Introduction

Wideband antennas play a great role in the commercialization of any wireless technology. With the increasing demand for various services depending on the data rate, modern wireless technology always comes with newly defined bands with enhanced channel capacity. This challenge can be properly handled by designing wideband antennas that can cover narrow bands, facilitating systems with less complexity, cost, and device size, and reducing the number of antennas required for a portable device. Moreover, these antennae consume less power than narrowband antennae, causing low interference. These antennae are very demandable for these benefits in applications like precision localization, medical imaging, ground penetrating radar, indoor multimedia communications, and sensor networks1,2,3,4,5,6,7,8. At the same time, if the antenna size is compact, it becomes an added advantage for the device manufacturers. Various techniques are investigated when designing compact and wideband antennae. Techniques are applied for microstrip antennas as well as other types of antennas. A square ring with a specific height with the utilization of a shorted strip gives enhanced bandwidth with increased gain though the percentage of the bandwidth is poor9. Various techniques are also applied in patch antennas. The defected ground structure (DGS) is one of the most prominent techniques for designing compact and wideband antennae. In this technique, various shapes are cut from the ground plane to cause one or more defects. In one research work, a triangle-shaped structure loaded with a section tapered slot is cut from the ground plane to reduce the antenna size. The compactness achieved here is slightly less than half wavelength but with high bandwidth10. The same level of miniaturization is achieved by utilizing an annular ring slot on the metallic ground plane, where the slot is the Koch snowflake structure of the third iteration. The compactness achieved here is slightly larger than half wavelength, and the bandwidth is also not so wide11. Some researchers cut a simple rectangular slot from the ground plane to get wideband property, whereas they achieved compactness by proposing a rectangular ring monopole antenna. But this multiband antenna shows narrow bandwidths at their lower bands12. Sometimes, the DGS structure is loaded with differently shaped structures to reduce the antenna size. Insha Ishteyaq et al. cut a rectangular block from the ground plane, and then a T-shaped structure is added at the upper edge of the DGS, which causes a low-frequency resonance. Though the bandwidth is much better than the previous work, it is still not enhanced13. Another promising technique for getting compactness is the excitation of different-shaped patches by a coplanar waveguide. Azimov and his team proposed a quarter-wave monopole where the inverted L-shaped stub of the monopole is responsible for the size reduction. The compactness achieved is less than one-fifth of the wavelength, but the bandwidth reported is very poor14. A semicircular shape with a C-shaped slot is also investigated with a coplanar waveguide to design a compact antenna. The coplanar ground used here is triangular15. Another triangular-shaped monopole connected with three inverted L-shaped stubs is proposed with a coplanar waveguide where the largest stub causes low-frequency resonance16. All these techniques suffer from narrow bandwidths. Some designs with coplanar waveguides achieve compactness as well as broad bandwidth. Mottahir et al. used asymmetrical coplanar ground comprising one inverted L-shaped and another inverted U-shaped structure to get wideband behaviour. The inverted U-shaped structure causes the bandwidth to shift at low frequency17. A slotted patch with a partial ground plane is another technique to get compact and wideband antennae. A group of researchers proposed a rectangular patch with an inverted L-shaped slot connected by a slotted circular ring backed by a partial ground plane. The L-shaped slot considerably reduces the current antenna path18. However, the bandwidth enhancement is negligible. On the other hand, bandwidth enhancement is achieved by Rezaul Azim and his group by cutting multiple slots in a rectangular patch where they use the partial ground plane as the reflector19. Researchers have also investigated structures with different shapes using partial ground planes. Tapan Nahar and his research team proposed a leaf-shaped antenna with a partial ground plane. They found considerable size reduction and bandwidth enhancement due to partial ground20. But the bandwidth enhancement is not too much. In another research, three circles of different radii are connected to get considerable miniaturization. Using a single circular patch resonates at a high frequency, but adding other circular patches shifted the resonance towards a very low-frequency region. At the same time, it is evident that the partial ground plane helps to enhance the bandwidth21. However, the bandwidth of the low-frequency region is not good. A circular-shaped patch is utilized as a main radiating element of a design where the partial ground plane with an open slot is used as the reflector. The partial ground plane generates low-frequency resonance, and the open slot enhances the bandwidth22. Moreover, a group of researchers investigated a multi-branch structure with a partial ground plane where the longest length branch reduces antenna size substantially. In addition, in this prototype, an L-shaped branch protrudes from the ground plane, considerably enhancing the bandwidth23. Another multi-branch structure was proposed by Arpan Desai et al., but they used a full ground plane. As a result, the multiband property is achieved where the bandwidths of those bands are very poor24. Besides these techniques, metamaterial structures are also used to design compact antennas, but the problem is that the bandwidth of those antennas is very poor, and the gain is minimal. Array structures can be used to enhance the gain. In this paper, a two-element array of a metamaterial elliptical split ring resonator (SRR) is designed with a unique matching circuit and a modified ground plane. The ground plane is modified in two stages. In the first stage, a slotted rectangular section is added at the left side of the partial ground plane to form a defected ground plane. In this stage, three parasitic blocks are placed at the lower right edge of the DGS section. The matching circuit with this ground modification achieves an impedance bandwidth of about 70%.

Antenna geometry and operation principle

Fig. 1
figure 1

Configuration of the proposed antenna. (a) Front view (Patch) (b) Back view (Ground).

Full size image

The proposed antenna is fabricated on a low-cost FR4 substrate with a thickness of 1.6 mm. The substrate dielectric constant is 4.3. The final structure of the antenna is presented in Fig. 1, and the detailed dimensions are listed in Table 1. The wideband antenna design is based on the resonance characteristics of the metamaterial unit cell. Later, an appropriate feeding network is applied to design a single-element antenna to get similar resonance behaviour. Then, a two-element array antenna is designed with an optimized matching network to get a closely spaced dual-band response. In the last stage, the DGS structure and parasitic rectangular blocks are added to the ground plane to widen the bandwidth and improve the overall matching of the entire band. A detailed analysis of the operation principle is given in the subsequent sections.

Table 1 Detailed dimensions of the proposed antenna.
Full size table

Elliptical slotted metamaterial unit cell analysis

The metamaterial unit cell is a rectangular-shaped split ring resonator (SRR) in which an elliptical-shaped slot is cut in the middle of the patch to form the ring. Later, a split of 0.7 mm thick and 1.03 mm long is cut at one of the sides along the Z-axis of the ring.

Fig. 2
figure 2

S-parameters of the metamaterial unit cell with different setups and the single-element antenna.

Full size image

The radiation characteristic of this unit cell is checked by the CST simulation software with two different boundary conditions since the radiation property for this resonator can give different results for these two boundary conditions25,26,27,28. The simulated results of the transmission coefficients with electrical polarization are presented in Fig. 2. Setup 1 of the boundary conditions, which applies the E-field along the x-axis and the H-field along the z-axis, the unit cell causes a single resonance at a high frequency around 6.4 GHz. In this case, the electric field is polarized along the no-slit bearing side. On the other hand, when the E-field is applied along the Y-axis and the H-field along the X-axis, which is Setup 2, the unit cell resonates at a low frequency of around 2.4 GHz, and in this case, a strong Electric field is polarized along the split. The single-element antenna is formed using this unit cell with an appropriate feeding network to get a similar electric polarization like setup 2 so that resonance occurs at a low frequency around 2.5 GHz. The bandwidth and gain of this antenna are very low. So, an array of two elements is designed, and additional circuitry is applied to improve these parameters.

Design evolution of the array antenna

Designing a two-element array improves the performance of a single-unit cell antenna, especially the gain and bandwidth. The final array antenna is achieved by applying some revolutionary steps, as shown in Fig. 3. Antenna 1 is a simple two-element array antenna. An appropriate feeding network is used to simulate this antenna, as shown in Fig. 3 (a). Only a single band within the 2 and 3 GHz frequency spectrum is found, as depicted in Fig. 4. The bandwidth and gain of this band are higher than the single-element antenna. However, a dip is also observed around the 3 GHz position. A further modification is applied to the design to increase the bandwidth and gain since the antenna is aimed at operating in the sub-6 GHz region. For this purpose, a T-shaped slotted structure is inserted between the two elements, as shown in Fig. 3 (b). This slotted structure causes two closely spaced bands within the 2 and 3 GHz frequency spectrum and another within the 4 and 5 GHz frequency region. It also introduces a resonance dip in the high-frequency (around 6 GHz) region. However, the impedance matching of these bands is poor. After that, the ground plane is modified to improve the impedance matching. An open-slotted rectangular block is added to the left edge of the partial ground plane (Antenna 3 in Fig. 3(c)), which rejects the previous low-frequency band (2 and 3 GHz) but significantly improves bandwidth and matching of the last high-frequency band within the 4 and 5 GHz. Now, the band with the enhanced bandwidth appears within the 3 and 5 GHz frequency positions. At the same time, it introduces a new band within the 5 and 5.5 GHz range and improves the band’s matching to around 6 GHz. Last, three parasitic rectangular blocks are placed just above the partial ground plane with a 0.5 mm spacing between them and 0.8 mm far from the right side of the open-slotted rectangular block, as shown in Fig. 3 (d). These blocks improve the matching of the bands within the 5 and 5.5 GHz and the band around the 6 GHz so that these two bands combine to work within the 5 and 6.3 GHz frequency spectrum. The blocks also change the matching of the low-frequency band’s higher cut-off frequency and the lower cut-off frequency of the high-frequency band in such a manner as to combine them to form a single band with a very high bandwidth within the 3 and 6.5 GHz frequency spectrum. The design evolution also affects the gain at every step. In the beginning, Antenna 1 gain is very low, with a peak gain of less than 0.5 dBi. The gain is enhanced when the matching network is added in Antenna 2, where the pick gain is around 2 dBi. As soon as the DGS structure is applied in Antenna 3, the gain is enhanced considerably, and the reported pick gain is 3.2 dBi. Finally, Antenna 4 causes some gain reduction in the band at the higher frequency region, but the gain reduction is little. A comparison table of bandwidth and gain between different antennas are listed in Table 2.

Fig. 3
figure 3

Design evolution of the proposed antenna. (a) Antenna 1 (b) Antenna 2 (c) Antenna 3 (d) Antenna 4.

Full size image
Fig. 4
figure 4

Comparison of (a) Reflection coefficient and (b) realized gain of different antennas.

Full size image
Table 2 Comparison of bandwidth and of different antennas.
Full size table

Parametric analysis

The proposed final design with high bandwidth and maximum gain is achieved by performing an extensive parametric analysis of three structures: the matching circuit between the array elements, the rectangular-shaped slotted block located on the ground plane, and the rectangular-shaped parasitic blocks on the ground plane. Among them, the second one has the most influence on the bandwidth and gain of the final prototype.

Matching circuit analysis

The T-shaped matching circuit is placed between the array elements. As shown in Fig. 4, this matching circuit causes multiple resonances in the low and high-frequency spectrum without loading DGS and parasitic circuits on the ground plane. So, this circuit can change the matching of the array elements. A parametric analysis is done on the dimensions of the structure to see the effect on the antenna impedance matching with the presence of DGS and parasitic structures, as shown in Fig. 5. It is found that when the slot thickness MT of the middle portion of the matching circuit decreases, the matching of the low and high-frequency dips improves except for the matching of the middle band. The optimized value of 0.7 mm is chosen because the gain of the middle band improves from the negative to positive value, as shown in Fig. 5 (a). On the other hand, when the slot length ML increases, the matching of the high-frequency dip and the middle bands improves, but the matching of the low-frequency dip degrades. An optimized value of 2 mm is chosen because the gain at the low-frequency dip is the best, with better matching through the entire bandwidth, as depicted in Fig. 5 (b). The upper portion of the matching network also significantly affects the low-frequency band. When the length of the slot, M2L, increases, the position of the resonance point of the low-frequency band shifts to the low-frequency region and at the same time, the matching of the middle band and high-frequency dip degrades. However, an optimized value is chosen to be 1.5 mm, for which the band starts just after 3 GHz, and the matching of the overall band is acceptable, as presented in Fig. 5 (c).

Fig. 5
figure 5

Parametric analysis of the (a) thickness and (b) length of the slots of the middle portion of the matching structure and (c) length of the slots of the upper portion of the matching structure on reflection coefficient.

Full size image

DGS structure analysis

The defected ground structure has many applications for improving antenna performance, especially in enhancing bandwidth29,30. In general, these structures have the potential to generate additional resonance points or to improve the impedance matching of multiple resonances so that several resonances can be positioned close enough to form a wide band31,32. In this case, the loading of the DGS structure also improves the matching of high-frequency bands and controls the position of their cut-off points to form a wide band.

Fig. 6
figure 6

Parametric analysis of the (a) length and (b) thickness of the slots of the DGS structure on the reflection coefficient.

Full size image

Parasitic blocks analysis

Fig. 7
figure 7

Parametric analysis of (a) the width and (b) the thickness of parasitic blocks on the reflection coefficient.

Full size image

A parametric study is done on the length and thickness of the DGS slots to clearly understand their effects on the operating bandwidth, as shown in Fig. 6. When the length SL of the slot increases, the matching of the high-frequency dips is radically improved, as presented in Fig. 6 (a). The optimized value of 5 mm joints for all three bands and acceptable impedance matching. Meanwhile, the thickness, ST, influences the lowest and highest dips of the band, as shown in Fig. 6 (b). As the value of ST increases, the matching of both deep improves. A value of 0.7 mm causes the best matching of the highest dip, but the bandwidth enhancement with good matching is more than that with the value of 0.5 mm. Further improvement of the impedance matching of the high-frequency band is done by loading three rectangular blocks. Another parametric analysis is done on the dimensions of the rectangular blocks to clarify their effects, as shown in Fig. 7. As the width, W3 of the blocks increases, the matching of the high-frequency dip improves, and the matching of the middle band degrades. However, there is no effect on the low-frequency band, as shown in Fig. 7 (a). The same effect is observed in the case of an increment of the thickness, L8, of the blocks. The above three parametric analyses reveal that the matching structure forms the low-frequency band. This structure’s dimension mainly affects this band’s matching and position. So, it can be said that it has a vital role in the size reduction of the prototype. On the other hand, the structures on the ground plane, namely the DGS structure and the parasitic blocks, are mainly responsible for forming the two high-frequency bands. In particular, the dimension of the DGS structure significantly improves the matching of the middle and high-frequency bands. Finally, the parasitic blocks are mainly placed to improve the matching of the high-frequency band. An analysis of the real input impedance and imaginary input impedance for the DGS structure and parasitic blocks are also presented in Fig. 8(a) and 8(b) respectively. From this analysis, it is evident that the addition of DGS and the parasitic blocks significantly keeps the value of real input impedance near 50 ohms and flattens the value of imaginary input impedance towards the value of 0 ohms overall the entire bandwidth. For this reason, the impedance bandwidth of the entire bandwidth improves significantly.

Fig. 8
figure 8

Effect of DGS and Parasitic blocks on the (a) Real Input Impedance and (b) Imaginary input Impedance.

Full size image

Electric field distribution

Fig. 9
figure 9

The electric field distribution of the proposed antenna at (a) 3.58 GHz and (b) 5.66 GHz.

Full size image

The generation of the resonances can also be described from their electric field distributions. The electric field distributions at 3.58 GHz and 5.66 GHz are given in Fig. 9. As anticipated, the matching circuit is responsible for the low-frequency resonance. The field distribution is more acute on the left patch, the partial ground plane’s left side, and the matching structure’s left portions. More clearly, the first resonance is mainly assisted by the matching circuit’s left portion, where the DGS circuit is not beneath the patch. On the contrary, at high frequency, i.e., at 5.66 GHz, the electric field density is higher on the right side of the array and the DGS structure, including the right portion of the partial ground plane and the parasitic blocks. At this frequency, the electric field density is reduced on the right element of the array and on the matching circuit, which means that the DGS structure controls the high-frequency resonance.

Fig. 10
figure 10

The reflection coefficient of the antenna without the matching circuit.

Full size image

The configuration and its reflection coefficient shown in Fig. 10 also prove that the array antenna cannot generate the low-frequency resonance even in the presence of the DGS structure if the matching circuit is not loaded. On the other hand, the reflection coefficient of Fig. 3 (b) configuration proves that the array antenna cannot produce high-frequency resonance even in the presence of the matching network if the DGS network is not loaded on the ground plane.

Fig. 11
figure 11

Surface current distribution of the antenna (a) without parasitic blocks and (b) With parasitic blocks.

Full size image

The surface current of the antenna with and without parasitic blocks are also investigated which is shown in Fig. 11. It is observed that in the feeding network just positioned above the parasitic blocks, the current flow is slightly disturbed but no considerable change of current flow in the ground plane is observed. But these parasitic blocks increased the current density throughout the entire structure which improves the impedance bandwidth of the high frequency band.

Equivalent circuit analysis

The equivalent circuit of the Proposed antenna can be developed from Fig. 8. Since Antenna 1 causes two dips in the lower frequency region, they can be modelled with two parallel RLC circuit connected in series with a coupling capacitor as shown in Fig. 12 (a)33. As the impedance matching the second dip is not good enough the values of R2 will be far from 50 ohm which is also observed in Fig. 8(a). In addition, the total reactance value of the second parallel RLC circuit is also far from the 0 value as depicted in Fig. 8(b).

Fig. 12
figure 12

Equivalent circuit of (a) Antenna 1 and (b) Antenna 2, Antenna 3 and Antenna 4.

Full size image

After loading the matching structure antenna 2 causes three dips which can be modelled as three parallel RLC circuits connected in series as shown in Fig. 12 (b). Later when the DGS structure and parasitic blocks are added the impedance matching improves for the entire bandwidth specially the parasitic blocks improve the impedance matching of the high frequency resonance. More specifically, Antenna 3 where the parasitic blocks are not added, the position of the low frequency resonance shifted to the higher frequency region. At high frequency, the inductance (XL1) of the reactive component of the input impedance of the first RCL circuit increases, and the capacitance (XC1) decreases considerably. These changed reactance values cause a change in the resistance (R1) value for a specific frequency within that band. For Antenna 3, the resistance and reactance values (R3, XL3, XC3) of the middle resonance changes negligibly. Later after adding the metallic parasitic blocks the resistance and reactance of the input impedance of the high frequency resonance (R2, XL2, XC2) changes due to the coupling of the metallic blocks with metallic ground plane which in turn improves impedance matching of that band.

Results and discussion

Fig. 13
figure 13

(a) The fabricated prototype of the antenna, (b) VNA and (c) Satimo measurement system of the proposed fabricated prototype.

Full size image
Fig. 14
figure 14

Simulated and measured (a) reflection coefficient and (b) realized gain of the proposed antenna.

Full size image

The analysis done above is validated with the experimental data of the prototype. For this purpose, the final optimized design has been fabricated and measured. The reflection coefficient of the prototype is measured using a performance network analyzer (PNA) series vector network analyzer (E8362C 10 MHz–67 GHz), and the radiation characteristics, efficiency, and realized gain are measured using a Satimo Star-Lab near-field antenna measurement system. The images of the two layers of the fabricated prototype and the measurement setups of the VNA and Satimo are shown in Fig. 13.

Fig. 15
figure 15

Radiation patterns of the proposed antenna at (a) 3.58 GHz, (b) 4.5 GHz and (c) 5.66 GHz.

Full size image

The simulated and measured reflection coefficients and realized gains are presented in Fig. 14. Both measured results are in good agreement with the simulation. The measurement results show that the proposed antenna exhibits a fractional bandwidth of 70% for |S11| ≤ -10 dB. The bandwidth’s upper and lower cut-off frequencies are 3.04 GHz and 6.32 GHz, respectively. However, a small portion within this band starts from 5.08 GHz to 5.32 GHz, where the maximum shift from the − 10 dB line is only 0.73 dB, which can be ignored. A peak realized gain of 3.07 dBi is experimentally recorded in this region.

The radiation patterns in the H and E planes of the proposed antenna are also investigated at three different frequency positions namely 3.58 GHz, 4.5 GHz and 5.66 GHz to examine the stability. The simulated and measured radiation patterns in the E and H planes are shown in Figs. 15. The simulated patterns are in good agreement with the measured patterns. The experimental E plan radiation patterns at the selected frequency positions are like dumbbell-shaped radiation pattern. The measured cross-polarization level at 3.58 GHz, 4.5 GHz and 5.66 GHz frequencies is low, less than − 16 dB, -13 dB and − 11 dB respectively. On the other hand, the measured H-field patterns at 3.58 GHz and 4.5 GHz are omnidirectional and at 5.66 GHz frequency it is almost omnidirectional. A comparison between the recently published works with the present is given in Table 3. It is evident from the table that the very compact design does not provide wide bandwidth. In particular, ref24 proposed a very compact design, but all bands’ bandwidths are very narrow. Only ref17 proposed design is compact, and the bandwidth of the low-frequency band is very high. But the work used CPW feeding, which has several significant drawbacks: band maximum, transmission loss, high cross polarization levels and lower radiation efficiency34,35. For this reason, this antenna gives one of the most prominent solutions for the sub-6 GHz applications of 5G with a wide bandwidth of almost 70% and a peak realized gain of 3.07 dBi.

Table 3 Performance comparison of the proposed antenna with other recently published compact and wideband antennas.
Full size table

Conclusion

This paper investigates a two-element array of an elliptical metamaterial unit cell with a matching circuit and a modified ground plane. The elliptical unit cell causes low-frequency resonance but with poor bandwidth and gain. A two-element array of this unit cell causes multiple deeps but with poor bandwidth and matching. The addition of the slotted matching circuit gives two nearby resonances in the sub-6 GHz region. The DGS section makes these resonances position as close as possible to achieve a high bandwidth. Finally, the parasitic blocks are added to improve the band’s matching near the higher cut-off frequency. The measured impedance bandwidth of the prototype is about 70%, ranging from 3.05 GHz to 6.32 GHz, with a peak realized gain of 3.07 dBi. The radiation patterns are stable, and the cross-polarization level in the E-plan is less than − 16 dB at the lower resonant frequency. The antenna’s bandwidth covers the 5G NR bands n77, n78, n79, n46, n47, and n48 of the sub-6 GHz frequency spectrum. Although the orientation of the radiation patterns in the E-planes at three selected frequencies are not the same, it will not cause significant signal loss in receiving end or more specifically portable device applications of patch antenna.