UWB antenna element design

In general, if only the change in the effective length of the antenna is considered, the antenna is half-cut at the axis of symmetry, which does not affect its performance. The principle is attributed to the magnetic wall effect at the symmetry plane of the antenna with a symmetric structure; thus, the halved antenna still has the frequency characteristics required by the resonance frequency. Figure 2(a) shows the design process of a single UWB antenna. A miniaturized antenna (#2) can be obtained by halving the original antenna (#1) along the axis of symmetry (SS′). As shown in Fig. 2(b), the reflection coefficient of the halved antenna is significantly worse than that of the original antenna because the input impedance of the halved antenna changes and its resonance frequency is basically unchanged. To improve the impedance matching of the halved antenna, the position of the antenna (#3) on the dielectric substrate is appropriately adjusted. Figure 2(b) shows that #3 has reflection characteristics similar to those of #1, and #3 has a wider operating band.

Band-notch structure design

A 90° clockwise rotation is utilized to create a four-port MIMO array after the basic UWB antenna unit (#3) is designed. Adjacent antenna units are then distributed orthogonally to each other, and the polarization characteristics correspond to orthogonal polarization. In this way, the antenna units can only receive weak signals excited by each other; thus, the isolation of the MIMO antenna is improved. The proposed MIMO antenna structure also exhibits an L-shaped and continuous bending-type slot structure on the radiator, and the direction of the current is opposite in these two slots. The radiated fields caused by the adverse current can cancel each other out, so the antenna cannot radiate efficiently in the notched frequency bands [Reference Chandel, Gautam and Rambabu13].

The length of the two slot structures is related to the center frequency of the notch band, and it can be calculated approximately according to equation (1) [Reference Khan, Iftikhar and Shubair20]:

(1)$$L = \displaystyle{\lambda \over 4} = \displaystyle{c \over {4f_0\sqrt {\varepsilon _{eff}} }}, \;\varepsilon _{eff} = \displaystyle{{\varepsilon _r + 1} \over 2}$$

In equation (1), L is the slot length, c is the speed of light, and ɛ _eff and ɛ _r are the effective permittivity and relative permittivity, respectively.

According to equation (1), the length of the slot is optimized to be approximately equal to the length of λ/4 (where λ is the free-space wavelength in the notch center frequency). When the antenna operates in the notch frequency band, these slots can be regarded as the λ/4 transmission line with a terminal short circuit at the corresponding frequency, i.e. the parallel resonant circuit. The equivalent circuit of the antenna is shown in Fig. 3(a). Whenever any slot resonates, the input impedance of the antenna at the corresponding notch band approaches infinity. At this time, the resonant circuit is equivalent to an open circuit, and the antenna cannot transmit energy normally; thus, a notch band is generated. In Table 1, the calculation and simulation slot length values when the notch center frequencies are 3.5 and 5.5 GHz, respectively, are compared. Table 1 shows that the difference between the two values is very small.

Fig. 3.

Comparison between the simulated S ₁₁ of antenna by CST and by the equivalent circuit model: (a) equivalent circuit model and (b) S ₁₁.

Table 1.

Comparison of theoretical value and simulation value of slot length

Figure 3(b) shows a comparison between S ₁₁ of the proposed antenna obtained by CST simulation and the equivalent circuit model. As shown in the figures, the trends of the curves agree reasonably well over the UWB band, especially in the dual-notched frequency bands. The discrepancy between the curves can be mostly attributed to the antenna that is replaced by a 50 Ω resistance in the equivalent circuit. In most cases, if the antenna is well matched, this approximation can be valid, but for a large bandwidth this hypothesis is not fulfilled.

All the parameters of the proposed antenna were optimized using commercial full-wave software CST Microwave Studio. The final detailed dimensions of the UWB MIMO antenna, in millimeters, are as follows: W = 42, W ₁ = 12, W ₂ = 1.5, W ₃ = 1.2, W ₄ = 0.4, W ₅ = 0.2, W ₆ = 2, W ₇ = 6, L = 42, L ₁ = 14.65, L ₂ = 8, L ₃ = 4.3, L ₄ = 2, L ₅ = 8.8, L ₆ = 2, L ₇ = 30, R = 7.7, h = 0.508.

S ₁₁-parameters

The simulated and measured S ₁₁ of the antenna prototype is given in Fig. 5(a), showing similar trends. The measured −10 dB impedance bandwidth is 2.9–16.5 GHz (relative bandwidth is about 140.2%), which is close to the simulated result (2.7–16.7 GHz, relative bandwidth is about 144.3%). The notch frequency bands of the MIMO antenna are 3.1–3.8 and 5.2–5.9 GHz. In addition, Fig. 5(a) also depicts the S ₁₁ of the slotless MIMO antenna, and the results are acceptable.

Fig. 5.

S ₁₁ and simulated surface current distribution of MIMO antenna: (a) S ₁₁ and (b) surface current distribution.

Figure 5(b) shows the simulated surface current distribution of the antenna at 3.5 and 5.5 GHz. Near the center frequency of the notch, a large number of currents in the opposite directions are confined along the slot structure, the electromagnetic energy cannot be completely radiated outward, and the notch function occurs [Reference Chandel, Gautam and Rambabu13, Reference Gautam, Yadav and Rambabu16]. Nevertheless, the current is evenly distributed in the non-notched frequency band, and the antenna energy is normally radiated outward.

To analyze the effect of slits on the band-notched characteristics, various parameters were assessed. First, the length of the notch antenna, slot 1, is varied to show the effect of the overall length of slot 1 on the WiMAX band-notched frequency in Fig. 6(a). As the figure shows, when the length of slot 1 is shifted from 16.8 to 18 mm, the notch band shifts toward a lower frequency. Therefore, to obtain the desired band-notched frequency, 17.4 mm is chosen as the final value. Similarly, Fig. 6(b) shows the S-parameters for the variation of length slot 2, while other parameters remain constant. As the length of slot 2 increases from 10.5 to 11.5 mm, the center of the notched frequency band shifts from 5.9 to 5.3 GHz. Therefore, 11 mm is chosen as the final value of the slot 2 length. Both values are approximately the same, which confirms the effectiveness of equation (1).

Fig. 6.

Effects of the length variation of the slits on S ₁₁: (a) slot 1 and (b) slot 2.

Isolation

Figure 7 shows the measured and simulated isolation of the MIMO antenna when port 1 is excited and the other three ports are terminated with 50 Ω matched loads. In the UWB frequency band, the couplings between ports 1 and 2 and that between ports 3 and 4 are less than −23 dB, but the coupling in the notch band is only less than −20 dB. This is because there is substantial energy accumulation in the notch frequency band and this energy cannot be radiated outward, which leads to the increase of the coupling between antenna elements. Because of the symmetrical geometry of the structure, the same characteristics appear when the other three ports are activated.

Fig. 7.

Isolation of the UWB MIMO antenna.

To understand the decoupling effect of the polarization diversity technology intuitively, Fig. 8 shows the simulated surface current distribution of the MIMO antenna at 10 GHz. When port 1 is excited, the surface current flows along the y-axis direction and forms a y-polarized field. Therefore, the surface current coupled to the other three antenna elements is very weak. Similarly, when port 2 is excited, the surface current flows in the x-direction, forming an x-polarized field. A large amount of current is concentrated in this antenna element, while the other three antenna elements have weak surface currents. The polarization diversity technology can effectively improve the isolation of MIMO antennas.

Fig. 8.

Simulated surface current distribution on the antenna at 10 GHz: (a) port 1 is excited and other ports are matched and (b) port 2 is excited and other ports are matched.

Radiation characteristics

In a MIMO system, multiple antenna elements with different radiation patterns can be employed to reduce the multipath effects. Figure 9(a) shows the three-dimensional (3D) radiation pattern of the MIMO antenna when ports 1–4 are excited at 5 GHz. As shown in the figure, the radiation patterns of Ant_1 and Ant_3 are 180° mirror images, and the radiation patterns of Ant_1 and Ant_2 are perpendicular to each other. Furthermore, Ant_2 has a good omnidirectional radiation mode in the yoz-plane, and a radiation pattern similar to the “8”-shape in the xoz-plane. In addition, the two-dimensional (2D) radiation patterns of Ant_1 and Ant_2 in the xoy-plane at 5 GHz are plotted in Fig. 9(b). The results show that the radiation patterns of ports 1 and 2 in the xoy-plane are also perpendicular to each other. Furthermore, Ant_1 has stronger radiation at 90° and 270°, whereas Ant_2 has null radiation in those directions. The above results verify that the proposed MIMO antenna has good diversity performance. Figure 9(c) shows that the pattern is nearly omnidirectional in the xoz-plane, which is suitable for UWB MIMO systems.

Fig. 9.

3D and 2D patterns of the antenna when exciting different ports: (a) 3D pattern, (b) xoy-plane, and (c) xoz-plane (only port 1 is excited, while the other ports are terminated with a 50 Ω load).

To illustrate the radiation stability of the proposed antenna, Fig. 10 shows the simulated radiation patterns of the antenna at 8, 11, and 14 GHz when port 1 is excited, while the other port is terminated with a 50 Ω load, and vice versa. At lower frequencies, the patterns were fairly “8”-shaped in the xoy-plane and omnidirectional in the xoz-plane, as shown in Figs 9(b) and 10(a) at 5 and 8 GHz. At higher frequencies, radiation patterns deteriorate at higher frequencies owing to the splitting of the radiation lobes. The deterioration is caused by an imbalance in the high-frequency current distribution [Reference Wang, Du and Yang7].

Fig. 10.

Simulated radiation patterns of the MIMO antenna: (a) 8 GHz, (b) 11 GHz, and (c) 14 GHz.

Figure 11 shows that, in the operating band, the maximum gain of the antenna is approximately 5 dBi. In the notch band, the gain plummets to −2 dBi, which effectively solves the problem of electromagnetic compatibility among the UWB, WiMAX, and WLAN. The radiation efficiency of the antenna also has the same variation trend.

Fig. 11.

Gain and efficiency of the UWB MIMO antenna.

MIMO performance parameters

In a MIMO antenna system, the envelope correlation coefficient (ECC) is used to evaluate the diversity performance. In an ideal uniform scattering environment, S-parameters – equation (2) – are often used to calculate the ECC [Reference Tripathi, Mohan and Yadav6, Reference Dastranj9]. However, the calculated ECC is closer to the actual situation with the far-field parameters – equation (3) [Reference Wang, Du and Yang7]:

(2)$$ECC = \rho _{eij} = \left\vert {\displaystyle{{\vert {S_{ii}^\ast S_{ij} + S_{\,ji}^\ast S_{\,jj}} \vert } \over {{\vert {( {1-{\vert {S_{ii}} \vert }^2-{\vert {S_{ij}} \vert }^2} ) \cdot ( {1-{\vert {S_{\,ji}} \vert }^2-{\vert {S_{\,jj}} \vert }^2} ) } \vert }^{1/2}}}} \right\vert ^2$$

(3)$$\rho _{eij} = \displaystyle{{{\left\vert {\int_0^{2\pi } {\int_0^\pi {( {XPR\cdot E_{\theta i}\cdot E_{\theta j}^\ast{\cdot} P_\theta + E_{\phi i}\cdot E_{\phi j}^\ast{\cdot} P_\phi } ) d\Omega } } } \right\vert }^2} \over {\int_0^{2\pi } {\int_0^\pi {( {XPR\cdot E_{\theta i}\cdot E_{\theta i}^\ast{\cdot} P_\theta + E_{\phi i}\cdot E_{\phi i}^\ast{\cdot} P_\phi } ) d\Omega \times \int_0^{2\pi } {\int_0^\pi {( {XPR\cdot E_{\theta j}\cdot E_{\theta j}^\ast{\cdot} P_\theta + E_{\phi j}\cdot E_{\phi j}^\ast{\cdot} P_\phi } ) d\Omega } } } } }} $$

where i and j are the numbers of ports, XPR is the cross-polarization ratio, and $P_\theta$ and $P_\phi$ are the θ and ϕ components of the angular density functions of the incoming wave, respectively.

The diversity gain (DG) is also an important parameter for evaluating the performance of a MIMO antenna. The DG can be calculated using equation (4). Figure 12 shows, except for the dual-notch frequency band, the ECC is lower than 0.004, and the DG is higher than the 9.9 dB for the MIMO antenna, proving that the antenna has a good diversity characteristic:

Fig. 12.

Simulated ECC and DG of the proposed design.

(4)$$\textstyle{DG = 10\sqrt { 1-ECC^2}} $$

Table 2 lists the performance parameters of MIMO antennas that have been reported in the literature, and compares them with those of the proposed antenna. Compared with those in the literature [Reference Rajkumar, Amala and Selvan10, Reference Mchbal, Touhami and Elftouh12], although the antenna in this study has a larger size, it has the advantages of high isolation and a simple structure (without using any decoupling structure). In addition, compared with previously reported ones [Reference Tripathi, Mohan and Yadav6, Reference Khan, Iftikhar and Shubair20], the design here has obvious advantages in terms of size and isolation, but the previous antennas [Reference Tripathi, Mohan and Yadav6, Reference Khan, Iftikhar and Shubair20] have better low-frequency performance. Although a previously designed antenna [Reference Dastranj9] is smaller, it is a two-element MIMO antenna. The maximum gains in previous reports have been shown in Table 2. Compared with those reported earlier [Reference Tripathi, Mohan and Yadav6, Reference Pannu and Sharma11, Reference Mchbal, Touhami and Elftouh12], the proposed antenna has a higher gain. Moreover, Table 2 shows the related parameters of the MIMO antennas (#1_MIMO and #3_MIMO) designed with a full antenna (#1) and a halved antenna (#3) as a single antenna. In the comparison, #3_MIMO has great advantages in terms of size and isolation.

Table 2.

Performance comparisons with previously reported antennas