Antenna evolution

The evolution of the antenna geometry and the corresponding |S₁₁| characteristics are presented in Fig. 2. The antenna design is initiated with a conventional geometry having dipole-like radiating arms attached to a rectangular ground plane (Stage 1) as shown in Fig. 2(a). The dipole arms are excited using a microstrip line to slot-line transition. The microstrip line on the rear side is configured as a L-shaped probe to achieve good impedance bandwidth. The antenna exhibits a resonance around 4 GHz with a weak additional resonance centerd around 2.6 GHz. The electromagnetic coupling between the dipole radiator and the director is enhanced by reducing the separation “s”. Further, the strip-line director is modified into a semicircular loop (Stage 2) as described in Fig. 2(b) to introduce additional resonant modes through electromagnetic coupling effects. There was a marginal improvement in the reflection coefficient behavior which is further enhanced through the attempts to modify the ground plane of the quasi-yagi antenna. Figure 2(c) shows the ground-modified quasi-yagi antenna wherein the top corners of the ground plane are shaped through a blend radius R (Stage 3). The optimum value of R is 10 mm to achieve strong resonance at the upper frequency. The ground plane modifications did not contribute to performance enhancement in the lower frequency edge. Hence, further structural modifications are performed to improve the antenna performance at the lower resonant frequency. Broadband impedance matching can be obtained using stepped impedance feed. Hence, in the next developmental stages, the L-feed is terminated with a square patch (Stage 4) and the rectangular slot-line in the front portion of the antenna is modified into a tapered slot (Stage 5) as shown in Fig. 2(d) and (e), respectively. The modifications in the feedline and the tapered slot have contributed to two strong resonances centerd at 2.5 GHz and 3.6 GHz. The optimum size of the square patch termination on the L-feed is fixed at “p” mm. In the final optimization stage, the dipole arms are tapered (Stage 6) to achieve necessary impedance matching at the intended dual frequencies. The resultant antenna geometry is shown in Fig. 2(f). The simulated return loss of the antenna at the ISM and 5G wireless local area network frequencies are 19 dB and 26 dB, respectively as shown in Fig. 2(g). The estimated bandwidth of the dual-band quasi-yagi antenna is 110 MHz and 125 MHz which are 4.5% and 3.5% at 2.45 GHz and 3.5 GHz, respectively. The proposed antenna has a realized bandwidth greater than the bandwidth required for the intended standards. Further, the data exchanged in WBANs typically requires a smaller bandwidth, unlike the information and entertainment systems. Hence, the realized bandwidth is sufficient for the said application. The optimized dimensions of the antenna are given in Table 1. Owing to the geometry, the antenna has a unidirectional radiation pattern which is a critical aspect for on-body antennas. The antenna is positioned on the human body to produce tangential radiation to perform on-body sensor communication.

Figure 2. Evolution of the dual-band antenna (a) Stage 1: Conventional quasi-yagi antenna (b) Stage 2: Modified reflector configuration (c) Stage 3: Modified ground plane configuration (d) Stage 4: Modified feed line with square patch termination (e) Stage 5: Modified slot-line (f) Stage 6: Modified dipole arms (Proposed antenna geometry) (g) Reflection coefficient characteristics during the developmental stages.

Table 1. Optimized geometrical parameters

Approximate equivalent circuit

The presented antenna has a quasi-yagi configuration with a dipole radiator as a driven element and a semicircular loop-like director element. The approximate equivalent circuit of the antenna is presented in Fig. 3. The equivalent circuit has two RLCcircuits. One of the RLCs represents the radiation equivalent corresponding to the dipole element while the other RLC represents the radiation emanating from the semicircular loop. Both the RLCs are coupled through a coupling capacitor C _c. The radiation from the dipole radiator is modelled as a series RLC circuit where the low-frequency and high-frequency characteristics are dominated by capacitive and inductive reactance respectively. At the resonant frequency, the antenna has a resistive property. Hence at both the resonant frequencies, the net impedance is resistive yielding the dual-band property in the proposed antenna element.

Figure 3. Approximate equivalent circuit of the proposed dual-band antenna.

MIMO antenna development

The quasi-yagi antenna is used to develop the flexible MIMO antenna. The antenna elements are distributed along the sides of the hexagon as shown in Fig. 1(a) to realize the MIMO scheme. The antenna elements are edge-to-edge separated by 20 mm and each element is tilted to the corner to realize the proposed six-element MIMO antenna. The ground plane of all the antennas is connected to form a single ground configuration. Further, a hexagonal cavity is created in the middle to run the cables without disrupting the antenna performance. The diagonal length of the proposed six-element MIMO antenna is fixed as 104 mm to keep the mutual coupling better than −20 dB at both lower- and upper-frequency bands. Figure 4 illustrates the isolation characteristics between the different ports of the antenna in the MIMO scheme. The higher isolation between the antenna elements is attributed to the feedlines which are oppositely oriented for each antenna element. This has created only a small surface current penetrating from one antenna to another antenna owing to the repelling characteristics of the electromagnetic fields. The ground plane of the antenna is shaped to reduce spurious currents which has significantly contributed to undesired port-to-port coupling. The shape slot line transition is another contributing factor to the enhanced isolation in the proposed MIMO antenna.

Figure 4. Simulated isolation characteristics between different ports with reference to Port 1.

S-parameter characterization

The prototype six-element antenna is developed using the photolithography technique and the prototype is subjected to testing for validation of the simulation results. The photograph of the fabricated six-element MIMO antenna and the measurement setup are shown in Fig. 5(a) and (b) respectively. The antenna is tested using the E5071C ENA series vector network analyzer (VNA) from Keysight. The VNA is fully calibrated using a mechanical calibration kit using the short-open-load-\through method to eliminate the cable-induced losses. The real-time |S₁₁| characteristics of the six-element antenna are described in Fig. 5(c). The measurement results are in good agreement with the simulation results. The measured impedance bandwidth of the antenna is 4.25% and 4.5% at the lower and upper frequency bands.

Figure 5. (a) Prototype of the fabricated antenna (b) Measurement scenario (c) Measured reflection coefficient characteristics (d) Measured isolation (S₂₁) characteristics during free space and on-body conditions.

Furthermore, the antenna functionality is tested under on-body conditions. The simulation of the on-body performance is carried out using a three-layer human body model of dimensions 60 × 60 mm with skin (1 mm) on the top, fat (10 mm) in the middle and muscle (100 mm) at the bottom as shown in Fig. 6. The simulated S-parameters under on-body conditions are presented in Fig. 5(c) and (d). Due to the increased complexity involved in the calculation of S-parameters for eight ports and the lack of computational resources, the source considered for the calculation is limited to Port 1 and Port 2. The simulated reflection coefficient characteristics are maintained by introducing an air gap between the antenna and the human body model. The simulated results are validated through experimental measurements. The MIMO antenna is positioned in the arm of a female subject with an arm diameter of 120 mm and the measurements are carried out. To avoid body-induced detuning effects, the human body and the MIMO antenna are separated by 1 cm using a Styrofoam spacer. The sensors in the WBAN come in a module and the module has the embedded electronics over which the antennas are loaded. Hence, in reality, the antennas are not directly positioned on the human body. Considering this, the 1 cm spacing is used. Owing to the high flexibility of the microwave laminate used for the development of the MIMO antenna, the antenna characteristics remained stable under body-induced conditions. The measured S-parameters showed an improved bandwidth compared to the free-space conditions to perform the necessary WBAN communications. The MIMO antenna is further subjected to on-body studies by locating the antenna in the abdomen region of the chosen female subject. The abdomen is chosen since the cluster note can be centrally positioned to establish communication with the neighbour nodes. The measured impedance bandwidth of the MIMO antenna during on-body implementation is 12% and 17% at the lower and upper-frequency bands. The proposed MIMO antenna exhibited good isolation characteristics. The estimated isolation is greater than 20 dB within both the operating bands of the proposed MIMO antenna.

Figure 6. Simulation setup for the on-body analysis.

Radiation characterization

The radiation performance of the proposed six-element MIMO antenna is tested in an anechoic chamber with standard gain horn antennas. The simulated and measured radiation pattern is described in Fig. 7. The radiation pattern depicted in Fig. 7 is the total gain of the antenna calculated along the θ plane. During the measurement of one antenna element, the other antenna elements are terminated in a 50-Ω load. As evident from Fig. 7, the antenna exhibits a directional radiation pattern where the patterns are oriented along different directions for each port resulting in the realization of pattern diversity. The radiation pattern calculated through Port 1 and Port 4 offers antenna coverage along theta = 90 degrees and θ = 270° respectively. The radiation pattern observed from antennas connected to Port 2 and Port 5 cover θ = 0° and θ = 180°, respectively. Slant polarization is realized using antenna excitation through the remaining Port 3 and Port 6 respectively. The antenna hence covers all spatial coordinates with high directionality behavior. Hence it can be concluded that the proposed antenna is a promising candidate for networks that involve the deployment of sensors at different regions in the WBAN scheme. The average gain and total efficiency of the antenna are plotted in Fig. 8. The average gain calculated across all six ports of the MIMO antenna is above 3.5 dBi in both operating bands. The simulated average efficiency of the antenna is greater than 75% at 2.45 GHz and 3.5 GHz. The efficiency of the antenna could not be measured due to the lack of a measurement facility.

Figure 7. Normalized radiation pattern (a) Port 1, 4 at 2.45 GHz (b) Port 2, 5 at 2.45 GHz (c) Port 3, 6 at 2.45 GHz (d) Port 1, 4 at 3.5 GHz (e) Port 2, 5 at 3.5 GHz (f) Port 3, 6 at 3.5 GHz.

Figure 8. Gain and efficiency characteristics of the proposed antenna.

MIMO characterization

The MIMO performance of the proposed six-element antenna is evaluated using the standard equations presented in reference [Reference Sharawi27]. The far-field equation-based calculation for envelope correlation coefficient (ECC, ρ_e), apparent diversity gain (ADG, G_app), effective diversity gain (EDG, G_eff) and the ratio of mean effective gain (MEG) are calculated and presented. The equations used for the calculation of the above metrics are presented in Equations (1–4).

(1)\begin{equation}{\rho _e} = \frac{{|\iint {\left( {\mathop {\mathop F\nolimits_1 }\limits^ \to (\theta ,\phi ) \bullet \mathop {\mathop F\nolimits_2 }\limits^ \to (\theta ,\phi )} \right)d\Omega }{|^2}}}{{\iint {|\mathop {\mathop F\nolimits_1 }\limits^ \to (\theta ,\phi ){|^2}d\Omega \iint {|\mathop {\mathop F\nolimits_2 }\limits^ \to (\theta ,\phi ){|^2}d\Omega }}}}\end{equation}

(2)\begin{equation}{G_{app}} = 10 \times \sqrt {1 - \left| {{\rho _e}} \right|} \end{equation}

(3)\begin{equation}{G_{eff}} = {\eta _{total}} \times {G_{app}} = {\eta _{total}} \times 10 \times \sqrt {1 - \left| {{\rho _e}} \right|} \end{equation}

(4)\begin{equation}MEG = \int\limits_0^{2\pi } {\int\limits_0^\pi {\left[ \begin{gathered} \frac{{XPR}}{{1 + XPR}}{G_\theta }(\theta ,\phi ){P_\theta }(\theta ,\phi ) \hfill \\ + \frac{1}{{1 + XPR}}{G_\phi }(\theta ,\phi ){P_\phi }(\theta ,\phi ) \hfill \\ \end{gathered} \right]} } \sin \theta d\theta d\phi \end{equation}

In Equation (1), the F ₁ and F ₂ are the field components along the azimuth (ϕ) and elevation (θ) directions corresponding to the two elements between which the ECC is calculated. The ADG is computed from the ECC using Equation (2). The radiation losses are included in the ADG value as given in Equation (3) to estimate the true diversity performance of the proposed six-element MIMO antenna. The MEG is a critical parameter that is used to establish the antenna gain properties in the MIMO arrangement. The calculation of MEG involves the fixation of cross-polarization ratio (XPR) which is assumed as 0 dB for isotropic conditions while for indoor and outdoor situations, the XPR is fixed as 1 dB and 5 dB, respectively. The power distribution is assumed as a normal distribution with mean and standard deviation equal to 10°.

The computed MIMO parameters are tabulated in Table 2. It is inferred that the field equation-based ECC is far less than the acceptable threshold of 0.5. The ADG and EDG values are within the critical limits of 7.0 and the ratio of MEG is much smaller than the acceptable value of 3 dB.

Table 2. MIMO parameters

The MIMO antenna performance is compared with the existing research works and presented in Table 3. The following are the merits of the proposed MIMO antenna.

1. 1. The proposed MIMO antenna is a low-profile solution unlike the research presented in references [Reference Shirvani, Khajeh-Khalili and Neshati11–Reference Zhou, Leng, Pan, Abdalla, Novoselov and Hu13] and [Reference Ananda Rao and Bhavani Konkyana15–Reference Ali, Sovuthy, Noghanian, Ali, Abbasi, Imran, Saeidi and Socheatra20]. The percentage profile reduction is more than 90% compared to references [Reference Shirvani, Khajeh-Khalili and Neshati11, Reference Mashagba, Rahim, Adam, Jamaluddin, Yasin, Jusoh, Sabapathy, Abdulmalek, Al-Hadi, Ismail and Soh12, Reference Ananda Rao and Bhavani Konkyana15–Reference Noghanian17] and [Reference Ali, Sovuthy, Noghanian, Ali, Abbasi, Imran, Saeidi and Socheatra20].

2. 2. The proposed antenna has a compact geometry and is more than 50% smaller compared to references [Reference Ashfaq, Faisal, Ullah and Choi10, Reference Mashagba, Rahim, Adam, Jamaluddin, Yasin, Jusoh, Sabapathy, Abdulmalek, Al-Hadi, Ismail and Soh12, Reference Zhou, Leng, Pan, Abdalla, Novoselov and Hu13] and [Reference Ananda Rao and Bhavani Konkyana15–Reference Ali, Sovuthy, Noghanian, Ali, Abbasi, Imran, Saeidi and Socheatra20].

3. 3. Bandwidth performance of the antenna is directly proportional to the substrate thickness. Despite the low-profile nature of the proposed antenna, the on-body bandwidth performance of 12% and 17% at 2.45 GHz and 3.5 GHz.

4. 4. The gain performance of the proposed MIMO antenna is better than references [Reference Ashfaq, Faisal, Ullah and Choi10, Reference Mashagba, Rahim, Adam, Jamaluddin, Yasin, Jusoh, Sabapathy, Abdulmalek, Al-Hadi, Ismail and Soh12, Reference Zhou, Leng, Pan, Abdalla, Novoselov and Hu13, Reference Ananda Rao and Bhavani Konkyana15, Reference Kumkhet, Rakluea, Wongsin, Sangmahamad, Thaiwirot, Mahatthanajatuphat and Chudpooti18] and [Reference Ali, Sovuthy, Noghanian, Ali, Abbasi, Imran, Saeidi and Socheatra20] without using additional electromagnetic bandgap structures and artificial magnetic conductors.

5. 5. The proposed MIMO antenna encompasses more elements unlike references [Reference Mashagba, Rahim, Adam, Jamaluddin, Yasin, Jusoh, Sabapathy, Abdulmalek, Al-Hadi, Ismail and Soh12–Reference Ali, Sovuthy, Noghanian, Ali, Abbasi, Imran, Saeidi and Socheatra20] within a small footprint resulting in improved pattern diversity characteristics.

Table 3. Comparison with WBAN antennas in literature