A Hybrid Mutual Coupling Reduction Technique in a Dual-Band MIMO Textile Antenna for WBAN and 5G Applications

This paper presents a hybrid mutual coupling reduction technique applied onto a dual-band textile MIMO antenna for wireless body area network and 5G applications. The MIMO antenna consists of two hexagonal patch antennas, each integrated with a split-ring (SR) and a bar slot to operate in dual-band mode at 2.45 GHz and 3.5 GHz. Each patch is dimensioned at <inline-formula> <tex-math notation="LaTeX">$47.2 \times 31$ </tex-math></inline-formula> mm<sup>2</sup>. This hybrid technique results in a simple structure, while enabling significant reduction of mutual coupling (MC) between the closely spaced patches (up to <inline-formula> <tex-math notation="LaTeX">$0.1\lambda$ </tex-math></inline-formula>). This technique combines a line patch and a patch rotation technique, explained as follows. First, a line patch is introduced at an optimized distance to enable operation with a broad impedance bandwidth at both target frequencies. One of the patches is then rotated by 90° at an optimized distance, resulting in a significant MC suppression while maintaining the dual and broad impedance bandwidth. The proposed MIMO antenna is further evaluated under several bending configurations to assess its robustness. A satisfactory agreement between simulated and measured results is observed in both planar and bending conditions. Results show that the MIMO antenna achieves an impedance bandwidth of 4.3 % and 6.79 % in the 2.45 GHz and 3.5 GHz band, respectively. Moreover, very low MC (<inline-formula> <tex-math notation="LaTeX">$S_{21} < -30$ </tex-math></inline-formula> dB) is achieved, with a low (< 0.002) envelop correlation coefficient, and about 10 dB of diversity gain at both desired frequencies using this technique. Even when bent at an angle of 50° at the <inline-formula> <tex-math notation="LaTeX">$x$ </tex-math></inline-formula>- and <inline-formula> <tex-math notation="LaTeX">$y$ </tex-math></inline-formula>-axes, the antenna bent maintained a realized gain of 1.878 dBi and 4.027 dBi in the lower and upper band, respectively. A robust performance is offered by the antenna against the lossy effects of the human body with good agreements between simulated and measured results.


I. INTRODUCTION
Antennas are increasingly being developed in a more compact format and on flexible materials. This is especially The associate editor coordinating the review of this manuscript and approving it for publication was Chan Hwang See.
attractive in wearable applications, as these antennas are used in many applications such as health monitoring, tracking and etc. [1]. These applications also demands that these antennas can operate across a wide range of frequencies, within the framework of the Internet of Things (IoT). Besides that, the combined IoT connectivity to high speed fifth generation (5G) systems will be advantageous in demanding systems such as healthcare using a single set of antenna. Moreover, flexible antennas can be worn while the users move freely, without affecting their morphologies requires these antennas to be wide in bandwidth. Moreover, realtime information transmission with minimal losses needed in these systems when applied for managing critical illnesses and procedures also adds to the challenge in flexible antenna design.
Considerable published work has reported on the effects of mutual coupling in dual band antenna arrays. However, the alleviating the effects of coupling (S 21 ) for dual-band MIMO textile antennas, specifically operating at 2.45 and 3.5 GHz is hardly discussed. This research proposes a unique hybrid technique to suppress the mutual coupling of the 2×1 MIMO antenna. This is aimed at achieving an S 21 of less than -30 dB with the closest possible placement of elements (an edge-toedge spacing of 0.1λ). This antenna is benchmarked against the performance of the same antenna with and without applying the proposed hybrid technique. Finally, the robustness of the proposed MIMO antenna against bending deformation is evaluated thoroughly. This paper is organized as follows. Section II outlines the characterization of the antenna element, after which a study on the mutual coupling effects antenna is presented. The three main parameters are analyzed and presented in three parts: distance analysis, mutual coupling reduction and gain, and envelope correlation coefficient. The measurement results of the prototype are presented in Section IV prior to the conclusions in Section V.

II. ANTENNA CHARACTERIZATION
The proposed antenna is designed to operate in dual band mode, centered at 2.45 for the wireless body are network (WBAN) lower band, and at 3.5 GHz for 5G as the upper band. Felt textile is used as the substrate and is sandwiched between top radiator and a full ground plane. It has a relative permittivity (ε r ) of 1.44, a loss tangent (tan δ) of 0.044, and a thickness (H ) of 3 mm. The conductive elements are formed using ShieldIt Super electro-textile from LessEMF Inc., which is 0.17 mm thick and features an estimated conductivity (σ ) of 1.18 × 10 5 Sm −1 . As an initial step, a rectangular-shaped patch radiator is designed to operate in the lower band, centered at 2.45 GHz, as shown in Figure 1.
Then, its design is modified with a slotted ring based on [26] to produce another resonant frequency at 3.5 GHz. The SR shaped slot in the middle of the rectangular-shaped patch antenna enables the bandwidth broadening of the upper 3.5 GHz band. The feeding points are placed on the bottom left edge for first element, labelled as Port 1, and bottom right edge for second element, labelled as Port 2, as illustrated in Figure 1. A detailed design procedure is presented in [27] which can be summarized in four steps, as follows: • First, the dimensions of the patch without SR-shaped slot are calculated based on the upper band resonance.
• Second, the probe feed structure is optimized to obtain a suitable matching in the upper band.
• Third, the SR shaped and the bar slot are added to broaden the bandwidth, and to provide operation in the respective bands.
• Fourth, the dimensions of the SR-shaped slot are tuned to provide operation in the respective lower band   as shown in Figure 1. All simulations and optimizations are performed using CST Microwave Studio software.

III. MUTUAL COUPLING
In this section, the distance between and orientation of the patch elements are studied to optimize the proposed antenna in terms of reflection coefficient and mutual coupling. Details of this study are discussed in the next subsections.

A. DISTANCE ANALYSIS
The distance between the patch elements affects the antenna performance in terms of reflection coefficient (S 11 ) and mutual coupling (S 21 ). The distance between antennas is varied from 0.5λ to 0.1λ, as illustrated in Figure 2. The results shown in Figure 3 indicated that the S 11 and bandwidth of the antenna is preserved in both bands with the variation of antenna gap. However, as shown in Figure 3(b), the S 21 values increased with decreasing distance, indicating higher coupling between the patches. The coupling increases from

B. MUTUAL COUPLING REDUCTION TECHNIQUE
A hybrid technique involving a line patch and rotation of antenna elements is then applied in this subsection. In the first step, a line patch of WL x LL is introduced at the optimized guided distance (D) of 0.1λ, as shown in the Figure 3. Next, the width (WL) and the length (LL) of the line patch are studied. They are varied as follows: WL = 0.73, 2, 4, 6, 8, and 10 mm, and LL = 40, 50, and 60 mm. Aimed at obtaining an optimized length and width of the line patch, the results are shown in Figure 4. While the size of the line patch did not affect the S 11 , the S 21 of the antenna varied with the variation of WL and LL, particularly at higher frequencies. From the analysis, the final optimized WL and LL value is 0.73 mm and 60 mm,  respectively. The next step in the proposed hybrid technique is by rotating the antenna elements to arrive at the final MIMO antenna design. The different rotations are illustrated in Table 1 and the optimized MIMO antenna is produced by    patch improved S 21 significantly, up to 60 % and 33 % at 2.45 GHz and 3.5 GHz, respectively.
On the other hand, Figure 6 illustrates the surface current distribution when one of the ports is excited. As seen in this figure, a single technique of either only adding the line patch in between the antenna elements or rotating the patch element reduced the current interaction with the other patch element. However, in both cases, part of the current still overflows to the adjacent patch. Combination of both techniques significantly reduced the coupling between the antenna elements.

C. BENDING
A comprehensive analysis on the effects of the bending on the proposed MIMO antenna is presented in this section. Simulations of the bending evaluation curvatures are performed at different angles (α) of 10 • , 20 • , 30 • , 40 • , and 50 • , which translates to 24.38, 30.48, 40.6, 69.8, and 121.9 mm radii, respectively, based on [28]. These bending values are selected to emulate the curvature of proposed MIMO antenna when wrapped around the arm in a regular body. Bending is investigated at two conditions, when bent at x-and y-axes for five different bending angles, as illustrated in Figure 7. The extreme condition is identified when the antenna is bent at y-axis with smallest angle/shortest radius, α = 10 • @ 24.28 mm. Measurements are then performed to observe The results obtained from the bent antennas are compared with simulations in flat condition, as illustrated in Figure 8. Decreasing the bending degree from 50 • to 10 • lowers the resonance in both bands, with a more significant change in the upper band. In contrast, different mutual coupling behavior can be observed when bent at the x-and y-axes. When bent at the x-axis at 2.45 GHz, lower S 21 is seen with increasing bending degrees. This behavior is contrary at 3.5 GHz. On the other hand, when varying the bending degree at y-axis, the S 21 fluctuates in the lower band, but is almost consistent in the upper band. As expected, bending at an angle of 50 • resulted in high mutual coupling at both frequency bands. Hence, it can be concluded that bending of the antenna at different degrees affected particularly the performance at the higher frequencies.

D. GAIN, RADIATION EFFICIENCY AND CORRELATION ANALYSIS
The proposed MIMO antenna is evaluated in terms of envelope correlation coefficient (ECC), diversity gain (DG), channel loss capacity (CCL) and total active reflection coefficient (TARC). The correlation between antenna elements is described by the ECC (ρ e ) and the diversity gain. They are used to evaluate the correlation levels of the channels [29], and is calculated using equation (1), as follows: A low ECC value indicates minimal correlation between antenna elements. Similarly, diversity gain (DG) is dependent on the spatial correlation coefficient between the patch elements. A low ECC (< 0.5) leads to high diversity gain, and both are related by equation (2), as follows: The simulated and measured ECC within the frequency of interest is presented in Figure 9. ECC at all resonant frequencies are below 0.05 in both flat and bent conditions, and satisfies the minimum (< 0.5) diversity criteria [21]. A low ECC leads to high diversity gain, which is demonstrated by the plot in Figure 10. For an ECC value of less than 0.1, the diversity gain is almost 10 dB.
On the other hand, CCL is the estimated maximum message transmission which can take place without any loss in the communication channel. The acceptable rate should be less than 0.4 bits/s/Hz. Calculated using equations (3) to (5), the CCL result is presented in Figure 11.
It shows that the proposed MIMO exhibits acceptable CCL for all bending conditions with varying operating frequency. and Another evaluated parameter for this antenna is TARC, defined as the ratio of reflected and incident power for a MIMO antenna system. For a two-port MIMO antenna, TARC is calculated using equation (6) and must be below −0 dB. For the proposed MIMO antenna at both operating VOLUME 9, 2021 frequencies, the TARC are below −5 dB as seen in Figure 12.
Radiation pattern of the proposed MIMO antenna is shown in Figure 13. It is observed that the main lobe of the radiation pattern is maintained while there are increasing in back lobe when the bending degree is decrease from 50 • to 10 • in both the lower and upper operating bands of the antenna. On the other hand, significant variation in the radiation patterns of the antenna is seen when bent at the y-axis. As the bending degree is reduced, the main lobe direction tilted to the left, with slightly higher back lobes pattern.

F. SPECIFIC ABSORPTION RATE (SAR) ANALYSIS
The SAR values for the proposed antenna are calculated using CST MWS by mounting the proposed antennas in proximity of a truncated Hugo human body model (on the upper arm). The proposed antenna is placed 1 mm away from these models, as seen in Figure 14. The SAR distributions averaged over 10 g of tissue are then calculated at 2.4 GHz and 3.5 GHz for this antenna with an input power 1 W when placed on the left upper arm. SAR levels for this antenna in planar condition presented in Figure 10 indicate that the maximum 10 g SARs are observed to be 0.0283 W/kg and 0.0162 W/kg at 2.45 GHz and 3.5 GHz, respectively. These simulated SAR results are verified against the measured SAR of the antennas in [30], which used the same textile materials and full ground plane as  the proposed MIMO antenna. The maximum 10 g measured SARs in [30] are 0.1 W/kg and 0.5 W/kg at 2.45 GHz and 5.2 GHz, respectively. A satisfactory agreement between the simulated and measured SAR is observed. Due to the use of the full ground plane, the SAR values for antenna in this antenna did not exceed 0.1 W/kg in both bands.

IV. EXPERIMENTAL EVALUATION RESULTS
The proposed MIMO antenna is then fabricated and experimentally assessed in the planar condition and when bent at both axes, as shown in Figure 15 (with 20 • of bending radius). The measurements are performed using Keysight Technologies E5071C E-series Vector Network Analyzer (VNA). A 50-coaxial cable has been used to connect SMA to  the VNA for measurements. Their S 11 and S 21 results are presented in Figure 16, with the solid lines representing the simulated performance, whereas measurement are represented by the dashed lines. The simulated S 11 for the proposed antenna in planar form are observed to be consistent with measurements in free space, as illustrated in Figure 16(a), except for a slight upwards shift in the lower band. Satisfactory agreements are also seen between measured S 11 for all bending configurations at y-axis, including their bandwidths. However, when bent at the most extreme condition (α = 10 • at y-axis), the proposed MIMO antenna showed a downwards shift in the lower band. In planar condition, its measured S 21 is about −35 dB in both bands, with improvements of 6 dB and 10 dB at 2.45 GHz and 3.5 GHz, respectively. On the other hand, t is observed that the measured S 21 is less than −30 dB when the antenna is bent at the y-axis for all bending conditions in the lower and upper bands. This indicates that the MC is reduced significantly even in the extreme bending condition. This validates the design's robustness against any y-axis bending and maintained its dual band characteristic.  Table 2 summarizes the performance of the proposed MIMO antenna in terms of S 11 , S 21 , impedance bandwidth, realized gain, radiation efficiency, and directivity when operating in flat condition in free space at 2.45 GHz and 3.5 GHz. As evident from these results, satisfactory performance for all parameters are observed in free space. A small difference exists between simulated and measured results due to the potential fabrication inaccuracies, the inhomogeneous thickness of the textile layers and inhomogeneous dielectric properties. The simulated and measured 2-D radiation patterns for the proposed antenna at 2.45 GHz and 3.5 GHz presented in Figure 17 indicate directional patterns with small back lobes. A good agreement between the simulated and measured radiation patterns have been observed.
Besides simulations, the prototype is measured in proximity of the human body on the chest and upper arm, as shown in Figure 18(c) and Figure 18(d), respectively. Comparison between simulated and measured S 11 and S 21 on the chest and upper arm are summarized in Figure 19. The impact of the human body on the antenna is minimal due to the shielding against coupling provided by the full ground plane. Measured impedance bandwidths of 6.95 % and 7.11 % are achieved in the lower and upper bands, respectively, when measured on the chest.
Meanwhile, when placed on the upper arm, the measured bandwidth is 7.78 % and 9.15 % in the lower and upper bands, respectively. Measured S 21 are consistently less than −30 dB when the antenna is mounted on the chest and upper arm in both lower and upper bands. A good agreement between the on-body simulated and measured S 11 and S 21 is seen, with small marginal shift observed due to nonidealities in the experimental environment. The low MC exhibited by the proposed antenna makes it suitable for off-body MIMO in WBAN and 5G applications.
In summary, Table 3 compares the performance of the proposed MIMO antenna with previous 1×2 MIMO antennas in terms frequency, flexibility, antenna size, technique, S 21 , and gap between elements. One of the most similar work in [8] presented a multiband wearable MIMO antenna with a comparable 0.1λ 0 inter-element gap with the proposed design. However, the metamaterial technique applied to the structure result in a more complex structure. It is also worth noting that the proposed work is the first work proposed on wearable MIMO antenna operating at 2.45 GHz and 3.5 GHz designed using a hybrid method to result in a relatively simple and compact structure. Besides the extensive validation on antenna deformation, the proposed hybrid technique also resulted in less than 30 dB of S 21 and a very small inter-element gap (0.1λ 0 ). Such method can potentially be applied to design MIMO antennas in space-constrained mobile devices.

V. CONCLUSION
This study proposes a hybrid method of mutual coupling reduction applied in designing a textile MIMO antenna for on body applications. This antenna is designed by combining two octagonal structures each integrated with a SR and bar slot. Mutual coupling of the MIMO antenna is significantly reduced by rotating the patch element and adding a line patch between the antenna elements. Most importantly, the resulting optimized structure is simple and can be implemented as a textile antenna. Due to this, the agreement between simulations and measurements is satisfactory. Moreover, evaluation of this antenna under different degrees of bending and bending axes indicated robust performance, with minimal changes in terms of reflection coefficient, mutual coupling, and radiation characteristics. Further assessments of this antenna in terms of MIMO parameters such as ECC, DG, CCL and TARC also validated that this antenna can be potentially applied in the next generation of 5G wearable devices.  Dr . Soh is a member of the IET and URSI. He volunteers in the IEEE MTT-S Education Committee. He is also a Chartered Engineer Registered with the U.K. Engineering Council. He was a recipient of the IEEE AP-S Doctoral Research Award, in 2012, the IEEE MTT-S Graduate Fellowship for Medical Applications, in 2013, and the URSI Young Scientist Award, in 2015. He was also the Second Place Winner of the IEEE Presidents' Change the World Competition, in 2013. Three of his (co)authored journals were awarded the IEEE AP/MTT/EMC Malaysia Joint Chapter's Publication Award, in 2018, 2019, and 2020, and another two journals were also awarded the CST University Publication Award, in 2011 and 2012.