Proposed antenna design and evolution

The geometry of the proposed compact SWB annular ring antenna using 45° clock-wise square patch inclusions for SHF applications is shown in Fig. 1.

Fig. 1. Geometry of the proposed SWB antenna.

The inclusions consist of square patches (S ₁₋₅) with inscribed etched circles (R ₁₋₄) and form the radiating elements of the antenna. The design is symmetrical along the vertical axis and it will be shown attractive to allow the current distribution to remain symmetric throughout the design evolution. It uses a partial ground plane with a 4-stair-type defect at the back of the substrate (Fig. 1) for good impedance matching over the BW and etched on a commercially available FR-4 substrate having dielectric constant ɛ_r = 4.4, substrate height h = 1.6 mm, and loss tangent tan δ = 0.02. A 50 Ω microstrip line fed the antenna whose dimensions were calculated using standard equations [Reference Balanis18] and iteratively optimized for optimum performance. CST Microwave Studio was used for the design and led to a compact prototype W, ~ L (Fig. 1) of 25 × 26 mm² (0.17λ × 0.18λ).

Initially, a conventional annular ring antenna, Fig. 2(a), was designed using standard equations [Reference Kumar and Ray19] and conveniently simulated. Observations on the surface maximum current distribution, shown in Fig. 2(b), suggested the removal of the area in which the currents are low. Most of the currents were concentrated toward the feed line with minimal concentrations opposite to the feed. This area of minimal concentrations was removed and resulted in a relatively smaller design; this is presented in Fig. 3(a) and termed as evolved 0 design.

Fig. 2. (a) Conventional annular ring antenna and (b) surface current distribution.

Fig. 3. Evolution of the proposed antenna: (a) evolved 0 design, (b) evolved 1 design, and (c) evolved 2 design.

The simulated reflection coefficient (S ₁₁) of the conventional annular ring antenna and evolved 0 to 2 are shown in Fig. 4. Results show evolved 0 design with f ₀ of 2.1–4.5, 10.95–12, 13.83–16, 17–20.45, 22.2–27.25, and 28.4–30 GHz measured at −10 dB and leading to a multi-band antenna.

Fig. 4. Simulated S ₁₁ parameters of evolved antennas.

Next, to achieve a SWB response by maintaining same physical size, a square patch of side S ₁ having a circular slot of radius R ₁ was included in the inner area of the annular ring of evolved 0 design; the resulting design is shown in Fig. 3(b) referred to it as evolved 1. Adding these structures help in increasing the electrical length of the antenna by providing multiple paths for the current to pass. As a result, the evolved 1 design resonated in the 2.05–6.28, 9.1–14.80, 15.65–20.90, 22.90–28.42, and 29.32–30 GHz frequency bands (measured at −10 dB).

Additional inclusions were performed to further increase the BW performance to form the final evolved design. This was made by two more clock-wise 45° rotated square patches of sides S ₂ and S ₃ having circular slots of radii R ₂ and R ₃ respectively, Fig. 3(c). This led to the evolved 2 design with satisfactory performance over the intended frequency range. However, impedance mismatch was observed (Fig. 4) at 8.5, 11.8, 14.5, and 20.7–22.7 GHz measured at −10 dB.

Finally, referred as proposed design (Fig. 1), two additional square patches of side S ₄ and S ₅ were included in the previous evolved 2 design out of which the fourth square (S ₄) is having a circular slot of radius R ₄. This resulted in an impedance matching away from the initially intended 50 Ω specially at the high frequency (f _h) of operation and to accommodate this, a stair-type defect was made in the ground plane. This led to a proposed compact SWB annular ring antenna design using 45° clock-wise square patch inclusions for SHF applications with a f ₀ between 2.07 and 30 GHz (measured at −10 dB). From the inset of Fig. 4, it can also be seen that, the lower cut-off frequency shifts mainly for the first inclusion and not others. Since the lowest cut-off frequency is caused by the outermost structures, it does not change much for inner inscribed shapes. On the contrary, the inner shapes improve the response at higher frequencies by generating multiple constructive resonances. This is further validated by Fig. 9.

The final dimensions of the proposed antenna are listed in Table 1.

Table 1. Dimensions of the proposed antenna

Fabrication and validation

The proposed antenna was fabricated and measured. A Power Network Analyzer (PNA-L) N5234B by Agilent Technologies was used for the experiments. Anechoic chamber setup for antenna measurement is shown in Fig. 5.

Fig. 5. Measurement setup inside an anechoic chamber.

The fabricated prototype and S ₁₁ parameters of the fabricated antenna (compared to simulated) are shown in Fig. 6.

Fig. 6. S ₁₁ parameters of the proposed antenna with the fabricated prototype.

Experimental results show a compact SWB annular ring antenna using 45° clock-wise square patch inclusions for SHF applications having a BW of 28 GHz over the f ₀ of 2–30 GHz. Measured and simulated results are in fair agreement as shown in Fig. 6. The simulated (co- and cross-polar) and measured (co-polar) radiation patterns of the antenna are shown in Fig. 7. Only relevant patterns measured at 5, 13, 21, and 28 GHz are shown. The patterns are all pretty much omnidirectional for the H-plane (ϕ = 90°) and dipole-like for the E-plane (ϕ = 0°); this is more apparent at 5 GHz, Fig. 7(a). At higher frequencies 13, 21, and 28 GHz, Figs 7(b)–7(d), the E-plane is not that dipole-like also H-plane is not omnidirectional due to the excitation of higher order modes [Reference Rafique and Din9], however, these modes did not present a negative influence in real applications. Also, the level of cross-polarized patterns is approaching toward co-polar patterns for some angles. This also occurs due to excitation of higher order modes at higher frequencies. Measured and simulated results are in fair agreement, validating the fabricated antenna.

Fig. 7. Simulated and measured radiation patterns of the proposed antenna at (a) 5 GHz, (b) 13 GHz, (c) 21 GHz, and (d) 28 GHz.

The gain was measured and corroborated with simulations as shown in Fig. 8(a). Peak gains of 8 dBi are recorded for the simulated and measured results with strong agreement between both the results. The gain pattern is in accordance with [Reference Hakimi, Rahim, Abedian, Noghabaei and Khalily14, Reference Rahman, Cao, Ullah and Khalil20–Reference Yeo and Lee23]. Figure 8(b) shows the input impedance of the antenna. Although it is normal to vary over the f ₀, fair agreement is seen with simulated results validating the results. The real and imaginary values are seen fluctuating around 50  and 0 Ω respectively what is perceived as an appropriate input impedance for the proposed SWB antenna.

Fig. 8. Simulated and measured (a) gain and efficiency and (b) input impedance of the proposed antenna.

Furthermore, to gain an insight into antenna performance for entire frequency range, current density at three different frequencies is shown in Fig. 9. It is evident from the figure that at 5 GHz, the current density was denser along the annular ring near the feed line while at higher frequencies, 15 and 28 GHz, uniformly dense over the radiator (annular ring + patch inclusions). That means, as the frequency increased more current densities were observed in the inner inclusions. Because of these inclusions, the current-carrying conductor (the radiator) was lengthier and the current density around the circumference got redistributed.

Fig. 9. Simulated maximum current densities of the proposed antenna: (a) 5 GHz, (b) 15 GHz, and (c) 28  GHz.

To corroborate the performance of the proposed SWB antenna is to characterize the percentage BW provided per unit electrical area, the BDR. For SWB antennas, a high BDR is preferred to ensure maximum BW per unit electrical area of the design [1]. Mathematically, the BDR is expressed as,

(1)$$BDR = {\% BW\over \lambda_l \times \lambda_w}$$

where $\% BW$ is the percentage BW of the antenna, λ_l and λ_w are antenna length and width corresponding to f _l.

Using this BDR measurement and other comparative measures, a comparison study was carried out with other reported literature prototypes and summarized in Table 2. Results show that the proposed design is miniaturized by 7.67$\%$ compared to the smallest prototype and providecomparable gain than reported in the literature. The proposed antenna provides a wider BW compared to [Reference Liang, Zhong and Wang1–Reference Liu, Esselle, Hay and Zhong7, Reference Rafique and Din9, Reference Chen and Row10, Reference Samsuzzaman and Islam12, Reference Tang, Ziolkowski and Xiao13, Reference Srifi, El Mrabet, Falcone, Ayza and Essaaidi21, Reference Yeo and Lee23–Reference Rahman, Islam, Mahmud and Samsuzzaman27] and better BW ratio and $\%$BW compared to [Reference Chen and Row10, Reference Gorai, Karmakar, Pal and Ghatak11, Reference Tang, Ziolkowski and Xiao13, Reference Hakimi, Rahim, Abedian, Noghabaei and Khalily14, Reference Srifi, El Mrabet, Falcone, Ayza and Essaaidi21, Reference Okas, Sharma and Gangwar24, Reference Okas, Sharma and Gangwar25, Reference Rahman, Islam, Mahmud and Samsuzzaman27, Reference Okan28] with a higher BDR compared to all designs except for [Reference Okas, Sharma and Gangwar24, Reference Okas, Sharma and Gangwar25] but these present narrower BW, therefore not overcoming the proposed design.

Table 2. Comparison study with the literature

Here, λ_l and λ_w are normalized length and width (in wavelengths) corresponding to the lowest operating frequency.

Time-domain analysis

To measure the performance of the proposed antenna in time domain, a time-domain analysis was performed as arranged in Figs 10(a) and 10(b) and using two identical proposed antennas as the Tx and Rx. The normalized amplitudes of both input and output pulses for the two different configurations are shown in Fig. 10(c). There are no apparent ringing effects in the input and output signals (Fig. 10(c)) and just output signals decaying over the time. Fidelity factor (FF), which defines the cross-correlation between transmitted and received signals, is calculated using (2):

(2)$$FF = \rm{max} \left[{\int_{-\infty}^{\infty} S_t( t) S_r( t + \tau) d\tau\over \int_{-\infty}^{\infty} \vert S_t( t) \vert ^2 dt \int_{-\infty}^{\infty} \vert S_r( t) \vert ^2 dt} \right].$$

where S _t(t) and S _r(t) indicate transmitted and received signals and τ the group delay. The computed values of FF are listed in Table 3.

Fig. 10. Time-domain analysis of the proposed antenna: (a) face-to-face configuration, (b) side-by-side configuration, (c) input and output signals, (d) group delay, (e) phase, and (f) isolation.

Table 3. FF of the proposed antenna

Group delay is a measure of the transition time taken by the signal to pass through a device and it signifies the negative rate of transfer function phase with respect to the frequency [Reference Wiesbeck, Adamiuk and Sturm32]. Mathematically, group delay is expressed as

(3)$$\tau_g ( \omega) = -{d\phi( \omega) \over d\omega} = -{d\phi( \omega) \over 2\pi df}.$$

Simulated results of τ_g(ω) for both the configurations are shown in Fig. 10(d). Practically, for SWB antennas group delay should be within 1 ns [Reference Cho, Kim, Choi, Lee and Park33] to ensure the phase linearity in the far-field region and no pulse distortion is present.

Isolation is defined by the ratio of incident power on input port and delivered power at the output port. This also characterizes the coupling between both the ports. A high value of magnitude of isolation, |S ₂₁|, represents an uncorrelated transmission between the ports. The simulated response for both the configurations is shown in Fig. 10(f). It can be observed from the figure that S ₂₁ ≤ −25 dB for the operating range.

Figure 10(e) shows phase variation of S ₂₁ with respect to frequency. A linear variation of phase signifies that there is no out of phase component in the received signal within the operating range. For the proposed antenna, a linear variation of the phase is recorded within its operating range.

MIMO configuration

To exploit polarization diversity, the proposed antenna is arranged in a 4×4 MIMO configuration, next. Four elements, E _i ∈ i = 1, 2, 3, 4, are etched on an FR-4 substrate of length L _4E = 56 mm, width W _4E = 56 mm, and thickness 1.6 mm with a structure shown in Fig. 11.

Fig. 11. MIMO configuration for polarization diversity: L _4E = 56 mm, W _4E = 56 mm, G _4E = 5 mm.

A polarization shift between adjacent elements that form the array is 90° given the arranged orientation of these elements (Fig. 11). A gap, G _4E, of 5 mm is used between the adjacent elements to accord with the convention that gap ≥ λ/2 for low mutual coupling.

The simulated reflection coefficient and isolation of the array are given in Fig. 12 and show confirmation of the proposed SWB response (from 2.07 to 30 GHz) with isolation between ports ≥ 22.5 dB.

Fig. 12. Simulated reflection coefficient and isolation.

To validate the results, the prototype was fabricated. This is shown in Fig. 13 and the corresponding reflection coefficient and isolation between ports in Figs 15(a) and 15(b), respectively.

Fig. 13. Fabricated prototype of 4 × 4 MIMO array: (a) front view and (b) back view.

Fig. 14. Measured reflection coefficients of the array.

Fig. 15. Measured isolation parameters of the array.

Measurements corroborated simulations; the insignificant discrepancies were due to fabrication tolerances. Since the proposed polarization diversity configuration has array elements which are arranged in a quadratic orientation, hence, the radiation characteristics are expected to mimic the same behavior. The E-plane (ϕ = 0°) radiation pattern of E ₁ is therefore equal to H-plane (ϕ = 90°) radiation pattern of E ₂ and vice versa. To validate this, the simulated and measured radiation patterns of the array at four distinct frequencies are shown in Figs 16(a)–16(d). It can be observed from the results that simulated and measured results are in good agreement and also satisfy the expected behavior of polarization diversity configuration.

Fig. 16. Simulated and measured radiation patterns of the array at (a) 5 GHz, (b) 13 GHz, (c) 21 GHz, and (d) 28 GHz.

The performance parameters of the array are evaluated next.

Envelope correlation coefficient and diversity gain

Envelope correlation coefficient (ECC) is one of the key parameters to evaluate the MIMO antenna performance that provides the degree of similarity between received signals. The ECC should ideally be zero and practically as low as possible within the operating frequency (f ₀) range. The simulated and measured ECC performance of the 4 × 4 array is shown in Figs 17(a) and 17(b), respectively.

Fig. 17. ECC performance of the array: (a) simulated and (b) measured.

In a uniform propagation environment, the ECC is formulated in terms of S-parameters [Reference Blanch, Romeu and Corbella34] as

(4)$$\rho_{ij} = \displaystyle{\vert S^\ast _{ii}S_{ij} + S^\ast _{\,ji}S_{\,jj}\vert ^2\over ( 1-( \vert S_{ii}\vert ^2 + \vert S_{\,ji}\vert ^2) ) ( 1-( \vert S_{\,jj}\vert ^2 + \vert S_{ij}\vert ^2) ) }$$

In general, the ECC ≤0.5 is acceptable [Reference Koohestani, Moreira and Skrivervik35] signifying the admissible threshold level of signal distortion. From Fig. 17, it can be noted that simulated value of ρ_ij for adjacent and diagonal elements is ≤ 0.0015 and ≤ 0.003 respectively. Furthermore, the measured values of ρ_ij are ≤ 0.002 and ≤ 0.005 for adjacent and diagonal elements within the operating range. This shows that no signal distortion is present for the array (polarization diversity application). Additionally, low value of ECC indicates good isolation between the array elements, which in turn is helpful for simultaneous operation.

Diversity gain (DG) indicates the improvement obtained from MIMO compared to SISO (single-input single-output) and calculated by the following relation [Reference Chandel, Gautam and Rambabu36]

(5)$$\rm{DG~( dB) } = 10\times \log_{10}( 10\sqrt{1-ECC^2}) $$

For satisfactory performance, DG ≥ 9.95 dB is desired [Reference Rafique, Agarwal, Nauman, Khalil and Ullah37]. The simulated and measured DG of the array is shown in Fig. 18 and is evident that DG ≈ 10 dB for both simulation and experimental results within the operating range.

Fig. 18. Diversity gain of the array: (a) simulated and (b) measured.

Mean effective gain

The mean effective gain (MEG), in practice, is defined as the ratio of power received by MIMO antenna elements to the power received by an isotropic antenna under fading environment. Analytically, the MEG for different element ports can be computed [Reference Kumar, Ansari, Kanaujia and Kishor38] as following with mentioned acceptable limit:

(6)$$\rm{MEG}_i = 0.5\left[1-\displaystyle\sum_{\,j = 1}^{N} \vert S_{ij}\vert ^2\right]< -3\, \rm{dB}$$

and

(7)$$\vert \rm{MEG}_i - \rm{MEG}_j\vert \leq 3\, \rm{dB}$$

The simulated and measured values of MEG₁ to MEG₄, and the corresponding differences between different ports are plotted in Figs 19(a)–19(f).

Fig. 19. Simulated and measured MEG of the array (a, b) between ports (1, 2), (c, d) between ports (1, 3), and (e, f) between ports (2, 4).

It is concluded from Fig. 19 that the difference between the MEG of two adjacent or diagonal array elements is below the acceptable limit of 3 dB.

Channel capacity loss

Channel capacity loss (CCL) denotes the highest transmission rate at which a signal can be continuously transmitted over a communication channel. For faithful transmission, the limit is 0.4 bits/s/Hz within the operating BW [Reference Choukiker, Sharma and Behera39]. The simulated and measured CCL of the array is shown in Fig. 20.

Fig. 20. Simulated and measured CCL of the array.

From the results, it can be concluded that CCL falls within the acceptable limit for the entire operating band.

Table 4 shows a comparison study of the proposed array with the literature. Benefiting from compact SWB antenna elements proposed in this paper, the results show that proposed array is miniaturized by 6.78% compared to the smallest prototype [Reference Raheja, Kanaujia and Kumar43] with wider BW and comparable isolation compared to [Reference Wu, Yuan and Wu40–Reference Kumar, Lee, Kim, Mohyuddin, Choi and Kim45].

Table 4. Comparison study of the proposed array with the literature

Here, λ_l and λ_w are normalized length and width (in wavelengths) corresponding to the lowest operating frequency.