Textile Bandwidth-Enhanced Coupled-Mode Substrate-Integrated Cavity Antenna with Slot

Abstract

A textile bandwidth-enhanced coupled-mode substrate-integrated cavity antenna with a slot is presented. The original coupled-mode substrate-integrated cavity antenna is of two close resonances for the odd and even coupled modes, and a rectangular slot is added on the top layer to introduce a third resonance. Parameters are optimized to merge the bands of the three resonances to realize a widened −10 dB impedance band to cover the Medical Body Area Network band, 2.45 GHz Industrial Scientific Medical band and Long-Term Evolution Band7. The proposed antenna can operate in a −10 dB impedance band of 2.32–2.69 GHz with a 14.9% fractional bandwidth according to the measurements on a fabricated prototype. Simulation and measurement results illustrate the robustness of the proposed textile antenna in the vicinity of the human body and cylindrical bending conditions. In addition, the simulated specific absorption rate of the antenna radiation in the human body is lower than the IEEE and EN limits.

1. Introduction

In the last few years, wearable technology has been investigated and widely applied in athlete training [1], medical monitoring [2], military equipment [3], security gear [4] and other areas. It plays an important role in the wireless body area network (WBAN) and is required to have several characteristics for practical applications [5]. First of all, a small profile, a light weight and a high flexibility are required for the comfort of the wearer. Secondly, the antenna should be integratable into clothes, with stable performance on the human body and in bending conditions. Textile antennas are a good candidate to meet such requirements [6,7,8,9].

The substrate-integrated waveguide (SIW) antenna [10] has been selected in the literature as a suitable candidate for textile wearable antennas due to its planar configuration with an easy integration into textile materials through embroidery [11,12]. The half-mode substrate-integrated waveguide/cavity (HMSIW/HMSIC) antenna [13,14], developed from the SIW antenna, comprises only half of the 2D footprint of the full SIW antennas, and is more suitable for wearable applications [15]. In certain applications, wearable antennas are required to operate in multiple bands including the Medical Body Area Network (MBAN) band (2.36–2.4 GHz), the 2.45 GHz Industrial Scientific Medical (ISM) band (2.4–2.835 GHz), and the Long-Term Evolution Band7 (LTE-7 band) (2.5–2.57 GHz for upload and 2.62–2.69 GHz for download) [16,17]. However, the basic SIW/HMSIW/HMSIC antennas are narrow-band due to their highly frequency-selective geometries [18,19]. Therefore, several bandwidth-enhancing strategies for the wearable/flexible SIW antenna or planar inverted-F antenna (PIFA) have been reported. In reference [20], a bandwidth-enhanced HMSIW antenna with a cork substrate was designed for wearable applications. By adding a slot in the top patch of the HMSIW antenna, two hybrid modes combining different proportions of the TM₁₁₀ and TM₁₂₀ modes were excited with close resonance frequencies to obtain an enhanced fractional bandwidth of 23.7% at 5.5 GHz. Reference [17] introduced a wearable SIW antenna with a miniaturized structure composed of two coupled eighth-mode cavities and an enhanced bandwidth of 414 MHz (16.2%) covering the 2.45 GHz ISM band and LTE-7 up- and downlink-bands. Reference [21] presented a wearable dual-band HMSIC antenna with an aperture slot to raise the bandwidth from 4.5% to 6.0% at 2.45 GHz. Reference [22] proposed an all-textile low-profile wearable PIFA with a slot and shorting pins with an enhanced fractional bandwidth of 18% at 5.5 GHz. The added slot and shorting pins, respectively, shifted the resonances of the TM_2,1/2 and TM_0,1/2 modes towards each other until two bands were merged into a widened band. Our previous work [23] also introduced two textile HMSIC antennas with added metallic/textile shorting vias to increase the lower resonance frequency towards the higher resonance frequency to realize bandwidth enhancement (14.7) at 5.5 GHz.

The coupled-mode patch antennas (CMPAs) proposed in reference [24,25,26] also provided inspiration for the bandwidth enhancement of the wearable SIW antenna. These CMPAs were of two resonances, representing the odd and even coupled modes (TM^HM_0,1,0 mode) in their two back-to-back half patches. References [27,28] (our previous works) also introduced two coupled-mode substrate-integrated cavity (CMSIC) antennas with similar back-to-back HMSIC structures and dual-band/mode characteristics. For both the CMPA and CMSIC antenna, the resonance frequencies for the odd and even coupled modes are naturally close to each other because the internal E-fields in each HMSIC follow the same distribution of the TM^HM_1,1,0 mode. It can be assumed that a larger bandwidth can be obtained when two resonance frequencies are close enough.

Based on this assumption, a textile bandwidth-enhanced CMSIC antenna with a slot is designed and presented in this paper. The proposed textile antenna combines the back-to-back HMSIC structure which leads to two resonances for the odd and even coupled modes and the added slot on the top layer which leads to a third resonance, and is of an enhanced −10 dB impedance band obtained by the fusion of three bands to cover the MBAN, 2.45 GHz ISM and LTE-7 bands. Therefore, this work provides a novel bandwidth-enhancing method for the wearable SIW antenna. Parameter analysis of several parameters is carried out through simulations. A prototype is fabricated through computerized embroidery to verify the design. The polarization, gain and efficiency of the proposed antenna across its −10 dB impedance band are studied through simulations and measurements. The performance of the proposed antenna on the human body or in bending conditions is also investigated through simulations and measurements. In addition, the specific absorption rate (SAR) in a human body phantom is simulated to evaluate the influence of the radiation of the proposed antenna on the human body.

2. Antenna Topology and Design

2.1. Baisc HMSIC and CMSIC Antennas

As shown in Figure 1a, the cavity of the basic HMSIC antenna (Antenna I) is defined by the conductive top layer, ground plane, three vertical sidewalls and a radiation aperture, and filled by the substrate material. A coaxial probe is employed as the feeding to excite electromagnetic field in the cavity. According to the classic cavity model, the top layer, ground plane and sidewalls are considered as perfect electric conductors (PECs), while the radiation aperture acts as a perfect magnetic conductor (PMC). By substituting these boundary conditions into Maxwell’s equations, the fundamental mode that can be excited in the cavity is the TM^HM_1,1,0 mode, and the deduced resonance frequency f_r is [19]:

$$f_r=\frac{c}{2\sqrt{{\epsilon }_r}}\sqrt{\frac{1}{4w^2}+\frac{1}{l^2}}$$

(1)

where c is the vacuum light speed and ε_r is the relative dielectric constant of the substrate.

As shown in Figure 1b, the basic CMSIC antenna (Antenna II) can be considered as two back-to-back HMSICs with a coupling aperture connecting two cavities. In the lower HMSIC, the fundamental TM^HM_1,1,0 mode can be excited by the feeding probe and coupled into the upper HMSIC through the coupling aperture. The coupling between two HMSICs can be electric or magnetic with different phases between radiation apertures [28].

The employed materials for these two antennas are modelled with the same physical and electrical parameters for comparison. The substrate is selective as the PF-4 foam (ε_r = 1.06, loss tangent tan δ = 0.0001) with a 3.2 mm thickness (h). A 100 mm × 100 mm ground plane is employed to guarantee the isolation between the antenna and the human body. Both the top layer and ground plane are modelled as planes with a 0.04 Ω/square sheet resistance to imitate the conductive fabric NCS95R-CR. The sidewalls are modelled as vertical planes with a 0.6 Ω/square sheet resistance [28] to imitate the three passes of the linear embroidery of the conductive threads with a 2 mm spacing between the adjacent two stiches. Through simulations, the other dimension parameters for Antenna I and Antenna II are determined as shown in

Table 1. Dimension parameters of Antenna I, Antenna II Additionally, Antenna III.

	w (mm)	l (mm)	xf (mm)	yf (mm)	gc (mm)	∆w (mm)	ws (mm)	ls (mm)	ys (mm)
Antenna I	41.8	83.6	10.8	-	-	-	-	-	-
Antenna II	39	78	11	-	24.5	2	-	-	-
Antenna III	42	84	18	13	26	9	7	25.5	1.5

.

The simulated |S₁₁| curves of Antenna I and Antenna II are presented in Figure 2. For Antenna I, the simulated fundamental mode is obtained at 2.39 GHz with a −10 impedance band of 2.35–2.43 GHz (73.5 MHz bandwidth, 3.1% fractional bandwidth). The simulated band covers the MBAN band, but does not cover other considered bands. For Antenna II, the simulated curve shows two close resonances at 2.39 GHz and 2.49 GHz with a widened –10 dB band of 2.34–2.51 GHz (171.5 MHz bandwidth, 7.1% fractional bandwidth). Compared to Antenna I, the simulated –10 dB impedance band covers the extra frequency band (2.45 GHz ISM band). However, the LTE-7 band is still not covered for Antenna II.

2.2. CMSIC Antenna with Slot

Based on the geometry of Antenna II, a CMSIC antenna with a slot (Antenna III) is proposed, as shown in Figure 1c. To be specific, a rectangular slot is added on the top layer of the lower HMSIC in Antenna II. The dimension parameters of Antenna III are determined based on simulations as shown in Table 1. The simulated |S₁₁| curve of Antenna III in free space is shown in Figure 3. Three resonances are observed at 2.39, 2.52 and 2.64 GHz. A −10 dB impedance band of 2.33–2.70 GHz with a 374.7 MHz bandwidth (14.9% fractional) is obtained. According to simulation results, all considered frequency bands are well-covered by the enhanced bandwidth of Antenna III.

The internal vector E-field distributions at the first and second resonance frequencies (respectively denoted as f_r1 and f_r2) are simulated and shown in Figure 4 to identify the coupled modes. Both of the simulated E-fields in the lower HMSIC at the first two resonance frequencies follow the distribution for the TM^HM_1,1,0 mode. At f_r1, the E-fields in two HMSICs are mirror-symmetrical. At f_r2, the E-fields in two HMSICs are axisymmetric about the y-axis. Therefore, it can be recognized that Antenna III is operating with the even and odd coupled modes at f_r1 and f_r2, respectively.

Figure 5 shows the simulated axial ratio (AR) of Antenna III across 2.33–2.70 GHz, and a 3 dB AR band around 2.60 GHz (2.58–2.62 GHz, 38 MHz bandwidth) is observed. The low AR (6 dB) at the third resonance frequency (denoted as f_r3) indicates the existence of the circular polarization. Antenna III is of a variable polarization at different frequencies within its −10 dB impedance band. Therefore, its 3 dB AR bandwidth is much smaller than its −10 dB impedance bandwidth.

Figure 6 shows the simulated internal vector E-field distributions of Antenna III at f_r3. It can be found that the E-field basically only exists in the lower HMSIC. The lower HMSIC can be approximately divided into a main and a parasitic quarter-mode substrate-integrated cavities (QMSICs) because of the added slot, and the TM^QM_1,1,0 mode is excited in the main QMSIC and coupled into the parasitic counterpart through the added slot. Between the E-fields in the main and parasitic QMSICs, there is a phase difference of about 40°. Based on the direction of the fringing E-field, the corresponding equivalent horizontal magnetic current is roughly marked as a red dashed polyline with an arrow for different phases. Both the composite vector E (blue solid line with arrow) and M (red solid line with arrow) rotate counterclockwise in the xy plane when the phase increases from 70° to 290°. This indicates that the circular polarization at f_r3 is in a right-hand form.

The maximum gain and radiation efficiency of Antenna III in free space across its −10 dB band are simulated and presented in Figure 7. In the investigated frequency range, the maximum gain and the radiation efficiency are higher than 5.5 dBi and 86%, respectively. The free-space radiation patterns of Antenna III at its three resonance frequencies are obtained through simulations, as shown in Figure 8. At the first and second investigated frequencies, the coplanar/cross polarized gain is Gain_θ/Gain_ϕ in the xz plane and Gain_ϕ/Gain_θ in the yz plane. At f_r1, the maxima of the coplanar and cross polarized gains are 7.6 and −8.3 dBi, respectively. At f_r2, the maxima of the coplanar and cross polarized gains are 8.8 and −3.8 dBi, respectively. At the third resonance frequency, however, the far-field radiation can be considered as the sum of the radiations of two parts of the equivalent magnetic current (red dashed polyline with arrow in Figure 6) along the radiation aperture of the lower HMSIC and the added slot. This leads to the increases in Gain_θ/Gain_ϕ in the xz/yz planes. Figure 8c shows that the simulated maximum of Gain_θ is higher than Gain_ϕ in the xz plane, and the simulated maximum of Gain_ϕ is higher than Gain_θ in the yz plane. Therefore, at the third resonance frequency, the coplanar/cross polarized gain is Gain_θ/Gain_ϕ in the xz plane and Gain_ϕ/Gain_θ in the yz plane, and the maxima of the coplanar and cross polarized gains are 4.7 and 0 dBi, respectively. In addition, the maxima of the right-hand circular polarized (RHCP) and left-hand circular polarized (LHCP) gains at f_r3 are 5.3 and −1.1 dBi, respectively. This further proves that the circular polarization at 2.64 GHz is in a right-hand form.

2.3. Parametric Analysis

Parameter w and l are key parameters determining the resonance frequencies of the HMSIC antenna (Antenna I) and the CMSIC antenna (Antenna II), and thus also determine the first two resonance frequencies of Antenna III. Parameter x_f and y_f determine the impedance matching based on the basic knowledge of antenna design. Apart from these parameters, whose influences are well-known, simulations are carried out for Antenna III to study the effects of other parameters, namely g_c, ∆w, l_s, w_s and y_s.

Figure 9a shows the simulated |S₁₁| curves of Antenna III with different lengths of the coupling aperture g_c. When g_c increases, the f_r2 and f_r3 basically remain unchanged, and the antenna impedance remains well-matched at these two resonances. The first resonance frequency, however, is decreased with an increasing g_c, and the −10 dB impedance band is increased.

Figure 9b shows the simulated |S₁₁| curves of Antenna III with different values of parameter ∆w. f_r3 suffers no influence from the change in ∆w in terms of the frequency and the impedance matching. With an increasing ∆w, both f_r1 and f_r2 increased, and the −10 dB impedance bandwidth decreased. This can be explained by the decrease in the cavity size in two HMSICs. In addition, the impedance matching at f_r1 and f_r2 is significantly affected by the change in ∆w.

Figure 9c shows the simulated |S₁₁| curves of Antenna III with different lengths of the added slot l_s. f_r1 and f_r2 are basically not affected by the change in l_s in terms of the frequencies and the impedance matching. When l_s increases, f_r3 is decreased without deteriorating the impedance matching, and the −10 dB impedance band is narrowed. The decrease in f_r3 can be explained by the increase in the radiating edge for the third resonance due to the increase in l_s.

Figure 9d shows the simulated |S₁₁| curves of Antenna III with different widths of the added slot w_s. When w_s is increased, f_r1 and f_r3 are slightly increased, and the −10 dB bandwidth basically stays unchanged. The increases in resonance frequencies can be explained by the shrinking of the effective cavity sizes in the lower HMSIC and the lengths of the radiating edges of the main and parasitic QMSICs due to the increase in w_s. f_r2 basically stays unchanged. In addition, the impedance matching at three resonance frequencies can be affected by the change in w_s.

Figure 9e shows the simulated |S₁₁| curves of Antenna III with different values of parameter y_s. The change in y_s leads to slight shifts in all three resonance frequencies and noticeable deterioration of the impedance matching at resonance frequencies.

2.4. Mutual Effects between Antenna and Human Body

In practical applications, the proposed textile antenna needs to be integrated into clothes and located in the vicinity of the human body. Therefore, it is important to study the mutual influence of the proposed antenna and the human body. A 300 × 300 × 40 mm³ four-layer (skin, fat, muscle and bone) phantom is modelled in simulations to imitate the human body, and the antenna is placed above the top center of the phantom with a 5 mm gap, as shown in the inset of Figure 3. The characteristics and thicknesses for each layer of the phantom at 2.4 GHz are obtained from reference [29], as listed in

Table 2. Characteristics and thicknesses of each layer of skin, fat, muscle and bone.

	Relative Permittivity	Conductivity (S/m)	Density (kg/m3)	Thickness (mm)
Skin	38.09	1.43	1100	2
Fat	5.29	0.1	910	5
Muscle	52.8	1.69	1041	20
Bone	11.4	0.381	1908	13

.

The simulated |S₁₁| curve of Antenna III with the human body is shown in Figure 3. Compared to the free-space scenario, f_r1 and f_r2 remain unchanged while a slight shift is observed for f_r3 (2.64 GHz in free space vs. 2.63 GHz on the human body). The −10 dB impedance band with the human body is still 2.33–2.70 GHz, and covers all of the considered frequency bands.

The maximum gain and radiation efficiency of Antenna III with the human body across its −10 dB band are simulated and presented in Figure 7. In the whole −10 dB impedance band, the maximum gain and radiation efficiency are above 5.6 dBi and 69%, respectively. Compared to the scenario without the human body, the radiation efficiency is reduced due to the absorption of radiation by the human body.

The radiation patterns of Antenna III on the human body at its three resonance frequencies are simulated and presented in Figure 10. The definitions of the coplanar and cross-polarized gains at each investigated frequency remain the same as in the free-space scenario. At 2.39 GHz, the maxima of the coplanar and cross polarized gain are 6.7 and −8.6 dBi, respectively. At 2.52 GHz, the maxima of the coplanar and cross-polarized gain are 8.6 and −2.6 dBi, respectively. At 2.63 GHz, the maxima of the coplanar and cross-polarized gain are 3.6 and 3.1 dBi, respectively, and the maxima of the RHCP and LHCP gains are 6.3 and −1.2 dBi, respectively. In addition, the backward radiations of all patterns on the human body are dramatically reduced compared to those in free space due to the reflection of the radiation by the human body.

The SAR is simulated and analyzed to investigate the influence of the radiation of Antenna III on the human body. The simulated 1 g average SAR distributions in the four-layer phantom at its three resonance frequencies with a 0.5 W input power are presented in Figure 11. The simulated SARs are below 0.61, 0.55 and 0.86 W/kg at 2.39, 2.52 and 2.63 GHz, respectively, and satisfy the limits of IEEE C95.1-2005 (<1.6 W/kg) and EN 50361-2001 (<2.0 W/kg). When the gap between the human body and the antenna reduces to 1 mm, the corresponding maxima of the SAR increase to 0.79, 0.69 and 1.07 W/kg, which are still well below the mentioned limits. In practical applications, the SAR can be further restricted by employing a larger ground plane, which can be made from conductive fabric or conductive threads weaved into the garment. Therefore, the safety of Antenna III on the human body is guaranteed.

3. Prototype Fabrication and Measured Results

A prototype of Antenna III is fabricated following the procedures listed below. The photos of the fabricated prototype are shown in Figure 12a,b.

• Design the pattern for the embroidered sidewalls with the PE-DESIGN software.
• Cut the PF-4 foam sheet and conductive fabric NCS95R-CR into designed shapes for the substrate, top layer and ground plane, respectively.
• Glue the top layer and the ground plane onto the top and bottom surface of the substrate.
• Embroider the sidewalls with the designed pattern using the computerized embroidery machine Brother NV950.
• Install the SMA connector with conductive epoxy CircuitWorks CW2460.

A vector network analyzer Tektronix TTR506A (as shown in Figure 12c) and an anechoic chamber are used to investigate the |S₁₁| curves and coplanar/cross polarized gain patterns of the prototype. Because of the limited equipment, the radiation efficiency, AR curves, AR bandwidth, RHCP/LHCP gain patterns and SAR on the human body are not measured and presented in this paper.

The measured |S₁₁| curves of Antenna III with/without the human body are shown in Figure 3. The measured |S₁₁| curves show a high agreement with simulated curves in terms of the resonance frequencies and the −10 dB impedance bands. In free space, the measured resonances are obtained at 2.39, 2.51 and 2.63 GHz. On the human body, the measured resonance frequencies are 2.39, 2.52 and 2.63 GHz. The measure −10 dB impedance band is 2.32–2.69 GHz (14.9% fractional bandwidth) for both free-space and on-body scenarios, and all of the considered frequency bands can be covered.

The |S₁₁| curves of Antenna III with cylindrical bending conditions are also obtained through measurements and are shown in Figure 13. A polyvinyl chloride (PVC) frame is fabricated to provide a cylindrical surface with a 5.5 cm radius to imitate the situation in which the antenna is fixed on the clothes and bent around the limbs or the shoulder. During measurements, the prototype is fixed onto the PVC frame using paper tapes as shown in the insets of Figure 13. With a bending condition around the x-axis, the measured resonance frequencies are 2.39, 2.49 and 2.60 GHz, and the measured −10 dB impedance band is 2.27–2.69 GHz, which still covers the MBAN, 2.45 GHz ISM and LTE-7 bands. With a bending condition around the y-axis, however, the impedance matching become worse at the f_r2 and f_r3, and the measured −10 dB band shrinks to 2.30–2.66 GHz which does not fully cover the LTE-7 band. This measured result contributes to the strategy of location and direction of the proposed textile antenna on the human body.

The measured patterns of the coplanar and cross polarized gain of Antenna III in free space at its resonance frequencies are shown in Figure 8. The measured coplanar polarized gain patterns are in good agreement with the simulated patterns. The maxima of the measured coplanar polarized gain at 2.39, 2.51 and 2.63 GHz are 7.0, 7.6 and 4.0 dBi, respectively. The minor differences between the simulated and measured |S₁₁| curves and maxima of the coplanar polarized gain can be explained by slight fabrication/measurement inaccuracies. Unfortunately, the radiation efficiency and the patterns of the LHCP/RHCP gains are not measured in this work due to the limited equipment.

4. Conclusions

A textile bandwidth-enhanced CMSIC antenna with a slot has been designed and fabricated to cover the MBAN, 2.45 GHz ISM and LTE-7 bands in wearable applications. Based on the basic CMSIC antenna with two close resonances, a third resonance is introduced by the added slot on the top layer, and parameters are optimized to merge three resonances into a wide −10 dB impedance band. The simulated −10 dB impedance band of the proposed antenna is 2.33–2.70 GHz, satisfying the design requirements. Simulations of the internal vector E-field, AR and radiation patterns indicate that the proposed antenna is linearly polarized at the first and second resonance frequencies and right-hand circularly polarized at 2.64 GHz. Additionally, the maximum gain and radiation efficiency of the proposed antenna in free space are above 5.5 dBi and 86%, respectively, across the −10 dB impedance band based on simulation results. Simulations of the proposed antenna on a four-layer human body phantom prove its stable coverage over the considered frequency bands and safety on the human body. A highly flexible textile prototype of the proposed antenna is fabricated using computerized embroidery. The measured results of the prototype show a satisfactory agreement with simulation results and verify the design. In addition, the cylindrical bending conditions in two different directions are investigated for the proposed antenna through practical measurements.

The comparison between the proposed antenna and other flexible bandwidth-enhanced SIW antennas and PIFAs is shown in

Table 3. Comparison with other wearable/flexible bandwidth-enhanced SIW antennas and PIFAs.

	Freq. (GHz)	Bandwidth (%)	Gainmax (dBi)	Efficiency (%)	Cavity Size (λε2)
[20]	5.5	23.7	4.3	85	0.364
[17]	2.45	16.2	4.7	67~82	0.096
[21]	2.45	6	not given	not given	0.220
[22]	5.5	18	5.9	74.1	0.662
[23]	5.5	14.7	8.8	≥95	1.073
This work	2.5	14.9	5.5~9.1	86~96	0.466

. The bandwidth of the proposed antenna is lower than [17,20,22] because of its low-loss PF-4 foam substrate [30]. However, the lower overall loss of the proposed antenna also leads to the higher radiation efficiency and maximum gain. The back-to-back HMSIC structure of the proposed antenna leads to a mediocre cavity size among all candidates. However, this work provides a new solution for the bandwidth enhancement on the wearable SIW antennas by employing the CMSIC structure. In addition, the proposed antenna shows variable radiation patterns and polarizations at different frequencies in the −10 dB band, and thus can be potentially reused for multiple applications in on-body/off-body WBAN communications.