**Small Antennas for Wearable Sensor Networks: Impact of the Electromagnetic Properties of the Textiles on Antenna Performance**

#### **Gabriela Atanasova 1,\* and Nikolay Atanasov 1,2**


Received: 24 August 2020; Accepted: 8 September 2020; Published: 10 September 2020

**Abstract:** The rapid development of wearable wireless sensor networks (W-WSNs) has created high demand for small and flexible antennas. In this paper, we present small, flexible, low-profile, light-weight all-textile antennas for application in 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 application in W-WSNs is also suggested. Based on the procedure, an antenna on a denim substrate is designed as a benchmark. It demonstrates very small dimensions and a low-profile, all while achieving a bandwidth (|S11| < −6 dB) of 285 MHz from 2.266 to 2.551 GHz, radiation efficiency more than 12% in free space and more than 6% on the phantom. Also, the peak 10 g average SAR is 0.15 W/kg. The performance of the prototype of the proposed antenna was also evaluated using an active test. To investigate the impact of the textile materials on the antenna performance, the antenna geometry was studied on cotton, polyamide-elastane and polyester substrates. It has been observed that the lower the loss tangent of the substrate material, the narrower the bandwidth. Moreover, the higher the loss tangent of the substrate, the lower the radiation efficiency and SAR.

**Keywords:** small antenna; textile antenna; wearable antenna; SAR; flexible antenna; low-profile antenna; sensor network; active test

#### **1. Introduction**

Wearable wireless sensor networks (W-WSNs) can be applied in diverse areas, including health care (clinical diagnostics, rehabilitation), sports (athlete activity profile, energy expenditure during training) and work safety (monitors for the temperature, humidity, CO2) [1,2]. These networks are a particular case where sensors are deployed on the user clothing and/or directly on the body to measure physiological signals of a human and/or to monitor its environment [3–5]. Hence, each W-WSN consists of multiple wearable sensor nodes which are capable of communicating with each other (on-body communications) or with external devices (off-body communications) allowing a connection with a monitoring centre (smartphone, local or cloud webserver). Several frequency bands (such as 2.36–2.38 GHz medical body area network (MBAN), 2.4–2.48 GHz and 5.75–5.82 GHz industrial, scientific, and medical (ISM) bands) and wireless technologies such as IEEE 802.11, IEEE 802.15, LTE, LoRA, etc. are used to connect wearable sensor nodes. These wireless technologies require each wearable sensor node to be equipped with a sensing element, processor, memory, power module, transceiver and an antenna [2,4,6].

The antenna plays a key role in the link performance of each wearable wireless sensor node because it determines the reliability of the wireless link and directly influences the power consumption of the node [2]. Moreover, small, flexible, low-profile, and light-weight wearable antennas based on materials which are deformable, twistable and stretchable are needed because the sensor node needs to be seamlessly worn [7]. Most of the proposed flexible wearable antennas are based on polymers [5,8,9], textiles [1,7,10–17] or flexible ceramics [18].

Among flexible materials, textiles are the most widely employed materials for wearable antennas due to their ease of integration on the clothes. The operating frequency bands and radiation efficiency of such a kind of antenna structure can be controlled by proper selection of the antenna topology, substrate thickness and substrate electromagnetic (EM) properties [1,19,20]. Consequently, design of the all-textile wearable antennas requires precise knowledge of the EM properties of textiles used for the antenna substrate at the frequencies of interest.

The second major challenge when designing wearable antennas is the performance reduction (such as shifting of the resonant frequency, changing of the input impedance, reducing the radiation efficiency of the antenna) caused by the specific environment, in which wearable antennas operate (close to the human body) [5,17,21,22]. Hence, antenna topologies with high body-antenna isolation are needed to guarantee a satisfactory performance in varying operating conditions and to reduce the specific absorption rate (SAR) [2].

In the literature, a diverse range of techniques have been reported for reducing interaction between the antenna and human tissue. One popular technique to reduce electromagnetic coupling between the antenna and human body is to use metamaterials such as electromagnetic bandgap [10–14] and artificial magnetic conducting surfaces [16,17]. Another technique is to use a reflector [1,5,8,15] or a full ground plane [7].

Also an essential aspect would be considered when designing antennas for wearable sensor nodes is miniaturization. Most of the proposed antenna designs still suffer from relatively large size [1,9,14–17] or high profile [5,10–12] and do not meet the requirements (for low-profile and small size) of the antennas for wearable wireless sensor nodes. Consequently, the design of antennas for wearable wireless sensor nodes is a complex task. Generally, in wearable antenna design, electrical, mechanical, and safety requirements described in [2] should be taken into account.

In this paper, we extend our conference paper [1] where the characterization of the EM properties of the regular textiles and study of their effects on performance (radiation efficiency, reflection coefficient magnitude, bandwidth, and maximum gain) of wearable backed antennas have been presented. Now, details about the design, manufacturing and performance measurements of small wearable backed antennas for applications in sensor nodes are provided. The proposed step-by-step methodology allows us to design and experimentally realize new small, low-profile, lightweight and flexible all-textile antennas with high body-antenna isolation. Compared with [1], additional contents of this expansion paper are as follows: (1) a step-by-step design, fabrication and measurement procedure of small wearable backed antennas for application in sensor nodes; (2) new small all-textile antennas for potential integration into everyday clothing; (3) a study (by simulations and measurements) of the antenna performance in free space and on a phantom of the human body.

#### **2. Design, Fabrication and Measurement of Small Antennas for W-WSNs**

Figure 1 shows the generic flowchart of the design, optimization, fabrication and measurement processes of small antennas for W-WSNs used in this work. As shown in Figure 1, the antenna design process starts with the definition of the design goal and target antenna specifications. The design goal of this work is to create new small low-profile all-textile antennas which provide sufficient radiation efficiency and appropriate isolation to the human body for off-body communications in W-WSNs, for potential integration into everyday clothing. The target antenna specifications and required performance associated with this goal are set out below in Table 1.

**Figure 1.** Block diagram of the design, optimization, fabrication and measurement process of small antennas for W-WSNs.


**Table 1.** Target antenna parameters and characteristics.

– – The frequency range pointed in Table 1 is chosen because most of the devices for off-body communications are designed for operation in the unregulated 2.36–2.38 GHz MBAN and 2.4–2.48 ISM frequency bands [3,5,6,10,11,22]. The minimum value of the radiation efficiency pointed in Table 1 is chosen based on a survey of radiation efficiency of internal antennas in real mobile phones [23]. The results from the measurements of the efficiency of the mobile-phone antennas show that average handset radiation efficiency varies between 4.5% and 20%. As expected, the antenna with the smallest size (antenna had a maximum length of 36 mm) has the lowest radiation efficiency (4.5%).

#### *2.1. Initial Antenna Design*

– The initial design of the antenna starts with the choice of an antenna geometry and a material for the substrate. There is no single specific type of geometry for wearable antennas. The most prevalent types are dipole [5,8], monopole [1,10,12,14–17] and patch antennas [7,9,11]. The choice of geometry needs to be guided from the design requirements such as simple structure, small size, low-profile and light-weight. Consequently, the antenna needs to be physically small.

In this work, an antenna geometry based on a loop antenna was chosen. This structure is one of the simplest small radiators and provides good matching with many types of feedings, such as coaxial cables and planar transmission lines.

' Another point to keep in mind during the initial design is the choice of materials for the antenna's elements and characterization of their electromagnetic properties, as shown in Figure 1. When designing antennas for potential integration into everyday clothing, the substrate material is predetermined. This is the textile from which the garment is made and in which the antenna will be embedded. Measurements of the complex permittivity of the chosen textile can be carried out using the resonant or non-resonant methods [1]. Conductive fabrics or threads are the most widely used materials for radiating elements. The information about dc conductivity (or sheet resistance) of the conductive textiles usually is available from the datasheets provided by the manufacturers.

In this study, conventional fabrics with natural fibre (denim and cotton), as well as synthetic fibre (polyester and polyamide-elastane) have been selected as substrates to develop small, low-profile, light-weight all-textile antennas. Conductive fabric has been chosen as a material for the radiating elements and reflector. More details about the EM properties of the selected textiles can be found in our conference paper [1].

The steps to design a small all-textile antenna that meets the specifications presented in Table 1 are depicted in Figure 2. Figure 2a depicts the structure of the proposed antenna in the first step of the design process. It is composed of a rectangular loop, coplanar waveguide (CPW) transmission line, substrate, and a reflector. The choice of the CPW to feed the antenna was because it offers a single-layer manufacturing process. The substrate is denim with a real part of relative permittivity (ε ′ *r* ) 1.878, and a loss tangent (tan δ) 0.0594 at 2.565 GHz. The thickness of the substrate is 1.5 mm (comprised of three-layer denim) with density from 1.54 g/cm<sup>3</sup> [1,2]. This substrate thickness enables us to obtain a small antenna with a low-profile. The reflector was implemented in the antenna's structure to reduce the effect of the human body over the antenna performance and SAR. It was chosen due to its simple form. Also, it is relatively easy for numerical modelling and manufacturing. The geometrical dimensions (length and width) of the loop strips and CPW were tuned to yield a resonance at around 2.47 GHz. Figure 2b shows the reflection coefficient magnitude of the proposed antenna during the design process. In the next step, an arc-shaped parasitic element was added at a close distance to the loop structure (see Figure 2a, Step 2). It is observed that the resonant frequency shifts to a lower frequency while the bandwidth and the impedance matching of the antenna are not changed. Finally, to tune the input resistance of the proposed antenna closer to 50 Ohms and broader bandwidth, four parasitic elements were inserted in the arc-shaped loop (see Figure 2a). This enables a good impedance match with a lower than −6 dB bandwidth (from 2.266 to 2.551 GHz) of approximately 285 MHz and a resonant frequency of 2.4 GHz, as shown in Figure 2b. ′ δ ' −

**Figure 2.** Antenna geometry: (**a**) Design steps; (**b**) Simulated reflection coefficient magnitudes versus frequency.

#### *2.2. Numerical Evaluation of the Antenna Performance in Free Space and on a Model of the Human Body*

Design, optimization and numerical evaluation of the antenna performance are carried out using the full-wave electromagnetic commercial software Remcom xFDTD (xFDTD, Remcom Inc., State College, PA, USA). This EM software packet is based on the finite difference time domain (FDTD) method which is extensively used for modelling antenna structures and human body models.

The following parameters are set in each simulation in order to obtain more accurate results. First, FDTD cell size of 0.5 mm × 0.5 mm × 0.5 mm is used for the antenna geometry. For the human body model and rest of the space, an inhomogeneous mesh having an increasing cell size from 0.5 to 1 mm is applied. All calculations continue until steady-state is reached. For the analyzed small antenna with a denim substrate, the steady-state is observed after 10,000 time-steps. A 7-layer perfect matched layer absorbing boundary condition is used.

Since the designed antenna is expected to be in close proximity or mounted directly on the different parts of the human body during the real operating conditions in a W-WSN, the design and optimization were made first in the free space. After that, geometrical dimensions of the antenna structure were optimized on a human body model. A homogeneous numerical model with dimensions 180 mm, 120 mm and 150 mm was employed to mimic the human body (ε ′ *r* = 40.805, σ = 2.33 S/m, and ρ = 1166 kg/m<sup>3</sup> ). The selected human model dimensions are those for the flat phantom pointed in standards EN 62209-2:2010 and IEEE Std.1528-2013 for the fixed frequency of 2.45 GHz so that the effect of the power reflection at the model surface on the peak 10 g average SAR is negligible (less than 1%). ′ σ ρ

The antenna was positioned on the surface of the phantom (the distance between the antenna and the phantom is 0 mm, as shown in Figure 3d) to study the effect of the human body on the antenna performance and SAR in the worst-case scenario.

**Figure 3.** Simulated: (**a**) Reflection coefficient magnitudes versus frequency; (**b**) Radiation efficiency and maximum gain versus frequency; (**c**) 3D radiation patterns of the antenna at 2.44 GHz in the free space and placed on the surface of the phantom; (**d**) Model of the antenna and phantom.

−6 dB The free space and on-phantom reflection coefficient magnitudes (|S11|) of the optimized antenna are presented in Figure 3a. In the free space, the antenna bandwidth (|S11|<−6 dB) is 285 MHz. As we can see the homogeneous phantom does not cause detuning effects on the resonant frequency. The |S11| curve and operating bandwidth for the case where the antenna is mounted on the phantom are the same as the case in free space.

The total radiation efficiency and maximum gain of the optimized antenna in the free space and on the phantom at the frequency bands of interest are displayed in Figure 3b. As we can see, the phantom causes a reduction in total radiation efficiency and maximum gain. The total radiation efficiency and maximum gain are decreased by a factor of about 0.5 when the antenna is placed on the phantom. Also, the total radiation efficiency and maximum gain show an increasing trend with increasing the frequency. Figure 3b shows that across the bandwidth of 2.36–2.5 GHz the simulated radiation efficiency varies from 12% to 16% when the antenna is in the free space and from 6% to 7.8% when it is placed on the phantom. –

From Figure 3c we can see that the antenna exhibits unidirectional radiation pattern. The front-to-back ratio is 12.20 dB in the free space and 22.55 dB on the phantom. Another essential question to be considered in designing antennas for W-WSNs is about concerning health hazard. Hence, to address this question, a more thorough evaluation and characterization of the SAR in the human body model were carried out. According to the safety guidelines and standard [24,25], the obtained peak 1 g and 10 g average SAR values should not exceed 1.6 W/kg and 2 W/kg.

Figure 4a shows the peak 1 g and 10 g FDTD-computed average SAR generated from the antenna in the phantom, in the MBAN and ISM bands when the antenna is positioned on the surface of the phantom (the distance between the antenna and the phantom is 0 mm). The results were normalized to a net input power of 100 mW. It can be seen that the peak SAR is frequency dependent. In general, the SAR increases as the frequency increases. For the considered input power, the peak 1 g and 10 g average SAR produced in the phantom are 0.533 and 0.148 W/kg, respectively at 2.48 GHz. The peak 10 g average SAR values are much lower than the maximum allowed value of 2 W/kg as required by the ICNIRP [24]. Moreover, the peak 10 g average SAR is found to be equal to ~2 W/kg when the net input power for the proposed antenna is 1353 mW. That is, the net input power as high as 1353 mW guarantee conformance with the safety guidelines imposed by the ICNIRP [24].

**Figure 4.** FDTD-computed SAR: (**a**) peak 1 g and 10 g average SAR, and whole-phantom averaged SAR versus frequency; (**b**) Distributions in xy-, yz- and zx-plane at 2.44 GHz and scale.

From the results presented in the Figure 4, it can be concluded that the proposed antenna exhibits low SAR values (peak 10 g average SAR is about ten times lower than the maximum allowed value of 2 W/kg) due to the shielding effect of the reflector and also due to the unidirectional radiation pattern of the antenna.

The SAR distributions on the surface and inside the phantom in the xy-, yz- and zx-planes are shown in Figure 4b. In the figure, we observe that the SAR values around the antenna edges are higher.

Also, under real-case scenarios, the distance between the wearable antenna and the human body may be changed, which may lead to a change in the antenna performance and SAR. Hence, numerical simulations were performed at distances 0 mm, 5 mm and 10 mm between the antenna and the human body model in order to investigate the robustness of the proposed design to effects of the human body loading.

Figure 5 presents the variation of the antenna radiation efficiency, maximum gain, peak 10 g average SAR and whole-phantom averaged SAR (at 100 mW net input power) versus the frequency for the distances 0 mm, 5 mm and 10 mm between the antenna and the phantom. As might be expected, the radiation efficiency increases with distance, while the peak 10 g average SAR and whole-phantom averaged SAR decrease. Also, the maximum gain shows an increasing trend with increasing the distance between the antenna and the phantom. From Figure 5b we see that the gain at a distance 10 mm from the phantom is larger than that in the free space. This is because the phantom, in this case, acts as a reflector, which enhances radiation in the direction opposed to the human body model.

**Figure 5.** Simulated: (**a**) Radiation efficiency; (**b**) Maximum gain; (**c**) peak 10 g average SAR; (**d**) whole-phantom averaged SAR at distances 0 mm, 5 mm and 10 mm between the antenna and the human body model.

Moreover, when the designed antenna is a part of a wearable wireless sensor node intended for a medical purpose such as tracking, recording, and monitoring of biomedical signals (used in a medical or home healthcare environment), it is essential to create an electromagnetically compatible device to minimize interference with other devices. In this case, the wearable wireless sensor node has to be designed and validated as a medical device (or medical electrical equipment according to the definition in IEC 60601-1). The standards which specify tests and requirements for the electromagnetic compatibility of the medical electrical equipment are IEC 60601-1-2 and IEC 60601-1-11.

Finally, examining the results from numerical simulation presented in Figures 3–5 as compared to the specifications of Table 1 illustrates that, the proposed antenna satisfies the requirements for application in W-WSNs. –

#### *2.3. Fabrication of the Antenna Prototype*

For the fabrication of the radiating elements of a textile antenna, it is possible to use the inkjet printing [26], embroidery [27] or cut-transfer-glueing process [2]. The choice usually is made on the base of the materials used for the radiating elements.

To fabricate the antenna's prototype, we used the cut-transfer-glueing process. By using the cutting machine, the antenna's elements were cut into the designed shapes with high accuracy (a tolerance of ±0.01 mm). A highly conductive fabric (DC conductivity 2.5 × 10 <sup>5</sup> S/m and 0.08 mm of thickness) was used for the conductive elements of the antenna. Next, both radiating elements and reflector were attached to the denim substrate by using a polymer tape that is activated by ironing. A coaxial cable (diameter 1.13 mm and length of 200 mm) with a 50 Ω U.FL connector was soldered to the CPW using a low-temperature solder wire. First, the soldering points of the coaxial cable with the conductive fabric were tin-plated at 250 ◦C (heating time 2.5 s), as shown in Figure 6. Then, the coaxial cable inner conductor was soldered on top of the middle conductor of the CPW while the coaxial cable braid shield was soldered to the CPW ground plane. ' ' 0 mm) with a 50 Ω U.FL connector was soldered to

**Figure 6.** Photographs of the antenna's prototype during the fabrication process. '

#### *2.4. Measurements*

The performance of the prototype of the designed antenna may be measured by using passive and active tests. In the passive tests, the prototype is connected to the measuring equipment (a network analyzer, signal generator, receiver, or spectrum analyzer) using a coaxial cable. In these tests, the antenna reflection coefficient magnitude (|S11|), bandwidth, gain, radiation pattern and radiation efficiency are measured. The detailed definition of the antenna parameters and their measurement methods are presented in [2,28,29].

Moreover, a full verification of the antenna design requires more extensive tests, which represents the behaviour of the antenna under practical operating conditions, called active tests [2,29,30]. In these measurements, a network simulator (or a radio communication test module) is used to set up a connection to the antenna under test, that is connected to a sensor node to reproduce a real-world performance. In order to conduct accurate and repeatable measurements, the testing needs to perform inside a chamber (anechoic or reverberation) with a controlled environment [2,28].

For antennas used near or on the human body such as antennas intended for sensor nodes in W-WSNs experimental measurements of the antenna parameters and characteristics on a physical model of the human body (called phantom) are also required. There are three types of phantoms for experimental use: solid, semi-solid, and liquid [31]. SAR measurements can be made using a robotic system, associated test equipment and a liquid phantom or using an infrared camera, associated test equipment and a solid or semi-solid phantom.

In order to assess the performance of the proposed antenna, |S11| was evaluated when the antenna is placed on a semi-solid phantom and in the case of free space condition. The semi-solid phantom (dimensions 180 mm (length) × 120 mm (width) × 150 mm (depth)) was fabricated accordingly to the recipe and technique described in [31]. The measurement of |S11| were performed using a Tektronix TTR503A vector network analyzer. As shown in Figure 7, when the antenna is placed on the semi-solid phantom, the |S11| remains almost unchanged. By comparing simulated and measured results, we observe a good agreement between them with a slight shift in the resonant frequency.

**Figure 7.** Simulated and measured reflection coefficient magnitudes versus frequency in the free space and on the semi-solid phantom.

In order to get a complete performance of the fabricated prototype, active tests were carried out. The first set of measurements was performed in a semi-anechoic chamber. Next, in order to take into account, the environmental variability, the measurements were repeated in a shielded room. All tests were carried out for both scenarios: (1) in the free space and (2) when the antenna is placed on the phantom.

−1 Two XBee S1 (DiGi International, Hopkins, MN, USA) modules were used in all tests as wireless nodes. One of the nodes was connected by a coaxial cable with a dipole antenna that has a resonant frequency of 2.44 GHz. The dipole antenna was in a vertical orientation. Next, the wireless node was connected to a personal computer via a UART-to-USB controller (Figure 8a,b) and was configured to generate a continuously repeated pseudo-random signal of 100 packets (each packet contains 50 bytes). The second wireless node was connected to the fabricated antenna prototype. The wireless nodes were connected, running XCTU software. On the XCTU, the transmission power, operating frequency and data rate, were set to −1 dBm, 2.41 GHz, and 9.6 kbits/s, respectively.

The semi-anechoic chamber was divided into 15 specific positions (three columns and five rows). The dipole antenna connected to the XBee node was stationary. At the same time, the other XBee node connected to the antenna prototype was located in each of described 15 specific positions in line-of-sight to the dipole. Both antennas were placed at a high of 1.33 m.

A range test was performed in order to determine the range and link quality between the nodes representing a real-world scenario. During the range test, XCTU sends data packets from the stationary XBee node to the remote node and waits for the echo to be returned from the remote node to the

stationary node. Also, during this test, the XCTU determines RSSI (Received Signal Strength Indicator) value and calculates the packet error rate.

(**a**)

(**b**)

**Figure 8.** 251 *Cont.*

**Figure 8.** RSSI: (**a**) Simple drawing of the configuration of the test setup; (**b**) Photograph of the test setup in the semi-anechoic chamber; (**c**) Distribution in the semi-anechoic chamber in the free space; (**d**) Distribution in the shielded room in the free space; (**e**) Distribution in the semi-anechoic chamber when the antenna is on the semi-solid phantom; (**f**) Distribution in the shielded room when the antenna is on the semi-solid phantom.

− −6 The distributions over the xy-plane of the measured RSSI in the semi-anechoic chamber and shielded room are shown in Figure 8. We can see that in the free space the measured RSSI is between −44 and −69 dBm depending on the specific antenna position in the semi-anechoic chamber (see Figure 8c). Also, when the distance between the antennas increases, the measured RSSIs decrease. When moving from semi-anechoic to shielded room scenario, it can be seen that in the free space the measured RSSIs vary between −37 and −56 dBm. A positive interference is clearly seen in Figure 8d when the distance between the antennas is 2 m (the measured RSSI is −37 dBm). Moreover, when the distance between the antennas is 6 m, the measured RSSI decreases to −56 dBm. Fluctuations of the RSSI values in the shielded room are due to constructive and destructive interference.

− − − − − − − − When the antenna is placed on the phantom, the measured RSSIs were in the range of −50 to −64 dBm in the semi-anechoic chamber (see Figure 8e) and in the range of −44 to −60 dBm in the shielded room (Figure 8f). Comparing the results with the free space scenario, we can conclude that the measured RSSIs when the antenna is on the phantom are lower (with about 6 dB) than those in the free space. These differences in RSSIs are attributed to the reduction in the radiation efficiency and gain when the antenna is on the phantom. We also observed that at all measurements, the packet error rate was zero.

Comparing the results, we see that at the same positions, the measured RSSIs in the semi-anechoic chamber are lower than those in the shielded room. This is since the more of the reflections in the semi-anechoic chamber are eliminated while in the shielded room, the propagation occurs by multiple reflections in the environment, which results in an additional energy contribution.

From the results in Figure 8, it is possible to conclude that the proposed antenna shows very good RSSI values both in the semi-anechoic chamber and shielded room which satisfies well the requirement of the receiver sensitivity in W-WSNs (−94 dBm) [32].

#### **3. Impact of EM Properties of the Textile Materials on the Performance of the Small Antennas for W-WSNs**

#### *3.1. Antenna Designs*

This section investigates the effect of the EM properties of different textile materials on the performance of small low-profile backed antennas. Three antenna structures: (1) Antenna with a cotton substrate, (2) Antenna with a polyamide-elastane substrate and (3) Antenna with a polyester substrate were developed. Here, each antenna has a substrate thickness of 1.5 mm, with a real part of the relative permittivity of 1.6321 (cotton), 1.5493 (polyamide-elastane) and 1.6202 (polyester), respectively. The loss tangent of the textile substrates is 0.0439 (cotton), 0.0146 (polyamide-elastane) and 0.0051 (polyester). In the design of the antennas (with substrates from cotton, polyester and polyamide-elastane) the configuration of the antenna with denim substrate was used. The difference being that at each substrate, the geometrical dimensions of the loop and parasite elements were optimized for maximum radiation efficiency and optimal impedance matching in the desired bandwidth. −

Figure 9 shows the structure and geometrical dimensions for each antenna. It is seen that in order to provide good impedance matching at the resonant frequency of 2.4 GHz, the perimeter of the square-loop on the denim substrate is 72 mm while the perimeter of the square-loop on the polyester substrate is 84 mm. Also, the loop strips width of the antenna with a polyester substrate are decreased to 1 mm to enhance the bandwidth.

**Figure 9.** Configuration of the antennas with denim, cotton, polyamide-elastane and polyester substrates.

From the results presented in the Figure 9, we can conclude, that the real part of the permittivity of the antenna substrate has consequences on the overall antenna size. As expected, with the increasing of the substrate material permittivity, the overall size of the antenna decreases [1,2,28].

A comparison with previously reported designs shows that the overall size of each of the four proposed antennas is between 25% and 90% smaller than the antennas in [1,9,10,12,14,15,26]. Moreover, the proposed antennas have a lower profile than [5,8,10–12,14].

#### *3.2. Results and Discussion*

Two scenarios were numerically studied when the antennas are: (1) in the free space and (2) placed on the semisolid phantom.

–

As shown in Figure 10a in free space, the antennas with denim and cotton substrates have a bandwidth (|S11| < −6 dB) of 285 and 269 MHz, respectively. In contrast, antennas with substrates from polyamide-elastane and polyester have a relatively narrow bandwidth of 178 and 130 MHz, respectively. We hypothesise that since the polyamide-elastane and polyester have a lower loss (loss tangent is between 0.005 and 0.015), the bandwidth of the antennas with substrates from these materials is more narrow versus the bandwidth of the antennas with denim and cotton substrates. −6 dB) of 285 and 269 MHz, respectively. In contrast, antennas with substrates

**Figure 10.** Simulated in the free space: (**a**) Reflection coefficient magnitudes versus frequency; (**b**) Radiation efficiency versus frequency; (**c**) Maximum gain versus frequency.

Figure 10b,c show the simulated radiation efficiency and maximum gain of the four antennas when they are in the free space. The behaviour of the radiation efficiency of the antennas, as a function of frequency, is shown in Figure 10b. From the results we can see that the radiation efficiency of the antennas with denim and cotton substrates remains almost unchanged (between 10% and 16%) across the MBAN and ISM bands. Furthermore, Figure 10b shows that the simulated radiation efficiency of the antenna with a polyamide-elastane substrate varies from 18% at 2.48 GHz to 30% at 2.42 GHz. The antenna with a substrate from polyester shows significant variations in the radiation efficiency from 28% at 2.36 GHz to 60% at 2.44 GHz.

The maximum gain of the antenna with a polyester substrate varies between −0.5 and 3.5 dBi across the MBAN and ISM bands, as seen in Figure 10c. The gain of the antenna with a polyamide-elastane substrate varies from −2.5 to 0 dBi. On the other hand, for antennas with cotton and denim substrates, small variations of only 1.5 dB in gain are observed. Also, the maximum gain of the antennas with denim and cotton substrates shows an increasing trend with increasing the frequency.

The simulated reflection coefficient magnitudes when the antennas were placed on the human body model are shown in Figure 11a. A good impedance matching is maintained for all four antennas despite the slight change in reflection coefficient magnitudes. The results show that the resonant frequency of the antennas with a substrate from cotton and denim fabric is not affected when positioned on the phantom. Similar results are observed in [1]. The resonant frequency of the antennas with a substrate from polyester and polyamide-elastane is slightly shifted up. Moreover, when the antennas are placed on the phantom, their bandwidths remain unchanged compared to the free-space scenario. It can be concluded that the resonant frequency and bandwidth are insensitive to detuning when the textile antennas backed by a reflector are positioned on the phantom.

−2.5

−

**Figure 11.** Simulated on the phantom: (**a**) Reflection coefficient magnitudes versus frequency; (**b**) Radiation efficiency versus frequency; (**c**) Maximum gain versus frequency.

The investigations have shown that the phantom significantly affected both the maximum gain and radiation efficiency of the antennas. For the antennas with denim and cotton substrates, the maximum gain and radiation efficiency are decreased by a factor of about 0.5 compared to the free space scenario. The simulated radiation efficiency of the antenna with a polyamide-elastane substrate falls between 5% and 13%, while simulated radiation efficiency of the antenna with a polyester substrate falls between 8% and 20%. The observed reductions in radiation efficiency and maximum gain are due to the power absorbed at the substrate and human body model.

Figure 12a,b compare the computed peak 10 g average SAR and whole-phantom averaged SAR generated from each of the four antennas in the homogeneous phantom, in the MBAN and ISM bands. The results are normalized to a net input power of 100 mW. The antennas with denim and cotton substrates exhibit very low SAR values (below 0.02 W/kg), as shown in Figure 12a. On the other hand, 10 g average SAR of the antennas with polyamide-elastane and polyester substrates varies significantly from 0.5 to 1.15 W/kg and from 0.5 to 1.95 W/kg, respectively. From the results presented in Figure 12a

it can be concluded that the peak 10 g average SAR for all four antennas is lower than that in the safety limit pointed in [24].

**Figure 12.** Simulated: (**a**) Peak 10 g average SAR (W/kg); (**b**) Whole-phantom averaged SAR (W/kg).

In MBAN and ISM bands, and for an input power of 100 mW, the whole-phantom averaged SAR for the antennas with denim and cotton substrates is below 0.002 W/kg (about 40 times lower than the maximum allowed value of 0.08 W/kg), as seen in Figure 12b. The obtained whole-phantom averaged SAR for the antennas with polyamide-elastane and polyester substrates is 10 and five times lower than the maximum allowed value of 0.08 W/kg, respectively.

Finally, the comparison between the antennas with substrates from natural fibres (cotton and denim) and substrates from synthetic fibres (polyamide-elastane and polyester) showed that the antennas with substrates from synthetic fibres give higher SAR values than the antennas with substrates from natural fibres (Figure 12).

Also, numerical simulations were performed for each antenna at distances 0 mm, 5 mm and 10 mm between the antenna and the human body model in order to investigate the effect of the EM properties of different textile materials on the robustness of the proposed designs to impact of the human body loading.

As might be expected, for each of the antennas the radiation efficiency and maximum gain increase with distance, while the peak 10 g average SAR decreases. From Figure 13b we also see that the gain at a distance 10 mm from the phantom for each antenna is larger than that in the free space. Consequently, we can conclude that the action of the phantom at a distance of 10 mm from the antenna is akin to a reflector, which enhances radiation in the direction opposed to the human body model.

Moreover, the computed maximum allowable net input powers which satisfy the ICNIRP restriction of 2 W/kg for peak 10 g average SAR for each antenna at different distances to the phantom are shown in Table 2. The maximum allowable net input power which satisfies the ICNIRP restriction at a distance of 0 mm is 1353 mW and 1062 mW for the antennas with denim and cotton substrates, respectively while for antennas with polyamide-elastane and polyester substrates is 176 mW and 104 mW, respectively.

Next, the prototypes of the antennas with cotton, polyamide-elastane and polyester substrates were fabricated using the procedure described in Section 2.3 and in [1]. Photographs of the fabricated prototypes are shown in Figure 14. The prototypes are very light (the weight of the prototype with a substrate from denim is 3.2 g, cotton 2.7 g, polyamide-elastane and polyester 2.6 g), which will allow the antennas to be easily integrated into clothing.

**Figure 13.** Simulated: (**a**) Radiation efficiency; (**b**) Maximum gain; (**c**) Peak 10 g average SAR versus frequency.

**Table 2.** Maximum allowable net input power which satisfies the ICNIRP restriction of 2 W/kg for peak 10 g average SAR in the homogeneous phantom at 2.48 GHz.


The measured |S11| for all four prototypes are given in Figure 15. As can be seen, a good impedance matching is maintained in both scenarios (in free space and on the phantom). Also, a good agreement can be found between measured and simulated (see Figures 10a and 11a) reflection coefficient magnitudes.

**Figure 14.** Photographs of the fabricated prototypes with a substrate from denim, cotton, polyamide-elastane and polyester.

**Figure 15.** Measured reflection coefficient magnitudes when the antennas are: (**a**) in the free space; (**b**) on the homogeneous phantom.

#### **4. Conclusions**

−6 dB −6 dB In this paper, we have presented a methodology for the design, fabrication and measurement of small wearable backed antennas for application in sensor nodes. Based on the presented design methodology, a new small-sized antenna with a denim substrate was proposed. Since the designed antenna is expected to be in close proximity to the human body during the real operating conditions in a W-WSN, the design and optimization were made both in the free space and on a human body model. In the free space, the antenna exhibits bandwidth (|S11|<−6 dB) of 285 MHz and radiation efficiency from 12% to 16% in MBAN and ISM bands. The simulated radiation efficiency varies from 6% to 7.8% when the antenna is placed on the phantom.

Concerning the health hazard, a more thorough evaluation and characterization of the SAR in the human body model were carried out. From the results, it can be concluded that the proposed antenna exhibits low SAR values (peak 10 g average SAR is about ten times lower than the maximum allowed value of 2 W/kg) due to the shielding effect of the reflector and also due to the unidirectional radiation pattern of the antenna.

The performance of the prototype of the proposed antenna was evaluated using passive and active tests. A good agreement between simulated and measured |S11| with a slight shift in the resonant frequency is observed. In order to get a complete performance of the fabricated prototype, range tests were performed in a semi-anechoic chamber and in a shielded room. Depending on the specific antenna position, the measured RSSIs vary between −44 and −69 dBm in the semi-anechoic chamber and between −37 and −56 dBm in the shielded room. Hence, it is possible to conclude that proposed antenna shows very good RSSIs which satisfy the requirement of the receiver sensitivity in W-WSNs (−94 dBm).

To investigate the impact of the textile materials on the antenna performance, the antenna geometry was studied on four textile substrates (denim, cotton, polyamide-elastane and polyester). The reflection coefficient magnitudes, bandwidth, maximum gain and radiation efficiency of the four antennas were numerically studied and compared to two scenarios: (1) in the free space and (2) on the semisolid phantom. The numerical investigations reveal that in MBAN and ISM bands, each of these four textile antennas achieves stable performance in both scenarios.

From the results, we can conclude, that the real part of the permittivity of the antenna substrate has consequences on the overall antenna size and resonant frequency. As expected, with increasing the substrate material permittivity, the overall size of the antenna decreases. Results show that the gain and radiation efficiency decrease with increasing dielectric losses in the textile substrate. Results for peak 10 g average SAR have revealed that the antennas with denim and cotton substrates exhibit SAR values below 0.02 W/kg. On the other hand, the peak 10 g average SAR of the antennas with polyamide-elastane and polyester substrates varies significantly from 0.5 to 1.5 W/kg and from 0.5 to 1.95 W/kg, in the frequency range of 2.36 to 2.48 GHz. Consequently, the peak 10 g average SARs are sensitive to substrate material dielectric loss.

In order to verify the results from the numerical simulations, prototypes of the antennas were fabricated, and their parameters were measured in a semi-anechoic chamber. A good agreement is found between measured and simulated reflection coefficient magnitudes of the antenna prototypes for two scenarios when the antennas are: (1) in the free space and (2) on the semisolid phantom. Therefore, the proposed small textile antennas backed by a reflector are promising candidates for integration into garments for applications in W-WSNs.

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

**Funding:** This research was funded by BULGARIAN NATIONAL SCIENCE FUND at the Ministry of Education and Science, Bulgaria, grant number KP-06-H27/11 from 11th December 2018 "Antenna technology for wearable devices in the future communication networks".

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **A Fully-Printed CRLH Dual-Band Dipole Antenna Fed by a Compact CRLH Dual-Band Balun**

#### **Muhammad Kamran Khattak, Changhyeong Lee , Heejun Park and Sungtek Kahng \***

Department of Information & Telecommunication Engineering, Incheon National University, Incheon 22012, Korea; Kamrankhattak01@gmail.com (M.K.K.); Antman@inu.ac.kr (C.L.); h.park.inu@gmail.com (H.P.)

**\*** Correspondence: s-kahng@inu.ac.kr; Tel.: +82-10-3423-0817

Received: 13 July 2020; Accepted: 31 August 2020; Published: 3 September 2020

**Abstract:** In this paper, a new design method is proposed for a planar and compact dual-band dipole antenna. The dipole antenna has arms as a hybrid CRLH (Composite right- and left-handed) transmission-line comprising distributed and lumped elements for the dual-band function. The two arms are fed by the outputs of a compact and printed CRLH dual-band balun which consists of a CRLH hybrid coupler and an additional CRLH phase-shifter. Its operational frequencies are 2.4 and 5.2 GHz as popular mobile applications. Verifying the method, the circuit approach, EM (Electromagnetics) simulation and measurement are conducted and their results turn out to agree well with each other. Additionally, the CRLH property is shown with the dispersion diagram and the effective size-reduction is mentioned.

**Keywords:** dual-band dipole; CRLH antenna; dual-band balun; CRLH balun; wireless communication

#### **1. Introduction**

Nowadays, wireless connectivity from equipment to other equipment and technological convergence becomes more vibrant. This results from a need to push the current limits. For instance, the 2.4 GHz band in the WLAN (Wireless local area network) frequency is commonly used, but this band is not sufficient to provide proper amounts of data because of excessive use and interference with other wireless communication methods using the same frequency band (e.g., Bluetooth, DCP, and ZigBee). So, to meet the frequency requirements, a number of multiband antennas have been proposed with various structures such as IFAs (Inverted-F antennas), bow-ties, slots and monopoles [1–4]. Pushpakaran et al. proposed a dog bone shape dual-band dipole antenna for WLAN applications. They presented a method for achieving a dual-band property by using the stacking technique [5]. Deepak [6] proposed a dipole antenna along with a folded element for multi-band operation. To eliminate the return current leak of the SMA (Subminiature version A) connector, they separate the dipole arms by using the double sides. Sim et al. reported multi-band asymmetric dipole antenna for WLAN operation. For the feeding of dipole antenna, they are using the 50-ohm coaxial cable as positive and negative on the two arms [7]. J. Huang used a tapered transition the balance feed line in [8]. The trapezoid dipole arms are printed on one side of the substrate and the other single dipole is formed on the opposite side. Nair [9] presented an F-shaped slot line to feed a dual-band dipole antenna which are fed by an SMA connector. H. Azeez modified a dipole with a pair of E-shaped conductive arms to generate multiple resonance [10]. Its feeding scheme is the same as [9]. Others put their radiating elements near a large metal ground like Alekseytsev who coupled a slit with an overpass metal-line to excite an open loop and a parasitic for two resonance frequencies [11]. More complicated configurations are built by Barani, such as one PIFA (Planar Inverted-F antenna) conjoined with PIFAs, with monopoles and slits on the ground to increase the number of bands [12]. Because the ground-edge mounted antennas occupy a large footprint, multi-band antennas are made as 3D structures. Tang adopted vertical loops passing through via-holes of two parallel lines [13]. Sreelakshmy put a thick disk on a vertical feed-line, and formed two asymmetric holes splitting one angular and radial mode into two, and used them for the dual-band characteristics [14].

The previously mentioned multi-band dipole antennas adopt SMA and coaxial feeds as an unbalanced signal port. To cope with that, the balanced to unbalanced (balun) structure is needed for a dipole antenna. Besides, to feed the dual-band dipole antenna, it is necessary to combine the dual-band balun. Liu et al. proposed the CRLH (Composite right- and left-handed) transmission line balun, in which they made the virtual ground for the odd mode and shunt inductor to remove a via. So, the phase difference between the two output ports is around 180◦ at 1.5 and 3.6 GHz [15]. Tseng et al. developed a CRLH balun to control the phase slope. As a result, they obtained the multi-band property by using the two branches where one arm consists of the RH-TL (Right-handed transmission line) and the other part is based on the LH-TL (Left-handed transmission line) with lumped elements [16]. According to odd-mode and even-mode analysis, the dual-band baluns are proposed with open-stub and BPF (Bandpass filter) [17,18]. Huang et al. reported a microstrip-based Marchand-type balun. They could achieve the high selectivity by using the roll-off of the filter [19]. Isolation between two bands can be better by making an SIR (Step impedance resonator) feedback loop between the two output branches by Li [20]. Multiple loops causing the phase difference are formed through layers with vertical lines as bridges [21]. Kahng et al. proposed a CRLH-based balun for common-mode current indicator which is employed in the chip and PCB EMC (Printed circuit board electromagnetics compatibility) problems. The balun consists of the branch-line coupler and the compact metamaterial phase shifters that are horizontally cascaded [22]. It works for a single band.

In this paper, a compact and fully printed metamaterial dual-band balun and metamaterial dual-band dipole antenna are proposed and put into one structure. The dual-band balun has the CRLH hybrid branch line-coupler and a CRLH phase-shifter. The conventional coupler and phase shift lines are replaced by miniaturized phase-shift lines having +90◦ for one frequency and −90◦ for another frequency for dual-band and low-cost fabrication. Instead of power-divider shapes, using lumped L and C, and multiple and long stages seen in others' works, the new balun has single-stage hybrid coupler and stage- lines and phase-shifter, and complete distribute elements and its design is elaborated on in detail. The dipole antenna is different from others by having two arms in the form of hybrid metamaterial lines and have the same omni-directional pattern at the two resonance frequencies; 2.4 and 5.2 GHz are chosen as the test case and the circuit approach is done first, and followed by the EM simulation and the measurement. Additionally, the CRLH property is shown with the dispersion diagram and the size reduction effect of the proposed balun is addressed.

#### **2. Compact CRLH Dual-Band Balun**

#### *2.1. CRLH Dual-Band Hybrid Branch-Line Coupler*

In this section, the design of a dual-band balun as fully-printed distributed-elements without lumped elements is explained. The size of the entire structure will be much reduced by devising a CRLH phase shift-line to have 90◦ at the lower frequency band and −90◦ at the higher frequency band, and by forming a compact branch-line coupler.

The circuit of the CRLH phase-shift line is set up by solving the equations to get the dual-band properties as the hybrid branch line coupler. Before starting the process of the design, as a goal, the following specifications in Table 1 are given as follows.


**Table 1.** The specifications of the composite right- and left-handed (CRLH) dual-band balun.

In Figure 1, *CR*, *CL*, *L<sup>R</sup>* and *L<sup>L</sup>* are determined by generating 90◦ at *f* <sup>1</sup>, and −90◦ at *f* <sup>2</sup>. It is a matter of course, the zeroth-order resonance (ZOR) should also be created around (*f* <sup>1</sup> + *f* <sup>2</sup>)/2 for the size-reduction. So, the mathematical equations for *CR*, *CL*, *L<sup>R</sup>* and *L<sup>L</sup>* are as follows [22,23]. Others end up with multiple stages, but a single-stage circuit is proposed here to reduce the size. − − − − − −

$$L\_R = \frac{\mathbf{Z\_c}\pi[\left(\frac{\omega\_1}{\omega\_2}\right)+1)]}{2\omega\_2[1-(\frac{\omega\_1}{\omega\_2})^2]}, \mathbf{C\_L} = \frac{\pi[\left(\frac{\omega\_1}{\omega\_2}\right)+1)]}{2\omega\_2\mathbf{Z\_c}[1-(\frac{\omega\_1}{\omega\_2})^2]}$$

$$L\_L = \frac{2\mathbf{Z\_c}[1-\left(\frac{\omega\_1}{\omega\_2}\right))]}{\pi\omega\_1[1+(\frac{\omega\_1}{\omega\_2})^2]}, \mathbf{C\_R} = \frac{2\pi[(\frac{\omega\_1}{\omega\_2})+1)]}{\omega\_1\mathbf{Z\_c}[1+(\frac{\omega\_1}{\omega\_2})^2]}$$

where ω<sup>1</sup> = 2π*f*1, ω<sup>2</sup> = 2π*f*<sup>2</sup> and Z<sup>c</sup> characteristic impedance in above equations. Solving the equations by setting Zc, *f* <sup>1</sup> and *f* <sup>2</sup> at 35.5 Ω, 2.4 and 5.2 GHz, *CR*, *CL*, *L<sup>R</sup>* and *L<sup>L</sup>* are as 2.52 pF, 0.64 pF, 3.17 and 0.81 nH, respectively. Additionally, to get the 50-Ω case, tackling the equations as *f* <sup>1</sup> and *f* <sup>2</sup> at 2.4 and 5.2 GHz the values are as follows: 1.78 and 0.46 pF, 4.46 and 1.14 nH, in the same order with the 35.5-Ω case. Using these elements, the phase and dispersion diagrams are plotted and show the CRLH characteristics including the ZOR point as in Figure 2. πω<sup>1</sup> [1+(ω<sup>1</sup> ω2 ) 2 ] ω1Z<sup>c</sup> [1+(ω<sup>1</sup> ω2 ) 2 ] <sup>ଵ</sup> = 2<sup>ଵ</sup> <sup>ଶ</sup> = 2<sup>ଶ</sup> Ω Ω Ω πω<sup>1</sup> ω2 ) ] ω1Z<sup>c</sup> ω2 ) ] <sup>ଵ</sup> = 2<sup>ଵ</sup> <sup>ଶ</sup> = 2<sup>ଶ</sup> Ω Ω Ω

**Figure 1.** Equivalent circuit of the proposed dual-band CRLH phase-shifter line.

**Figure 2.** Circuit simulated results of the CRLH phase-shift line: (**a**) phase; (**b**) dispersion diagram.

−

−

In Figure 2a, the phases of 90◦ and −90◦ are achieved at 2.4 and 5.2 GHz. This phase-shift line complies with the specifications in Table 1 and will put into the 90◦ -branches of the hybrid branch-line coupler for miniaturization in terms of the physical size. Besides, there is the dispersion diagram in Figure 2b where the curve shows the LH region (β < 0) and the RH region (β > 0), along with the zeroth-order resonance (ZOR) near 3.8 GHz. β β

The finalized physical dimensions of the 35.5 Ω phase-shift line and geometry are given as table II and Figure 3a. Additionally, the physical dimensions of the 50 Ω phase-shift line and geometry are given as Table 2 and Figure 3b. The single-stage fully printable CRLH phase-shift line is designed using the CST-MWS as a full-wave EM simulator. Figure 3 shows the proposed CRLH phase-shift line geometry and EM simulated data of the phases. Both 35.5 and 50 Ω have an interdigital and shorting structure for CR, CL, L<sup>R</sup> and LL. The phases of 90◦ and −90◦ are realized at 2.4 and 5.2 GHz at 35.5 and 50 Ω. These will be substituted for the 90◦ -branches of the hybrid-branch-line coupler with a view to the effective size reduction of the dual-band balun. Ω Ω Ω − Ω

**Figure 3.** The physical geometries of the phase-shifter for (**a**) 35ΩΩ and (**b**) 50ΩΩ of their phases (**c**).


**Table 2.** Physical dimensions of the proposed CRLH phase-shift line.

**Ω**

Each of the segments in Figure 4a is realized with Table 2 for Figure 3a,b and becomes Figure 4b. As a crucial building block of the proposed balun, the function of the dual-band branch-line coupler is checked with the electric-field distributions at the frequencies of interest as in Figure 4c. The RF energy from port 1 is split into the output ports, and port 2 is turned on first as the 0◦ - field-shot, and then port 3 is turned on after 90◦ -lapse for 2.4 and 5.2 GHz. Nonetheless, at 4 GHz as the stopband, the electric field is not detected at the output ports in the 0◦ -field-shot and 90◦ -shot. The geometry is physically realized as in Figure 4d where the length is about 2 cm.

**Figure 4.** *Cont*.

(**c**)

(**d**)

**Figure 4.** The physical geometry of the dual-band CRLH hybrid branch-line coupler as a single-stage geometry: (**a**) schematic; (**b**) EM (Electromagnetic) design; (**c**) E-field distribution; (**d**) fabricated prototype.

Figure 5a shows S11, S<sup>21</sup> and S<sup>31</sup> of the EM simulation From S11, the impedance is matched at the two frequencies. S<sup>21</sup> and S<sup>31</sup> present the equal power-division at the outputs. Figure 5b has the measured S11, S<sup>21</sup> and S31. Similarly, to the simulated results, the power-division as well as the impedance match are obtained at 2.4 and 5.2 GHz. Figure 5c reveals the phase difference obtained as expected for the branch-line coupler. Somewhat incomplete parts of the function will be mitigated in the stage of the balun.

**Figure 5.** The frequency response of the dual-band CRLH hybrid branch-line coupler (**a**) EM simulated; (**b**) measured; (**c**) phase difference.

#### *2.2. CRLH Dual-Band Balun*

The branch-line coupler of Figure 4a is developed to the metamaterial balun of Figure 6a in the level of the schematic. The implementable geometry of the CRLH hybrid branch-line coupler extended with the CRLH phase-shift lines is shown in Figure 6b. Prior to fabrication, electromagnetic observation is conducted on the suggested device. The purpose of this observation is to see this device working as a desirable balun that shows the 180◦ -phase difference between ports 2 and 3 from the standpoint of the electric field. In Figure 6c, for 2.4 and 5.2 GHz, the RF energy is divided to ports 2 and 3, and port 2 and port 3 take turns by a 180◦ -time lapse. Meanwhile, at 4 GHz, bot output ports appear dark, which means no energy passes the circuit. As to the fabricated balun, its overall size is <sup>14</sup> <sup>×</sup> <sup>20</sup> <sup>×</sup> 1.2 mm<sup>3</sup> and realized on the FR4 consistent to the former design stage. Considering 27 <sup>×</sup> <sup>27</sup> <sup>×</sup> 1.2 mm<sup>3</sup> as the nominal size of other dual-band baluns with FR4, the size is reduced more than four times. The proposed balun shows similar results to circuit design as shown in Figure 7. The dual-band operating frequency is 2.4 and 5.2 GHz as satisfying the specifications in Table 1. The curves of S<sup>21</sup> and S<sup>31</sup> are almost the same each other as the −3.8 dB. This also results from the use of FR4 substrate. Here comes 180◦ as the phase difference at *f<sup>1</sup>* and *f2*. −

(**b**)

**Figure 6.** *Cont*.

(**d**)

**Figure 6.** The physical geometry of the dual-band CRLH balun as a single-stage geometry: (**a**) schematic; (**b**) EM design; (**c**) E-field distribution; (**d**) fabricated design.

The frequency response of this compact balun works as expected. In Figure 7, both the simulated and measured s-parameters meet the design specifications. Additionally, the simulation and measurement results are in good agreement. The 180◦ -phase difference is seen. With regard to the size of the dual-band balun from the dual-band hybrid branch-line coupler, there is no change even though there is one new component added to the hybrid branch-line coupler. Since the new component as the dual-band phase-shift line is made as part of output port 3, it does not work as something negative in the size-reduction. Next, a dual-band dipole printable on the substrate is elaborated on.

(**c**)

**Figure 7.** The frequency response of the dual-band CRLH balun: (**a**) EM simulated; (**b**) measured; (**c**) phase difference.

#### **3. CRLH Dual-Band Dipole**

The motivation of this part of the work is to make a dipole resonate at two frequencies not in the harmonic relationship and have the same omni-directional pattern at different resonance frequencies as a dipole. The proposed CRLH dipole antenna which has a dual-band property and operating at 2.4 and 5.2 GHz shown in Figure 8. In Figure 8a the highlighted cyan section indicates the upper path and yellow section express the lower path as an equivalent circuit for Figure 8b where R<sup>R</sup> = 67 Ω, LR1 = 3.9 nH, LR2 = 25.5 nH, LL1 = 12 nH, LL2 = 2 nH, C<sup>R</sup> = 0.38 pF, C<sup>L</sup> = 0.5 pF. These result in the resonance at 2.4 and 5.2 GHz as in Figure 8c. In contrast to other dual-band dipoles, to maintain the omni-directional radiation pattern by suggesting a bow shape structure, a new structure is designed by adding a new path and making the entire geometry as the hybrid metamaterial arms on FR4 substrate. The upper current path is used to excite the high frequency and lower side path is matching at the low frequency. So, the proposed antenna connected from the top as the lumped inductor through the bottom path lumped capacitors as a loop shape. Geometrical elements ws, w<sup>f</sup> , w1, w2, wg, g1, g2, ls, l<sup>f</sup> , l1, l2, l3, l4, and l<sup>g</sup> are 100, 3, 80, 3, 150, 5, 150, 85, 20, 22, 15, 10, 85, and 3, respectively in mm. Compared to 140 mm or larger in length for commercial dipoles which are volumetric with the dielectric enclosure, the proposed antenna is smaller and planar. Using the geometrical parameters, the simulated and measured S<sup>11</sup> as well as the far-field pattern is provided next. Ω

**Figure 8.** *Cont*.

<sup>(</sup>**c**)

**Figure 8.** The CRLH dual-band dipole antenna design: (**a**) equivalent circuit; (**b**) S<sup>11</sup> of the equivalent circuit; (**c**) structure in the EM CAD (Computer aided design) program.

Figure 9a,b expresses the resonant currents. The EM simulated and measured S<sup>11</sup> in Figure 9c agree well and comply with the objectives. Dispersion curves are obtained from the circuit model and EM model of the new dipole as in Figure 9c. While 5.2 GHz has a positive wavenumber, 2.4 GHz has a negative wavenumber, which reveals the metamaterial property, helpful to size-reduction. Additionally, the far-field patterns with antenna gains 3.9 dBi and 2.38 dBi for the two frequencies seem very similar as given in Figure 9e.

(**b**)

**Figure 9.** *Cont*.

**Figure 9.** CRLH dual-band dipole: (**a**) current at 2.4 GHz; (**b**) current at 5.2 GHz; (**c**) S11; (**d**) dispersion curves; (**e**) measured beam-patterns.

#### **4. CRLH Dual-Band Antenna with CRLH Dual-Band Balun**

Figure 10a is a sketch of the idea how the balun is used to feed the circuit model of the dipole antenna. The realized CRLH dual-band balun and antenna structure are provided in Figure 10b,c. The measurement harness at an anechoic chamber is seen in Figure 10d. The antenna gains are observed and they are above −1 dBi at the two frequencies. Table 3 presents the simulated and measured antenna gains. Figure 11a,b shows the gains and radiation characteristics of the CRLH dual-band antenna at 2.4 and 5.2 GHz are obtained similar to Figure 9. As to the antenna efficiency, it is over and similar to 50% in simulation, and becomes lower than 40% in measurement as checked in Table 3. They are changed from the simulated results in a small scale, due to the manual soldering for wiring of the semi-rigid coaxial cables and connector adapter as a major reason as well as the deviation of FR4 dielectric constant from the vendor's data. For example, it was found out the inductor was soldered in the lab which has the effect of adding an extra inductance and pushes the second resonance frequency downward by approximately 0.4 GHz. The fabricated antenna gives a not-very-high efficiency due to the aforementioned errors, but the efficiency at the two frequencies is usable for communication as guessed from practices. With regard to the approach of combining the dual-band balun with the dual-band dipole, the gain of the antenna falls by 0.7 dB at 2.4 GHz and 1.1 dB at 5.2 GHz in the simulation because of adding the balun. This is loss, but the antenna gain is nearly 1 dBi. The loss grows in the measurement as 5 dB at 2.4 GHz and 4 dB at 5.2 GHz, even though the antenna gain for the two frequencies is below 0 dBi. The loss from the combination is accounted for by the 1-dB insertion loss of the manufactured balun, and the 3.5 dB of the mismatch from cabling and connector attachment. Precision etching with a low-loss substrate and cascading all the blocks without the cables will mitigate this loss problem. As for the far-field patterns of the antenna is broad and close to the omni-directional shape. This can be an obvious advantage, as a small planar balun feed and a planar dipole enabling the same radiation function working at different frequencies, while commercial baluns and dual-band dipoles are bulky and give difficulty in being integrated to flat wireless systems for diverse use-cases such as IoT (Internet of Things) apparatus. −

(**a**) **Figure 10.** *Cont*.

(**d**)

**Figure 10.** The CRLH dual-band dipole fed by the CRLH balun: (**a**) schematic; (**b**) hybrid schematic; (**c**) fabricated geometry; (**d**) far-field pattern measurement facility.


**Table 3.** Antenna gain and efficiency.

**Figure 11.** The radiation patterns of the CRLH dual-band dipole with CRLH balun: (**a**) simulated and measured results at 2.4 GHz; (**b**) simulated and measured results at 5.2 GHz.

#### **5. Discussion**

It is meaningful to check the technical status of the proposed idea in reflection of others' in the same subject. First, comparison is made between the new balun and other baluns aiming at multiple bands.

− − Through Table 4, the features of the new metamaterial balun are mentioned with those of selected papers as above. The kinds of features are type of structure, frequency bands, size, metamaterial or not, and origin of the design. While Tseng et al proposed the mixture of distributed and lumped elements, this work and the others are made from distributed elements, which provides ease of manufacture [16]. Except for this report, the circuits work for 2.4 and 5.2 GHz, and both works adopt the concept of metamaterials. They are obviously different in the following observations Tseng [16] uses multiple stages of lumped elements as a metamaterial line and a conventional delay-line to be a power-divider not a hybrid coupler. However, the proposed structure has no lumped elements and single-stage metamaterial-line segments for a hybrid coupler not a power-divider. Size-wise, the proposed balun is

much smaller than others'. Huang, F et al presented a little bit smaller than the proposed geometry because their line is diagonally folded into the center of the area and the input port is connected to the common corner, but this work has not folded the branch-line coupler. The other circuits originate from the power-divider, filter, and T-junction, but the backbone of this work is the branch-line coupler similar to Yang et al's design [18]. Please note, Yang's device [18] is not a metamaterial but a coupler with stubs. This comparison is convincing that the proposed balun is novel and appropriate for the objectives. Second, it is worth checking characteristics and benefits of the proposed dual-band dipole in comparison with others in the same subject.


**Table 4.** Comparison of the proposed CRLH dual-band balun and the previous works.

The antennas are looked into in terms of type, frequency, size, whether metamaterial or not, layering, and with or without a balun as in Table 5. Deng, Li, and Deepak's strucutres [1,2] and [6] are PIFA, bow-tie and SIR, respectively, but most of them take the form of a dipole or its modification. In contrast to all, only this work uses a metamaterial dipole. While Li, Huang, Sim, and Azeez's antennas [2,4,7] and [10] have more than three bands, 2.4 and 5.2 GHz are obtained by the other dipoles. To check the sizes as miniaturization effect, the areas of all the antennas are compared, which leads to a finding that the proposed antenna is substantially small. As given a question which antennas are metamaterials, this work (with Huang and Pushpakaran's antennas [4] and [5]) answers the question. Huang and Pushpakaran's antennas [4] and [5] are patch with a slit and slit coupled patches, respectively, and they have no evidence of metamaterials. However, this work reveals metamaterial properties with Figures 2b and 9d. With regard to layering, which is related to cost, Li, Huang, Pushpakaran, Huang, and Alekseytsev's antennas [2,4,5,8] and [11] have two or three layers. Nonetheless, this work is realized with a single layer of dielectric, which is cost effective. Most of them do not have baluns, but Huang et al's antenna [8] uses a 2-faceted balun comprising an upper tapered line and a lower line, and this work is connected with the CRLH balun of a single layer. The proposed antenna and CRLH balun have distinct features as explained.

**Table 5.** Comparison of the proposed CRLH dual-band dipole antenna and the previous works.


#### **6. Conclusions**

A dual-band balun and a dual-band dipole antenna for the 2.4 and 5.2 GHz wireless communication are designed and combined to meet the demands on higher throughputs in IoT mobile connectivity with standing and portable electronic products. These two blocks are novel and small in that different from others, distributed type and single-stages of CRLH TX-lines are basically used to meet the requirement on the dual-band performance. In detail, the balun originates from a branch-line coupler and its segments which are 90◦ -lines in the conventional designs are replaced by metamaterial phase-shifters that show −90◦ -phase at 2.4 GHz and +90◦ -phase at 5.2 GHz as a small structure. The one-stage compact balun eventually makes the output ports out of phase by 180◦ at the target frequencies. In order to coincide with the two frequencies of the balun, an ordinary dipole is changed to a CRLH structure where a negative phase and a positive phase are generated at 2.4 and 5.2 GHz as the resonance condition of the dipole. The size does not grow, since the negative phase occurs at the lower frequency. The two bands of the dipole are fed by the balanced currents from the CRLH balun. This enables the cascaded blocks to radiate the far-field wave omni-directionally at the two frequencies with the acceptable antenna gains greater than −1 dBi, despite the error in the experiment due to the manual soldering of cabling as a major reason and the dielectric constant deviating from the vendor's data as a minor one.

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

**Funding:** The lead authors Changhyeong Lee and Sungtek Kahng on behalf of the contributors acknowledge that this work was supported by the Post-Doctoral Research Program (2017) in the Incheon National University. Besides, it should be mentioned that this work was carried out with the support of "Cooperative Research Program for Agriculture Science & Technology Development (Project No. PJ014762)" Rural Development Administration, Republic of Korea.

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).
