Compact BLE Antenna With a Modified PIFA Configuration for Wearable EMG Monitor

An antenna mounted underneath the top cover of a wearable device, fed by an ultralow-profile connector, is proposed for Bluetooth-low-energy (BLE) communication at 2.4 GHz. As the antenna is to be used for a wearable device, it is essential that it should be compact in size and tolerant against the nearby medium change. The modified planar inverted-F antenna (PIFA) that we propose consists of a main patch and a parasitic element to broaden the antenna bandwidth. The end of the parasitic element is shorted to miniaturize the antenna to $25\times10.8$ mm, fitting well inside a watch-type wearable device. Also, the complete ground layer of the antenna makes it radiate well in the outward direction, and minimally interact with the back-side medium. This results in the reflection coefficient being insensitive to the medium change on the backside. The proposed antenna has a peak gain of 3.62 dBi along with 20% efficiency and impedance bandwidth of 80 MHz (2.4–2.48 GHz). To examine the communicative operation of the antenna in practice, the received signal strength indicator (RSSI) of the complete prototype device with the antenna is measured in various postures and orientations, demonstrating reliable connectivity within a typical indoor distance of 10 m. Lastly, the antenna is embedded in a wearable device, demonstrating electromyography’s wireless monitoring.

on-body communication was reported for the operation in the ISM band [10]. The antenna has a complete ground plane to reduce the back radiation. However, the antenna is a multilayered structure and the size is 37 × 30 mm 2 , making the antenna expensive for fabrication and burdensome to be embedded inside a watch-type device. Recently, metamaterials and metasurfaces are also actively adopted to realize compact antennas with high performances [12], [13], [14], [15], [16], [17], [18]. They are the artificial structures composed of subwavelength macro cells with the advantage that their effective medium properties can be manipulated. Nevertheless, compact antenna design with high reliability for wearable devices lacks attention in the literature yet.
This article presents an antenna design for the watch-typed wearable device for BLE communication. Fig. 1(b) shows the cross section of the wearable device, presenting various electronic components, such as a battery, electronic circuitry, and the antenna mounted underneath the top cover of the device. All the components are enclosed inside the graycolored frame, which consists of top and bottom covers, with its total size as compact as 50 × 40 × 12 mm 3 . The proposed antenna fits well within such a small area and is fed by a commercial, ultrasmall connector to ease the assembly with the main circuitry board. The antenna has a complete ground layer that ensures the antenna to operate whether the medium under the device is air or body tissue, making the antenna suitable for wearable purposes.
To confirm the operation of the antenna for communication, the antenna is included in a complete prototype device and the received signal strength indicator (RSSI) is measured to check the connectivity [19], [20]. We show the RSSI levels higher than a typical receiver sensitivity of −95 dBm for various postures and orientations of the device within a typical indoor distance of 10 m. The antenna is integrated with a wireless electromyogram (EMG) monitoring device [21], [22]. It is demonstrated that the activity of the forearm muscle can be wirelessly monitored over the BLE communication.
The article is organized as follows. The studies of antenna design and its characterizations with versus without tissue are discussed in Section II. With the communication module equipped with the proposed antenna, the wireless connectivity for various environments is discussed in Sections III and IV. Finally, the article is concluded in Section V.
II. ANTENNA DESIGN STUDIES AND CHARACTERIZATIONS To design the antenna for wearable devices, the size consideration of the antenna to fit in a compact device is an essential part of the investigation. Also, the antenna needs to radiate well in the surrounding directions except for the direction toward the body tissue to minimize the medium dependency. Fig. 2 shows the two antennas designed and investigated for this work. The first antenna in Fig. 2(a) shows a folded meander dipole antenna as a reference one with a size of 18.5 × 4.1 mm 2 . Over the substrate of FR4 with a relative permittivity of 4.4 and 0.2-mm thickness, the detailed dimensions of the reference antenna are tabulated in Table I. Due to its small size (∼0.15λ 0 ), this antenna structure is widely adopted for miniaturized mobile devices [2], [3], [4]. One of the key factors to reduce the size is that one arm of the antenna is folded on the other side of the other arm [7]. The radiation pattern is almost omnidirectional as that of a regular dipole antenna. As a result, when the reference antenna is located in the wearable device, it largely interacts with and is affected by a medium that may be placed under the wearable device.
In contrast, the proposed antenna has a complete ground plane on the bottom layer as shown in Fig. 2(b). The same substrate material with the same thickness as the reference antenna has been used for the proposed one. The proposed antenna is designed to be minimally affected by the medium behind as shall be demonstrated by simulations and measurements.

A. Antenna Design Procedure
The antenna design follows the concept of the planar inverted-F antenna (PIFA), which reduces the antenna size by adopting a shorting pin near the feeding pin [23]. The proposed design of modified PIFA is considered to be installed on the inner side of the device top cover making its effective radiation toward the outside of the device.
The available size and location for the antenna underneath the device top cover can be seen in Fig. 3(a), denoted by the orange-colored borderline. The allowable size on the  bottom was approximately 26.5 × 12 mm 2 . Here the proposed antenna is fed with the U.FL connector, a type of ultrasmall connector produced by Hirose Electronics, so that the antenna can be easily connected to the device circuits with a low profile [24]. The footprint for the U.FL adaptor followed the guideline provided by the product (U.FL-R-SMT-1) shown in Fig. 3(b). The connection of the antenna feed via the regulated footprint eases the stable soldering and assures the robust operation of the antenna. Considering the footprint, the area of 5 × 4.7 mm 2 was additionally reserved for the port connection at the top-left corner of the antenna as seen in Fig. 3(a).
Upon the size specification allowed for it, the antenna is designed with a step-by-step procedure considering the plastic cover of 0.5-mm thickness and a conducting layer in the PCB circuitry. 1) Step I: A rectangular patch was made on one side of the substrate and another side was used as a complete ground plane. The initial dimensions of the patch are 23 × 7 mm 2 . Based on the well-known transmission-line model for the rectangular patch antenna design [25], the feeding point was moved along the center line as can be seen in Fig. 4 to have the impedance matching at 3.32 GHz, which is near the desired operating frequency (2.45 GHz). 2) Step II: To improve the impedance matching and bandwidth, a parasitic element is added as seen in Fig. 4(b). When the structure is simulated using a commercial electromagnetic solver (the CST Microwave Studio), this choice yields a better impedance matching with a small frequency shift of the resonant frequency to 3.2 GHz. The bandwidth enhancement is due to the two similar current paths produced by the patch itself and by the added parasitic element [26], [27], [28], [29], [30]. 3) Step III: To reduce the size of the antenna the shorting pin method is used [31]. The shorting pin is made at the end of the parasitic patch in Fig. 4(c). The S-parameters for the three steps are plotted in Fig. 4(d). It can be seen that the resonance frequency appears close to the target frequency.
The effect of shorting pin location on the resonance behavior of the antenna is investigated in Fig. 5(a). As the shorting pin moves from point A to F, the length of the current path becomes longer to decrease the resonant frequency, which is clearly seen from Fig. 5(b). 4) Step IV: The elaboration of the feed point design to connect the U.FL port is made. Fig. 6(a) shows the top view of the complete antenna on a substrate with 100% transparency to show both the upper patch and the pattern on the ground plane for the U.FL connection. The signal pad is connected to the feed line with the via connection. The feed line is extended from the edge toward the inside of the main patch by creating a slot to make an inset feed configuration. As the depth of the inset goes deeper than the initial position of the feed point in Step III, the resonance behavior of the antenna changes as Fig. 6(b). It is observed that the position of the feed point effectively changes the current path length and can be used as a knob to shift the resonant frequency. 5) Step V: After the antenna configuration is determined, the parametric study is carried out to finalize the antenna geometry. For this purpose, the feeding slot width S w , the feed linewidth w f , the gap between the main patch and the parasitic patch wn 2 , and the shorting pin position S n shown in Fig. 7 are chosen as the parameters to tune the resonance frequency and the bandwidth. When those parameters vary, the S-parameter results are shown in Fig. 8(a)-(d). As the feeding slot width S w and feeding linewidth w f increases from 1.3 to 1.7 mm and 0.3 to 0.7 mm, the resonant frequency rarely changes but the impedance matching improves at certain cases   in Fig. 8(a) and (b). A large frequency shift can be observed when the gap variation wn 2 changes from 0.2 to 0.6 mm and also when the shorting pin position S n does from 1.0 to 4.0 mm. The above study helps to fine-tune the resonance behavior of the antenna at the required frequency band.
The simulation model to consider the tissue medium below the wearable device embedding the antenna is studied after the antenna optimization as shown in Fig. 9(a). It is observed that due to the shielding by the ground plane of the antenna itself and the conducting sheet of the PCB circuitry, the resonance behavior remains the same in Fig. 9(b). The final dimensions of the proposed antenna are summarized in Table II.

B. Measurement Results of the Proposed Antenna
The reference and the proposed antenna are fabricated as shown in Fig. 10(a) and (b), respectively. The reference antenna is vertically placed at the edge of the wearable device as shown in Fig. 11(a)     of a visual interface for the device. Most importantly, the complete ground plane on the other side of the two-layered antenna can effectively block the back radiation, and hence the antenna performance is less affected by the medium beneath the wearable device.
To evaluate and compare the effect of a nearby medium for two antennas, the antennas are installed inside the wearable device (see Fig. 11) and the S-parameters are measured ON and OFF the body. A vector network analyzer (VNA; Anritsu MS46122A) was used to measure the S-parameters. To mimic the on-body situation, a pork meat piece of about 20-mm thickness of the same size of the device was placed below the wearable device. Fig. 12 shows the S-parameter results for both antennas without and with tissue. Fig. 12(a) shows   that the S-parameter of the reference antenna shifts when the device is placed on the tissue. The shift of resonant frequency can affect the performance of the antenna depending on the user environment, which is not desirable. In contrast, the proposed antenna shows a stable S-parameter regardless of whether the device is ON or OFF the tissue in Fig. 12(b). It can be inferred that the performance of the proposed antenna is hardly affected by the user environment. Based on the  Besides, the radiation patterns were measured in an anechoic chamber along with the tissue (pork meat) and shown in Fig. 13. Fig. 13(a) shows the principle planes (x y, x z, and yz) of radiation from the device when the device is placed with its normal direction along the z-axis. As conventionally defined in the spherical coordinate [2], the angle θ and φ refer to the polar and the azimuthal angle, respectively. It can be observed that the peak radiation is observed in broadside directions around +50 • for yz plane, around +90 • in x z plane. For the x y plane, it is almost uniform and nulls are observed at ±90 • .
The proposed antenna is compared with the reported antennas utilized for the smartwatch, body-area network, and telemedicine applications at 2.4 GHz, in terms of the size, feeding techniques, and the design configuration in Table III. Compared to the existing solutions, the proposed antenna exhibits the smallest size by the techniques of modified PIFA with the shorting pin.
Since the antenna design is compact in size and demonstrates little medium dependency, the proposed antenna is chosen for the watch-type wearable device.

III. CONNECTIVITY FOR VARIOUS CONDITIONS
Since the antenna targets a wearable application, we examine the reliability of connectivity depending on the various postures of the user and the orientations of the device in this section. It is inconvenient to measure the S-parameter of the antenna for various scenarios using the VNA. Therefore, the connectivity is evaluated instead by measuring the RSSI, with the antenna assembled within a prototype device in Fig. 11. The measurement of the RSSI also better reflects the actual operating condition of the device since it does not require a coaxial cable from the VNA which prevents the complete closure of the device cover. The RSSI can be measured with a Bluetooth-enabled tablet in which an in-house developed BLE communication code is executed. While the wearable device is set to transmit power of 6 dBm level, the tablet as an external reader records the strength of the received signal at a fixed position.
The measurement scenarios for the wearable device to work on the human body are shown in Fig. 14. The external reader is placed on a computer table with a height of 70 cm. In the first scenario, the device is worn on the wrist facing the external reader and the distance D x between transceivers is varied up to 10 m [see Fig. 14(a)]. The RSSI measured by the external reader is presented in Fig. 14(b). It can be observed that the device can operate well up to the range of 10 m, where the RSSI value drops to −90 dBm. Considering the typical BLE receiver sensitivity of about −95 dB [20] and the typical distance between indoor transceivers, the measurement results reveal that the device is suitable for indoor operation.
Further measurements are performed in various scenarios in which the device is worn on the wrist, inserted in the front pocket, or attached on the back side of upper waist as shown in Fig. 14(c), (e), and (g). The second scenario is shown in Fig. 14(c) where the user is standing at 1 m distance from the reader and turning in the clockwise direction as shown in the inset of the figure. In this scenario, the device is worn on a wrist like a watch or inserted in the front pocket. The orientation of the device can be informed by considering the illustration of the coordinate space shown in Figs. 13(a) and 14(c). For both situations, the RSSI is recorded and plotted in Fig. 14(d). The different orientation of the device inside the pocket makes the maximum RSSI appear at a different rotation angle from the case when the device is worn on the wrist. One can also observe that the RSSI level gets lower when the user is rotated by around 180 • obviously because the signal is severely blocked by the body of the user. Nevertheless, the RSSI does not drop below −90 dB, which indicates that BLE communication remains connected.
The third scenario is shown in Fig. 14(e). In this scenario, the user is in the attention position, where the device is again worn on a wrist or inserted in the front pocket. The user rotates in the same way as in Fig. 14(c) and the signal strength is measured as Fig. 14(f). For the wrist case, the rotation angle with the maximum strength shifts to 90 • because the orientation of the device has changed from Fig. 14(c). In other words, when the user is rotated by 90 • , the reader is in the broadside direction of the antenna. The front-pocket case has the similar trend as the front-pocket case of Fig. 14(c) because the orientation of the device has not changed.
The last scenario is when the device is attached on the back side at the upper waist as presented in Fig. 14(g). The device is placed with two different orientations as shown by the illustrations of the coordinate space in Fig. 14(g). When the user makes the rotation, the signal strengths are plotted in Fig. 14(h). It is observed that in the case of the waist measurements, the maximum RSSI occurs at θ = 180 • because at this angle, the device and reader are facing each other. In all the experiments of the above scenarios, the RSSI remains in the affordable range that the device is connected, which indicates the reliable operation of indoor communication as a wearable device.

IV. SYSTEM INTEGRATION FOR EMG MONITORING
The proposed antenna is integrated with a wearable electromyography monitoring device. The EMG is a common clinical test used to assess the function of muscles and the nerves that control them. The EMG can be detected either directly by inserting electrodes or indirectly with surface electrodes positioned on skin regions immediately above the muscle tissue [43]. Here, the EMG signal detection is performed using the surface detection method and the electric signal of the muscle is then transferred to the remote reader located 10 m far from the human body via the BLE. The device location on the forearm can be seen in Fig. 15(a) and the decoded EMG signal at the remote reader is shown in Fig. 15(b). The forearm is chosen as the test position because the muscle of the forearm is large enough to attach the electrode and it aligns with the objective of our research to develop the antenna for a watch-type wearable device. In general, however, the device can be placed at any position on the human body as experimented in Fig. 14. It can be seen that when the muscle is in an active period, the signal burst with a maximum amplitude of approximately 2 mV is obtained. In contrast, during the rest period, the amplitude level is negligible following the baseline.

V. CONCLUSION
In this work, the antenna design has been studied for the BLE commutation of a wearable device. The proposed antenna was compared with the meandered dipole antenna as a reference antenna which is widely used for BLE communication due to its small size. While the reference antenna has a smaller size, it suffers a change in frequency response when placed on the tissue. On the other hand, the proposed antenna shows little difference whether it is placed on a tissue or in the air, because of the inclusion of a complete ground plane beneath the main radiation patch. The measurement has been carried out for the validation of the concept. The proposed antenna has an impedance bandwidth of 80 MHz (2.4-2.48 GHz), with a peak gain of 3.62 dBi and an efficiency of 20%. The antenna embedded in the wearable device enables a reliable connection for various scenarios of orientations and postures of the user. The maximum range in the device transmit power level of 6 dBm was observed to be 10 m with the signal strength of >−90 dBm, demonstrating a suitable operation for indoor communication. Since the antenna is connected through the standard U.FL connector, the design can be adopted easily in other wearable applications with high reproducibility. Lastly, the wireless EMG monitoring device using the proposed antenna was demonstrated showing the wireless recording of the EMG signal at a distance of 10 m.