Antenna design

The geometry of the proposed MIMO antenna is shown in Fig. 1. Four orthogonally placed identical radiating elements and ground planes are printed on the upper surface of the 1.6 mm-thick flame retardant-4 epoxy substrate (tanδ = 0.02) with a relative permittivity of 4.4. A single radiating element excited by the ACS is composed of a quarter-elliptical patch in which two L-shaped slots are etched on its surface to achieve dual-band notch characteristics. The finally optimized values of the proposed antenna are shown in Table 1.

Figure 1.

The top view of the geometry of the proposed four-port ACS-fed MIMO antenna.

Table 1.

Optimized dimensions of the proposed four-port ACS-fed MIMO antenna

To clarify the process of the proposed antenna design, three forms of the proposed MIMO antenna are shown in Fig. 2. The results of the comparison of S parameters of three forms of MIMO antenna are shown in Fig. 3.

Figure 2.

Three configurations derived from the evolutionary process of the proposed MIMO antenna: (a) Ant.1, (b) Ant.2, (c) Ant.3.

Figure 3.

Comparison of S parameters of each proposed antennas: (a) S11, (b) S12, and (c) S13.

On the basis of Ant.1, a pair of L-shaped slits are etched side by side of each radiating element. This modification changes the effective current path and impedance bandwidth. Simultaneously, the lowest resonant frequency of the antenna shifts to a lower frequency, and the frequency range of 5.1–5.8 GHz and 7.0–8.1 GHz are suppressed. Furthermore, the coupling between adjacent elements is significantly reduced at 5.2–6.1 GHz, while the coupling between opposite elements is significantly reduced at 6.8–8.4 GHz. The total length of the L-shaped slot is designed to fulfill the following equations:

(1)\begin{equation}{L_1} = {L_{s1}} + {W},\end{equation}

(2)\begin{equation}{L_2} = {L_{s2}} + {W},\end{equation}

(3)\begin{equation}{L_{slot}} = {c \over {4{f_{notch}}\sqrt {({\varepsilon _r} + 1)/2} }}.\end{equation}

where L ₁ is the total length of the upper L-shaped slot, L ₂ is the total length of the lower L-shaped slot, L_slot is the total length of the L-shaped slot, c is the speed of light, f_notch is the desired frequency of the notch band, and ε_r is the relative permittivity of the substrate.

To further optimize the isolation performance between the antenna elements, Ant.3 introduces a meander line isolator to reduce the coupling between adjacent elements at 3–4.6 GHz and 8.2–12 GHz. The end branches of the meander line are coupled with the radiation elements, ensuring isolation greater than 18 dB over the entire operating band and extending the impedance bandwidth effectively. More than this, due to the existence of the isolator, isolation between opposite elements is greater than 20 dB over the entire bandwidth.

Figure 4 illustrates the effect of the variations in parameters on antenna performance. As “L_s1” increases, the frequency range of the first notch band changes from 5.1–5.8 GHz to 4.6–5.6 GHz. Conversely, as “L_s2” decreases, the frequency range of the second notch band changes from 7.0–8.1 GHz to 7.4–8.3 GHz. It indicates that the length of the L-shaped slot can be flexibly adjusted according to the frequency of the notch band. In addition, the decrease in “M₂” means that the distance between the isolator and the radiation elements increases. In this case, it is difficult for the isolator to absorb the current on the radiator, resulting in enhanced coupling between antenna elements.

Figure 4.

(a) Effect of varying “L_s1” parameter of L-shaped slot on S11. (b) Effect of varying “L_s2” parameter of L-shaped slot on S11. (c) Effect of varying “M₂” parameter of the meander line isolator on S12. (d) Effect of varying “M₂” parameter of the meander line isolator on S13.

The effect of introducing an isolator can be observed by comparing Fig. 5(a) and (c), (b) and (d). It is worth noting that the antennas in Fig. 5(a–d) only activated port 1. The frequencies chosen are those where the differences in S parameters are relatively clear. When port 1 is activated, ports 2, 3, and 4 are terminated with matched load. The current is mainly distributed to the radiating element 1 and the meander line of its adjacent isolator, while there is almost no current distribution on the surface of ports 2, 3, and 4. It can be seen that the isolator absorbs part of the current and plays a decoupling role.

Figure 5.

Surface current distribution of (a) Ant.2 at 3.0 GHz, (b) Ant.2 at 10.1 GHz, (c) Ant.3 at 3.0 GHz, and (d) Ant.3 at 10.1 GHz.

Figure 6(a) and (b) shows the effect of the etched double L-shaped slots side by side on the antenna current distribution, where 5.6 GHz and 7.4 GHz are the center notch frequencies of the proposed antenna. The current is mainly concentrated around the upper L-shaped slot at 5.6 GHz. Similarly, at 7.4 GHz, the surface current is mainly concentrated around the lower L-shaped slot, indicating the significant effect of the notch structure.

Figure 6.

Surface current distribution of (a) Ant.3 at 5.6 GHz and (b) Ant.3 at 7.4 GHz.

Experimental results

To validate the accuracy of the simulated results, an experimental prototype of the MIMO antenna is fabricated. An Agilent E5080A network analyzer and a Satimo Starlab near-field measurement system are used to test the performance characteristics of the proposed antenna. Figure 7 provides the photograph of the antenna prototype and pattern measurement environment. Figure 8 illustrates the result of the comparison of measured and simulated S parameters. There are some differences between the simulation results and the measurements results, which are caused by the unavoidable wear and tear during the production and welding processes of the antenna and the fluctuation of the relative dielectric constant of FR4 dielectric material in the high-frequency range.

Figure 7.

(a) Fabricated prototype of the four-port MIMO antenna. (b) Pattern measurements in anechoic chamber.

Figure 8.

Comparison of measured and simulated results: (a) S11, (b) S12, and (c) S13.

In this section, the radiation patterns of the proposed MIMO antenna are investigated. As Fig. 9 illustrates, the simulated and measured two-dimensional radiation patterns of the antenna in free space along the E-plane (xz plane) and H-plane (yz plane) at sample frequencies of 4.2, 6.8, and 9.2 GHz, respectively, were given, which represent low, intermediate, and high frequencies within UWB bandwidth. The radiation patterns are measured when port 1 is activated and the other ports are terminated by the 50 Ω load. It can be observed that the radiation patterns act as an omnidirectional in the E-plane and the back lobe (180°–360°) of H-plane at each sample frequency, while there is relatively weak radiation in the front lobe (0°–180°) of H-plane. Cross-polarization is governed by the asymmetric nature of ground plane and the feeding structure, which induces the orthogonal modes that are perpendicular to the surface of the radiator. The co-polarization to cross-polarization ratio is maintained as −16 dB at 4.2 GHz, −17 dB at 6.8 GHz, and −21 dB at 9.2 GHz. Essentially, it can be assumed that the antenna can transmit and receive signals well in various wireless networks.

Figure 9.

Radiation patterns in the E-plane and H-plane at the frequencies of (a) 4.2 GHz, (b) 6.8 GHz, and (c) 9.2 GHz.

Figure 10 provides the peak gain and radiation efficiency results of the proposed MIMO antenna. The efficiency curves indicate that the proposed antenna has the ability to radiate most of its energy within the operating frequency band, but at notch frequencies, due to impedance mismatch, the energy of the signal source forms a standing wave on the transmission line, resulting in incomplete absorption of energy, so the efficiency has a sharp decline at notch frequencies. Similarly, the peak gain varies from 2.6 to 5.9 dBi over the entire operating frequency range, and it drops to −3.8 dBi in the rejected band. In general, the proposed antenna possesses an acceptable gain and radiation efficiency.

Figure 10.

Peak gain and radiation efficiency of the proposed ACS MIMO antenna.

Another important parameter used to evaluate the performance of UWB antennas is group delay, which is the time delay generated by the input signal when traveling through the antenna. The time-domain test setup is shown in Fig. 11. During the measurement process, two identical prototypes are placed face to face at a distance of 25 cm. As can be seen in Fig. 12, the group delay of the proposed antenna remains nearly constant over the entire UWB frequency spectrum, except for notch bands. Ideally, the group delay of the antenna must remain constant while the phase is strictly linear. Thus, the proposed antenna possesses good pulse handling capability and linear phase.

Figure 11.

Time-domain analysis test setup.

Figure 12.

Group delay of the proposed antenna when positioned face to face.

MIMO antenna performance

To verify the capability of the proposed antenna for MIMO applications, performance evaluation parameters such as envelope correlation coefficient (ECC), diversity gain (DG), and total active reflection coefficient (TARC) are discussed.

Envelope correlation coefficient

ECC is a metric that describes the degree to which communication channels are isolated or correlated with each other. The lower the ECC value, the lesser the correlation between antennas. The result of ECC is illustrated in Fig. 13, and the values can be derived from the following equation (4):

(4)\begin{equation}EC{C_{ij}}{\textrm{ = }}{{{{\left| {S_{ii}^*{S_{ij}} + S_{ji}^*{S_{jj}}} \right|}^2}} \over {\left| {(1 - {{\left| {{S_{ii}}} \right|}^2} - {{\left| {{S_{ji}}} \right|}^2}) \cdot (1 - {{\left| {{S_{jj}}} \right|}^2} - {{\left| {{S_{ji}}} \right|}^2})} \right|}}.\end{equation}

Figure 13.

ECC versus frequency for the proposed antenna.

Diversity gain

The DG is defined as the improvement in signal-to-noise (SNR) ratio from the combined signal from all the antennas of the diversity system relative to the SNR from a single antenna. The result of DG is illustrated in Fig. 14, and the values are calculated using the following equation (5):

(5)\begin{equation}D{G_{ij}} = 10\sqrt {1 - {{\left| {EC{C_{ij}}} \right|}^2}} .\end{equation}

Figure 14.

DG versus frequency for the proposed antenna.

Total active reflection coefficient

In a MIMO system, the mutual impedance between adjacent antennas and self-impedance of individual antenna elements are affected by antenna radiation. Therefore, analyzing only the individual S-parameter values of the antenna elements is insufficient. However, a new parameter called TARC can indicate changes in mutual impedance and self-impedance of array antennas, making it useful for investigating MIMO system properties. The TARC result is shown in Fig. 15, and the values can be computed using the following equation (6):

(6)\begin{equation}TARC = {{\sqrt {\sum\limits_{i = 1}^N {{{\left( {\sum\limits_{j = 1}^N {{S_{ij}}} } \right)}^2}} } } \over {\sqrt N }},\end{equation}

Figure 15.

The results of the measured TARC.

where N is the number of antenna elements and S_ij are the reflection coefficients between various ports of the MIMO antenna.

Comparison with other already reported MIMO antennas

A comparison between the proposed antenna and other similar antennas is shown in Table 2, which includes size, impedance bandwidth, isolation, ECC, and the number of notch bands. Compared to those previously reported, the proposed antenna is relatively small and achieves high isolation. It is worth noting that λ in Table 2 is the wavelength of the lowest operating frequency.

Table 2.

Performance comparison of the proposed antenna with existing literature