane surrounding the photomixer, therefore, the cavity created by FSS superstrate and ground plane can enhance the optical intensity on the surface of the LT-GaAs.

Since the photomixer is used as a source to excite the truncated GaAs THz DRA, the metallic ground plane has been deployed on top of the LT-GaAs photoconductive layer. However, in order to illuminate the photomixer by the laser beams, a central slot feed is used to accommodates the photomixer as shown in Figure 3b. Moreover, the electrodes have been defined as optical gold (Palik) for CST simulation purposes. The dimensions of the parameters shown in Figure 3b have been defined as: *M* = 0.5 µm, *B* = 0.2 µm, *W* = 0.1 µm, *e* = 0.5 µm, *P* = 2.3 µm, *L* = 1.13 µm, *D* = 0.8 µm, *Tend* = 4 µm, *H* = 1 µm, *G* = 13.68 µm, and thickness of 0.1 µm. It should be noted that the thickness of the LT-GaAs photoconductive layer is 0.44 µm with a relative dielectric constant of 12.9. The material of the 2D-PhC FSS has been assumed as GaAs while the periodicity, central air hole radius, and thickness have been chosen as *a* = 0.76 µm, *r* = 0.3 *a* and *h* = 0.2 *a*, respectively. Figure 4 illustrates the reflectivity of the 2D-PhC FSS, where it can be noted that there is a stopband at wavelength range of 750 to 780 nm.

**Figure 4.** Reflectivity of the 2D-PhC FSS.

A 2D-PhC layer with 19 × 19 unit cells has been suspended at a height of 0.3 µm above the photomixer to act as an FSS superstrate. The incident laser beams have been modeled as a linearly polarized plane wave with a 1 V/m electric field component along the direction of photomixer electrodes. The electric field magnitudes between the central electrode pair on the LT-GaAs' surface is illustrated in Figure 5 with the comparison to that at the absence of 2D-PhC FSS superstrate. In addition, the cross-section of the surface electric field distribution at the central pair of photomixer electrodes are presented in Figure 6. From these results, it can be observed that the utilization of the 2D-PhC superstrate has improved the electric field on the electrodes from 2.1 to 9.9 V/m, which represents an enhancement factor of 4.7. As explained earlier, the generated THz power is proportional to the 4th power of the electric field, therefore the corresponding enhancement factor of the generated THz power is 487. Besides, the same methodology has been applied to an identical photomixer albeit with a InGaAs photoconductive layer, where the electric field on the InGaAs's surface has increased form 2.42 to 11.5 V/m by utilizing an FSS superstrate with unit cell's dimension of *a* = 0.72 µm *r* = 0.27 *a* and *h* = 0.19 *a* at a height of 0.23 µm above the photomixer. The results are presented in Figure 5, where it can be noted that approximately same enhancement factor has been achieved compared to the LT-Gaas photoconductive layer. Compared with the cases of increasing the electrodes E-field from 2.1 to 3.4 V/m and 4.365 V/m using 2D-PhC, with central hole [42] and plasmonic rod [43], respectively, as reflectors underneath the photoconductive layer, employing a

2D-PhC as FSS superstrate has substantially enhanced the optical to THz power conversion efficiency. However, the overall efficiency of the system depends on the laser to THz power conversion efficiency as well as the antenna's radiation and matching efficiencies that will be investigated next. In the following section, the optimized photomixer will be used to excite a THz DRA that is truncated from the supporting bulky GaAs substrate. A DRA has been chosen due to the high radiation efficiency of more than 90% at the frequency range of interest.

**Figure 5.** Optical E-field magnitude between the central electrodes of a photomixer on the surfaces of LT-GaAs and I GaAs photoconductive layers.

**Figure 6.** The optical E-field distribution (**a**) without FSS (**b**) with FSS.

#### **3. THz Dielectric Resonator Antenna Design**

#### *3.1. Antenna Configuration*

The presence of the bulky GaAs supporting substrate reduces the input impedance and absorbs most of the generated THz power, which impairs the matching and radiation efficiencies. A typical radiation efficiency for a dipole above a thick dielectric substrate is 40% or less depending on the thickness and dielectric constant of the substrate [44]. On the other hand, a rectangular dielectric resonator antenna offers a considerable enhance-

ment in the radiation efficiency owing to the absence of surface waves and ohmic losses. Therefore, the bulky GaAs substrate can be truncated to act as a dielectric resonator antenna that operates at the higher order resonance mode. However, the truncated DRA should be large enough to maintain the required physical support to the photomixer device. For fabrication purposes, the width to height aspect ratio of the truncated DRA should be greater than 3. Otherwise, a fragile configuration will be achieved that is difficult to fabricate. Furthermore, the utilization of a DRA as a substrate results in a much smaller configuration compared to traditional structures that are based on utilizing a hemispherical Si lens to extract the THz power. As the configurations of the photomixer and corresponding 2D-PhC FSS superstrate are relatively small enough at the THz spectrum, they will have a negligible impact on the performance at the THz frequency range.

As a result, the GaAs substrate has been employed as the THz antenna that also provides the mechanical support to the device at the same time. The GaAs DRA is illustrated in Figure 7 with dimensions of *WDRA* = 250 µm, *HDRA* = 60 µm, as well as a relative dielectric constant of 12.9, and has been placed on a gold ground plane with a size of *Wground* = *Wsup* = 400 µm. For further gain enhancement, an additional GaAs dielectric superstrate has been employed with dimensions of *Wsup* = 400 µm and *Tsup* = 60 µm. As illustrated in Figure 7, the original optical superstrate has been placed on the feed side of the DRA to capture the illuminating laser beams, while this THz superstrate is placed above the opposite side of the DRA to enhance the radiated THz power. Therefore, the two superstrates will not impact each other as they separated by the DRA and the gold ground plane that accommodates the photomixer. The distance between the DRA and the new THz superstrate can be determined as *Hsup* = (0.25\*((*ϕ*<sup>1</sup> + *ϕ*2)/*π*) + 0.5) *λ* [45], where *ϕ1*, *ϕ<sup>2</sup>* represents the reflection coefficient phases of the superstrate and ground plane. Therefore, the distance between the GaAs DRA and the THz superstrate has been calculated as *Hsup* = 30 µm.

**Figure 7.** THz DRA and superstrate configuration.

DC bias is required to generate the THz power, therefore, the ground plane has been divided into two halves by a narrow slot with a width of Wseperate = 0.5 µm to work as two large DC biasing pads as illustrated in Figure 8. Since the generated THz power can leak through the DC biased pads and transmission line, a coplanar stripline (CPS) network has been employed to work as a choke filter to minimize the THz current leakage, as well as improving the matching. The configuration of the CPS and feeding slot, which accommodates the photomixer and excites the DRA, has been included in Figure 8. The feeding slot has a length of *Lslot* = 65 µm and width of *Wslot* = 5 µm. The dimensions of the CPS network have been chosen as *LTx* = 120 µm, *WTx* = 1 µm, *Lstub* = 91 µm, *Wstub* = 0.5 µm, *gstub* = 50 µm, *Wgap* = 0.5 µm, and *gTx* = 3 µm, respectively.

**Figure 8.** Top view of feeding slot and CPS.

Finally, the feeding photomixer has been modeled as a discrete port with a 10 kΩ input resistance that is in parallel with a 3fF lumped capacitance. Both of the discrete port and lumped capacitance have been deployed at the center of feeding slot in order to be connected with the CPS and DC bias pads.

#### *3.2. Results and Discussion*

The input impedance of the DRA with and without CPS network has been studied as shown in Figure 9, where it can be noted that the input resistance has been improved from 430 to 700 Ω by utilizing the CPS, which corresponding to an enhancement of matching efficiency from 15.8 to 24.5%. Furthermore, the resonance mode of the DRA has been investigated as illustrated in Figure 10, where it can be noted that the TE<sup>711</sup> mode has been excited. The radiation patterns of the DRA are presented in Figure 11, where the broadside gain has been improved from 6.5 to 9 dBi by incorporating the THz GaAs superstrate. As a result, the radiated THz power has been enhanced by a factor of 2. Therefore, the performance of the THz photomixer based antenna has been improved considerably by combining several factors such as the improving the optical to THz power conversion efficiency as well as enhancing the radiation efficiency by utilizing a DRA and employing a CPS that improved the matching efficiency.

**Figure 9.** Input resistance of photomixer based slot fed THz DRA with or without CPS.

**Figure 10.** The H-field component inside the DRA with TE711mode at (**a**) XY plane; (**b**) XZ plane.

**Figure 11.** Radiation pattern of DRA with and without superstrate at (**a**) ϕ = 0 ◦ ; (**b**) ϕ = 90 ◦ .

Table 1 compares the performance of the proposed DRA with published THz antennas and THz DRAs, where it can be observed that presented DRA offers a higher antenna gain as compared with other DRAs though, it is slightly larger than the reported DRAs. Compared with other antenna types, the proposed THz DRA achieves a similar gain with much smaller dimensions. Such a miniaturized high gain antenna can be used to optimize the performance of any THz application with a limited system space.


**Table 1.** Performance summary and comparison with prior works.

#### **4. Conclusions**

The presented work introduces a photomixer based slot fed terahertz dielectric resonator antenna with enhanced optical to THz power conversion, as well as improved matching and radiation efficiencies. The interaction of two incident continuous wave

laser beams and photomixer has been studied. A general analytical expression for the generated THz power has been derived, which demonstrates that the generated THz power is proportional to the 4th power of the electric field on the surface of photoconductive layer. Therefore, by utilizing a 2D-PhC as FSS, the optical to THz conversion efficiency has been improved by a factor of 487. Consequently, the optimized photomixer has been accommodated in a central slot, which has been used to excite the THz DRA that has been truncated from the thick supporting GaAs substrate. A coplanar stripline has been implemented to minimize the leakage of THz power through DC bias pads and transmission line, as well as improving the input resistance of the antenna. As a result, the input resistance of the DRA has been improved from 430 to 700 Ω, which corresponds to 15.8 and 24.4% matching efficiency, respectively. Finally, a THz GaAs superstrate has been employed with the THz DRA, which leads to an enhancement of the antenna gain from 6.5 to 9 dBi. The presented results demonstrate that the proposed design outperforms other counterparts reported in the literature. As demonstrated by (13), further enhancement of the optical to THz conversion efficiency can be achieved by reducing the carrier lifetime of the photoconductive layer such as changing the configuration of the photomixer electrodes to minimize the traveling distance of carriers. The performance can be improved further by altering the shape and dimensions of the slot to increase the order of the excited mode as this has the potential of providing higher gain. In addition the packaging of the proposed configuration needs to be considered so that a physical support is provided to the Thz superstrate in conjunction with improving the handling and stability of the device.

**Author Contributions:** Conceptualization, X.L., W.Y., and S.K.; methodology, X.L. and W.Y.; software, X.L.; validation, X.L.; formal analysis, X.L.; investigation, X.L.; writing—original draft preparation, X.L.; writing—review and editing, S.K.; supervision, S.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article or supplementary material.

**Acknowledgments:** The authors would like to thank Professor Richard Hogg, James Watt School of Engineering, University of Glasgow for the valuable discussions and advice with respect to the practical dimensions of the truncated THz DRA.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

