This paper analysed the effects of bending on the performance of a textile antenna wherein the antenna under test was made of felt substrate for both industrial, scientific, and medical (ISM) band and WBAN applications at 2.45 GHz. Moreover, the conductive material was used for the patch, and the ground plane used a 0.17 mm Shieldit textile. Meanwhile, the antenna structure was in the form of rectangular, with a line patch in between elements to abate the mutual coupling effect. The measured operating frequency range of the antenna spanned from 2.33 GHz to 2.5 GHz with a gain of 4.7 dBi at 2.45 GHz. In this paper, the antenna robustness was examined by bending the structure on different radii and degrees along both X- and Y-axis. Next, the effects on return loss, bandwidth, isolation, and radiation characteristics were analysed. This paper also discovered that the antenna’s performance remained acceptable as it was deformed, and the measured results agreed well with the simulation.

1. Introduction

The concept of wearable antenna design was originated from the idea of having wearable computing systems as part of clothing that provides the wearer with hassle-free continuous data transfer. Extensive research studies have investigated the realisation of wireless body area networks (WBANs), specifically on the textile antenna design in medical and nonmedical applications. On top of that, the emergence of 5G technology creates multiple prospects of WBAN applications, including child protection, tracking, military, real-time health monitoring, biosensors, tourism, and security [14].

Furthermore, the incorporation of antennas and flexible electronics into clothes as a body-worn device eliminates the need for carrying a device, thus providing comfortability to the wearer [1]. Besides, in terms of sustainability of the wearable device, it should be lightweight, offer great flexibility, inexpensive, and robust. Considering these characteristics, this paper proposed a simple structure of rectangular patch multiple-input multiple-output (MIMO) antenna [59]. It should be noted that MIMO systems are extensively used in wireless communication to optimise the system with greater signal receptions and enlarge channel capacity [10].

For a flexible antenna, bending is a vital parameter as it could experience bending and crumpling during practical application. Additionally, a wearable dual-band antenna with an artificial magnetic conductor (AMC) structure on the back is proposed to alleviate body coupling [3]. It could be observed that the antenna reflection coefficient was conserved better during Y-axis bending compared with X-axis bending. Works in [4] showed an interesting finding on the bending effect of the antenna performance with different dielectric values for substrate, where the results showed that the substrate with dielectric constant close to air (ɛr = 1) has stable performance. Research on the impact of different substrate thicknesses (from 2 mm to 10 mm) towards bending conditions was explored in [11]. In addition, bending the antenna resulted in moving S11 to a higher frequency, where a thickness of 6 mm was less affected by the changes when the antenna was deformed. The antenna characteristic is preserved during 64 mm radius X-axis bending with the proposed metamaterial-inspired isolator that combines the defected ground structures (DGSs) and modified split-ring resonators (SRRs) [12]. A study on cylindrically bending hollow and solid effects on fabric-based antenna showed that gain and radiation patterns are almost identical for solid and hollow cylinders under bending, with slightly decreased bandwidth on solid cylinders [13].

This paper presents a bending assessment on different radii and degrees to gauge the impact on the performances of the antenna in [14]. Bending in both X- and Y-axis planes was performed by placing the antenna on cylinders with different radii (25, 45, 55, and 65 mm), and their effects were investigated. Simulations against the human phantom model and folded bending were also analysed on the antenna. To the best of the author’s knowledge, there is no research available on the folded bending for wearable antenna design.

2. Materials and Methods

The antenna bending is characterised by the bending radius, R, as illustrated in Figure 1. Two types of bending were explored in this paper, the circular radius bending and degree bending in two principal planes, namely, X-axis and Y-axis bending. Figure 1(a) shows the antenna in flat; Figure 1(b) shows the antenna in circular bent in X-axis; and Figure 1(c) shows the antenna in degree bending in Y-axis. In this paper, the antenna is designed using computer simulation technology (CST) software, where R varied from 65 mm to 25 mm (worst case).

The proposed antenna structure is initiated from a rectangular patch, with the patch dimension of 56.6 × 47 mm2. The antenna is modelled using felt substrate with a dielectric constant of 1.44 and thickness of 3 mm, where Shieldit super with 0.17 mm thickness is used as the radiating part. Resonating at a frequency of 2.4 GHz, the antenna is improved by converting into 2 × 1 MIMO with the total dimension of 140 × 70 mm2. A small gap of 0.1 λ between patch elements was achieved with the introduction of a line patch of 4 mm width that assists on mutual coupling suppression [15]. A detailed description of the proposed antenna is described in [14].

3. Results and Discussion

3.1. Circular Bending

The analysis starts with a circular bending where the antenna was bent in the Y-axis and X-axis with the radii of 65, 55, and 45 mm. Figure 2 shows the scattering parameter results when the antenna was bent in a different radius. Compared with the result of the antenna in a flat state, the S11 curve shifted to the right as the radius decreases from 65 mm to 25 mm for Y-axis bending as in Figure 2(a). On the other hand, the isolation of the antenna (shown by the S21 result) was improved for a smaller radius of bending.

Meanwhile, for X-axis circular bending, the resonance frequency shifted to the right as the radius decreased from 65 mm to 25 mm, as shown in Figure 2(b). On the positive side, the isolation of the antenna (shown by the S21 result) was improved by a smaller radius of bending due to the separation between radiating elements that get bigger as the degree bend increases.

Figure 3 shows the variation of off-body radiation pattern concerning the bending radius in E-plane and H-plane. In the Y-axis bending, the pattern is almost similar to the direction of radiation. However, the back lobe gets bigger as the bending degree increases (decrease in radius). On another note, the radiation pattern for X-axis bending tuned the angle of the main beam for about 30° in E-plane, while the pattern exhibits back lobe in H-plane when the antenna is bent into a different degree. Regarding gain for both axis bending, the gain decreases as the bending degree increases from radius 65 mm to 25 mm.

Table 1 compares the gain for different circular bending radii for both X-axis and Y-axis. Decreasing the circular radius bending will degrade the gain of the antenna. There are noticeable 16% and 12% gain reduction in the X-axis and Y-axis bending from flat to bending of 25 mm radius, respectively. To validate the MIMO performance, the envelope correlation coefficient (ECC) and diversity gain (DG) were monitored, as shown in Figure 4. ECC implies the degree of independence of the radiation patterns of two antennas [16], while the DG is the signal-to-noise ratio. Both parameters were related and calculated from S-parameter of radiation far field pattern results. Low ECC will result in high DG, where ECC < 0.5 is indicated as a good MIMO antenna [15]. In this work, calculated from S-parameter values from 2.4 GHz to 2.5 GHz range based on the proposed antenna, the values of ECC were below 0.01 with diversity gain close to 10 in all bending conditions.

3.2. Human Phantom Modelling

A cylindrical body part with four layers of skin, fat, muscle, and bone to resemble a human upper arm model was designed for on-body analysis. To analyse the effect of antenna performance towards bending in the human body’s vicinity, the antenna is bent according to the 55 mm radius human model in the X- and Y-axis as illustrated in Figure 5. Notably, the model has different dielectric properties and conductivity based on the type of human tissues and frequency. Table 2 shows the values at 2.4 GHz.

The simulated result is presented in Figure 5(a), where bending of the antenna to the human model has shifted the resonance to a higher frequency, which also degrades the matching of the antenna. Furthermore, the antenna resonates at 2.49 GHz (−20 dB) when the structure is bent in Y-axis to the human model, while for X-axis bending, S11 was shifted to 2.58 GHz with the value of −13 dB. As for mutual coupling, S21, the Y-axis bending improved the value from −28 to −36 dB.

On the other hand, the angle of the radiation pattern was shifted about 30° for X-axis, while it maintained the same direction during the Y-axis bending. These results are proved in Figure 6. It should be highlighted that the specific absorption rate (SAR) assessment was used to determine the limit of radiation exposure to the human body. The effect of SAR can be reduced by a large ground plane. Apparently, the proposed antenna is less affected by the human body proximity compared with the omnidirectional antenna as the full ground plane shields the radiating pattern to the human body, which also contributed to low specific absorption rate (SAR) values [1719]. Hence, in this work, no SAR simulation was conducted.

3.3. Folded Bending

The bending characteristic of the antenna was also analysed in terms of folding degree, ranging from 25° to 90°. In the Y-axis folding degree, the reflection coefficient was shifted to the higher frequency as the bending degree increases from 0° (flat) to 90°. The result is shown in Figure 7, where the resonant frequency shifted from 2.42 GHz to 2.512 GHz when the antenna is bent to 90°, whereas the isolation is shiftless at −26 dB.

In contrast, the result for X-axis bending is almost consistent from 25° to 90° bent regarding the impedance bandwidth with less than 1% different in S11 result, with better matching as the bending degree increased. A similar trend could be seen in the S21 result, where the isolation improved from −26 dB to −38 dB. Also, the radiation pattern is another parameter used to evaluate the antenna performances, and the different degree bending patterns are depicted in Figure 8. There are prominent changes in the Y-axis bending antenna pattern, where the values decline with a shifted angle of about 30° in H-plane. Conversely, the front lobe pattern maintained in the X-axis degree bending was slightly bigger in the back lobe as the degree bending increased. Table 3 tabulates the gain result for different degrees in X- and Y-axes. For most of the results, the gain decreases as the bending degree increases. Nevertheless, for Y-axis bending of 45°, there is a 5% increment in the gain. Another important observation is that the MIMO performance matrices are maintained during the folded bend of the antenna (see Figure 9), where the antenna exhibits lower ECC and good diversity gain.

3.4. Measurement Result

For the purpose of verifying the antenna characteristic, the antenna was fabricated, and its performance was measured. The antenna was fabricated by manually cutting the textile based on the size as in the simulation, as displayed in Figure 10. The fabricated antenna showed good conformity between simulated and measured results. However, the measured S11 shifted to the right of about 2.8%, and there was a 1 dB increment of S21, as shown in Figure 11(a). The shifted operating frequency is due to the fabrication in accuracy. On another note, the measured radiation pattern agreed well with the simulated result, as in Figure 11(b). Figure 11(b) shows the measured radiation pattern which agreed well with the simulated result.

For the bending measurement setup, a PVC pipe with a diameter of 90 mm and a plastic film was folded into a circular structure of 55 mm radius. This setup was done to hold the antenna into the circular structure. The measurement setup is depicted in Figure 12, where the antenna is measured using a 2-port measurement of Agilent Vector Network Analyser (VNA). From the observation on the graph plotted in Figure 13, the S-parameter results exhibit similar patterns for simulation and measurement but with a small divergence in the values. Meanwhile, for the 45 mm circular bend, the measured resonance frequency is unchanged irrespective of the bend axis; however, it reduced the matching and impedance bandwidth compared with the simulation. In the 55 mm circular bend, the centre frequency of both measured S11 in X- and Y-axis was shifted to 2.49 GHz, with 4% impedance bandwidth. Contrastingly, the S21 values were improved to −31 and −29 dB compared with the −26 dB in simulation for X- and Y-axis, respectively.

For on-body measurement, the antenna was attached to the human upper arm, and the antenna ports were connected to Port 1 and Port 2 of VNA for S-parameter evaluation. The results shown in Figure 14 indicate that the measured S11 was shifted to the left if compared with the simulation result for X- and Y-axes bending. The variance in the results could be a result of simulation, where the antenna was bent directly to the human skin while during the measurement, the antenna and the human skin were separated with a thin fabric layer. In view of isolation, a similar trend was demonstrated for simulation and measurement results with the lowest S21 value given by a simulation of X-bend.

The performance of the measured antenna is summarized and compared with the other related works in Table 4. The data are based on circular bending experiments, except for [20], where there was no analysis on the bending case. Also, there was no significant deviation on the finding during the on-body experiment, as reported in [21]. In contrast with the proposed antenna in this paper, the body attachment experiment shifted the S11 for about 250 MHz to a lower frequency, as stated in [22]. It should be noted that with a simple and small MIMO gap, the proposed antenna performed well within the operating frequency. On top of that, this paper investigated the performance of the wearable MIMO antenna by folded bent, which has not been reported in previous works. The on-body measured results indicated that the antenna works well with close-proximity to the human body.

4. Conclusions

This work presented a simple two-element wearable MIMO antenna working in 2.45 GHz for WBAN applications. The antenna performances were discussed in terms of S-parameter, radiation pattern, gain, ECC, and diversity gain. From the results, it could be seen that with 0.1λ MIMO spacing, the antenna performed well in circular and folded bendings. Also, there were no significant changes in antenna performance in terms of S11 for circular bending in X-axis, while better isolation was observed for the same axis bending. An isolation of at least –26 dB was observed during circular bending in both x- and y-axes. Conversely, in the case of folded bending, more variations in reflection coefficient S11 result could be seen for Y-axis bend with an almost constant value of S21. Furthermore, the proposed antenna exhibits a good diversity as the ECC is below 0.01 for all bending conditions.

Data Availability

No data were used to support this study.

Conflicts of Interest

The authors declare that they have no conflicts of interest.