Low-Profile Broadband Dual-Polarized Dipole Antenna for Base Station Applications

A low-profile broadband dual-polarized antenna is investigated for base station applications. It consists of two orthogonal dipoles, fork-shaped feeding lines, an artificial magnetic conductor (AMC), and parasitic strips. By utilizing the Brillouin dispersion diagram, the AMC is designed as the antenna reflector. It has a wide in-phase reflection bandwidth of 54.7% (1.54–2.70 GHz) and a surface-wave bound range of 0–2.65 GHz. This design effectively reduces the antenna profile by over 50% compared to traditional antennas without an AMC. For demonstration, a prototype is fabricated for 2G/3G/LTE base station applications. Good agreement between the simulations and measurements is observed. The measured −10-dB impedance bandwidth of our antenna is 55.4% (1.58–2.79 GHz), with a stable gain of 9.5 dBi and a high isolation of more than 30 dB across the impedance passband. As a result, this antenna is an excellent candidate for miniaturized base station antenna applications.

Dual-polarized planar dipole antennas have attracted considerable attention because of their wide bandwidth and pattern stability. Low-profile designs can be achieved using bowtie-shaped crossed-dipole antennas and Huygens dipole antennas, with profiles of 0.088λ 0 [8] and 0.0483λ 0 [9], respectively (where λ 0 represents the wavelength in free space at the center frequency). However, they can provide narrow −10-dB impedance bandwidths of only 15.6% [8] and 0.462% [9]. To enhance the antenna bandwidth, parasitic elements are used in antenna design. The common parasitic element can be L-shaped metal strips [10] or quadrangular loops [11,12]. They act as additional resonating elements, creating a coupling effect [10] or adding a new resonant mode [11,12]. As a result, the overall bandwidth significantly improved. The parasitic U-shaped grooves are also utilized to broaden the bandwidth [13,14]. Grooves can modify the current distribution on the main radiating element and, thus, improve the antenna impedance. However, the profile of those dipole designs is around a quarter wavelength due to the conventional conductor reflector. The profile is required to eliminate the effect of reflected electromagnetic waves Figure 1 shows the configuration of the wideband dual-polarized antenna. With reference to Figure 1a,b, this antenna comprises crossed-dipoles, double-layer AMC, fork-shaped feeding structures, and parasitic strips. Two crossed dipoles are positioned perpendicularly with slant angles of ϕ = ±45 • and are represented as dipoles 1 and 2. Figure 1c shows the exploded view of the proposed antenna. It can be seen from the figure that the feeding lines and crossed dipoles are printed on Layers 1 and 2 of Substrate 1 (Sub. 1), respectively. Figure 1d zooms in view of the arms of dipoles and feeding structures. As can be seen from the figure, each arm of the dipole is a hexagonal ring with a width of w 1 = 3.6 mm and four mirror-symmetrical etched slots. It should be mentioned that one of the dipole arms Is soldered directly to the feeding structures, while the other arm is fed via space coupling. This method is designed to obtain a wide −15-dB impedance matching bandwidth. The novel AMC consists of two metal layers, one substrate, and air layers. Periodic patches are its first metal layer printed on Layer 3 of Sub. 2, with the interval gap of g = 2 mm and Sensors 2023, 23, 5647 3 of 14 patch width of w = 13 mm. The second metal layer is aluminum ground. Sub. 1 and Sub. 2 are FR4 substrates (ε r = 4.4, tanδ = 0.02). Crossed dipoles are placed at a height of 5.5 mm from the AMC surface. Plastic pins are used to locate layers precisely. Figure 2 shows the different layers of the dual-polarized antenna. With reference to the figure, our parasitic strips are printed on Layer 2 of Substrate 1 to reduce cross-polarization.  Figure 2 shows the different layers of the dual-polarized antenna. With reference to the figure, our parasitic strips are printed on Layer 2 of Substrate 1 to reduce cross-polarization.   Figure 2 shows the different layers of the dual-polarized antenna. With reference to the figure, our parasitic strips are printed on Layer 2 of Substrate 1 to reduce cross-polarization.   Figure 3 shows the configurations and reflection coefficients of different AMC units. The proposed AMC is a patch-type with a patch length and width of w = 13 mm, a substrate-layer height of hd = 1.2 mm (0.0088λ0), and an air-layer height of ha = 12 mm (0.088λ0). AMC1 and AMC2, which are conventional AMCs with the same patch size and gap width but different substrate-layer heights (hd1 = 1.2 mm, hd2 = 9.5 mm), were simulated for comparison. Generally, the in-phase reflection bandwidth is the frequency range in which the reflection phase is between −π/2 and π/2. With reference to the figure, the in-phase reflection bandwidths are 9.8%, 39.4%, and 54.7% for AMC1 (4.57-5.04 GHz), AMC2 (1.67-2.49 GHz), and the proposed AMC (1.54-2.70 GHz), respectively. Compared with the proposed AMC, AMC1 lacks the air layer, giving a zero-phase reflection frequency of 4.82 GHz that is out of the operating bandwidth of 2G/3G/LTE. It is worth mentioning that AMC2 has almost the same zero-phase reflection frequency, however, it has a narrower inphase bandwidth and a worse reflection coefficient than the proposed AMC. Therefore, our AMC is a good alternative to the metal ground due to its wideband.  Figure 3 shows the configurations and reflection coefficients of different AMC units. The proposed AMC is a patch-type with a patch length and width of w = 13 mm, a substratelayer height of h d = 1.2 mm (0.0088λ 0 ), and an air-layer height of h a = 12 mm (0.088λ 0 ). AMC 1 and AMC 2 , which are conventional AMCs with the same patch size and gap width but different substrate-layer heights (h d1 = 1.2 mm, h d2 = 9.5 mm), were simulated for comparison. Generally, the in-phase reflection bandwidth is the frequency range in which the reflection phase is between −π/2 and π/2. With reference to the figure, the in-phase reflection bandwidths are 9.8%, 39.4%, and 54.7% for AMC 1 (4.57-5.04 GHz), AMC 2 (1.67-2.49 GHz), and the proposed AMC (1.54-2.70 GHz), respectively. Compared with the proposed AMC, AMC 1 lacks the air layer, giving a zero-phase reflection frequency of 4.82 GHz that is out of the operating bandwidth of 2G/3G/LTE. It is worth mentioning that AMC 2 has almost the same zero-phase reflection frequency, however, it has a narrower in-phase bandwidth and a worse reflection coefficient than the proposed AMC. Therefore, our AMC is a good alternative to the metal ground due to its wideband.  Figure 3 shows the configurations and reflection coefficients of different AMC units. The proposed AMC is a patch-type with a patch length and width of w = 13 mm, a substrate-layer height of hd = 1.2 mm (0.0088λ0), and an air-layer height of ha = 12 mm (0.088λ0). AMC1 and AMC2, which are conventional AMCs with the same patch size and gap width but different substrate-layer heights (hd1 = 1.2 mm, hd2 = 9.5 mm), were simulated for comparison. Generally, the in-phase reflection bandwidth is the frequency range in which the reflection phase is between −π/2 and π/2. With reference to the figure, the in-phase reflection bandwidths are 9.8%, 39.4%, and 54.7% for AMC1 (4.57-5.04 GHz), AMC2 (1.67-2.49 GHz), and the proposed AMC (1.54-2.70 GHz), respectively. Compared with the proposed AMC, AMC1 lacks the air layer, giving a zero-phase reflection frequency of 4.82 GHz that is out of the operating bandwidth of 2G/3G/LTE. It is worth mentioning that AMC2 has almost the same zero-phase reflection frequency, however, it has a narrower inphase bandwidth and a worse reflection coefficient than the proposed AMC. Therefore, our AMC is a good alternative to the metal ground due to its wideband. The proposed AMC unit can be equivalent to an LC parallel resonant circuit using the equivalent circuit analysis method (ECM) [23,24], as illustrated in Figure 4. Periodical patches provide the grid inductance L p . Another inductance L d is created by the total substrate, including the air and the dielectric. Gaps between adjacent patches form a grid capacitance C p . As a result, the resonant frequency and bandwidth of the proposed AMC can be represented by Equations (1) and (2), which show the relationship between parameters on the resonant frequency and bandwidth coefficient.

In-Phase Reflection Bandwidth of the AMC Unit
where η 0 = µ 0 /ε 0 is the wave impedance in free space. Compared with AMC 1 and AMC 2, the proposed AMC has higher profiles and, thus, a larger dielectric inductance L d . Since the derivative of Equation (2) with respect to L d is positive, the bandwidth is positively related to the dielectric inductance L d . Therefore, AMC 2 and the proposed AMC have a wider bandwidth. In addition, the dielectric substrate of AMC 2 is thicker, which increases dielectric losses and manufacturing costs. The proposed AMC has the widest bandwidth. The surface impedance of AMC can be represented by: Sensors 2023, 23, x FOR PEER REVIEW 5 of 14 The proposed AMC unit can be equivalent to an LC parallel resonant circuit using the equivalent circuit analysis method (ECM) [23,24], as illustrated in Figure 4. Periodical patches provide the grid inductance Lp. Another inductance Ld is created by the total substrate, including the air and the dielectric. Gaps between adjacent patches form a grid capacitance Cp. As a result, the resonant frequency and bandwidth of the proposed AMC can be represented by Equations (1) and (2), which show the relationship between parameters on the resonant frequency and bandwidth coefficient.
is the wave impedance in free space. Compared with AMC1 and AMC2, the proposed AMC has higher profiles and, thus, a larger dielectric inductance Ld.
Since the derivative of Equation (2) with respect to Ld is positive, the bandwidth is positively related to the dielectric inductance Ld. Therefore, AMC2 and the proposed AMC have a wider bandwidth. In addition, the dielectric substrate of AMC2 is thicker, which increases dielectric losses and manufacturing costs. The proposed AMC has the widest bandwidth. The surface impedance of AMC can be represented by: At the parallel resonant frequency (using Equation (1)), the denominator of Equation (3) is equal to zero, and thus, the surface impedance ηs tends to infinity. The reflection coefficient of AMC can be calculated by where equals the impedance of air. Therefore, the reflection coefficient equals one, and the reflection phase is zero. AMC achieves the in-phase reflection at the resonant frequency.

Brillouin Dispersion Diagram of the AMC Unit
Generally, a surface wave will be excited when electromagnetic waves reflect from At the parallel resonant frequency (using Equation (1)), the denominator of Equation (3) is equal to zero, and thus, the surface impedance η s tends to infinity. The reflection coefficient of AMC can be calculated by Γ = (η s − η 0 )/(η s + η 0 ), where η 0 equals the impedance of air. Therefore, the reflection coefficient equals one, and the reflection phase is zero. AMC achieves the in-phase reflection at the resonant frequency.

Brillouin Dispersion Diagram of the AMC Unit
Generally, a surface wave will be excited when electromagnetic waves reflect from the ground, deteriorating antenna performance. Figure 5 shows the dispersion curves of the proposed AMC unit. With reference to the figure, the electromagnetic wave under 2.65 GHz served as a slow wave and bounded in the AMC surface. However, as the frequency increases beyond 2.65 GHz, the electromagnetic wave gradually becomes a fast wave and leaks into free space. Consequently, antennas loading the proposed AMC and the conventional metal ground have nearly the same realized gain before 2.65 GHz, as shown in Figure 6a. Figure 6b shows the simulated S-parameters of the antenna. It can be seen from the figure that both the reflection coefficient and isolation are improved by using the AMC. It should be mentioned that the proposed antenna can effectively decrease the antenna profile from 0.25λ 0 to 0.13λ. It should be mentioned that the proposed antenna can effectively decrease the antenna profile from 0.25λ0 to 0.13λ.  Figure 7 shows the design evolution of the crossed dipole and compares the simulated reflected coefficient of Dipole I, II, and III.  It should be mentioned that the proposed antenna can effectively decrease the antenna profile from 0.25λ0 to 0.13λ.  Figure 7 shows the design evolution of the crossed dipole and compares the simulated reflected coefficient of Dipole I, II, and III.   Dipole III is obtained by adding rectangular slots on Dipole II. The introduced rectangular slots effectively make two different resonant modes close (seen in Figure 8c), resulting in a wide-impedance passband of 55.4% (1.58-2.79 GHz). Figure 9 also shows the Smith charts of antennas with different dipoles. The Smith chart of Dipole I has no resonant loop, while the Smith charts of Dipole II and Dipole III have 1 and 2 resonant loops, respectively. Dipole III achieves a wider impedance bandwidth by cutting arm corners and centers and adding rectangular slots.     Dipole III is obtained by adding rectangular slots on Dipole II. The introduced rectangular slots effectively make two different resonant modes close (seen in Figure 8c), resulting in a wide-impedance passband of 55.4% (1.58-2.79 GHz). Figure 9 also shows the Smith charts of antennas with different dipoles. The Smith chart of Dipole I has no resonant loop, while the Smith charts of Dipole II and Dipole III have 1 and 2 resonant loops, respectively. Dipole III achieves a wider impedance bandwidth by cutting arm corners and centers and adding rectangular slots. The equivalent circuit model of Dipole III is shown in Figure 10a. According to previous studies [25,26], the feeding structure is modeled as a parallel LC resonator circuit (Lf and Cf) and a π-shaped network (Cg1, Cg11, and Cg12). Dipole III is obtained by cutting corners and opening rectangular slots in Dipole I. Dipole I can be expressed as a parallel RLC resonator circuit (Rd1, Ld1, and Cd1). The corners and rectangular slots can be expressed as a parallel RLC resonator circuit (Rs1, Ls1, and Cs1). When one loop dipole is excited, the other acts as a coupled loop resonator. Consequently, they are configured in parallel within the circuit. The adjacent gaps between two dipoles are modeled as a π-shaped network (Cg2, Cg21, and Cg22).

Evolution of the Antenna
By utilizing the proposed equivalent circuit and circuit analysis software, we can expedite the design process of the proposed antenna, particularly when attempting to bring two resonant modes into proximity. The values of circuit parameters can be obtained by The equivalent circuit model of Dipole III is shown in Figure 10a. According to previous studies [25,26], the feeding structure is modeled as a parallel LC resonator circuit (L f and C f ) and a π-shaped network (C g1 , C g11 , and C g12 ). Dipole III is obtained by cutting corners and opening rectangular slots in Dipole I. Dipole I can be expressed as a parallel RLC resonator circuit (R d1 , L d1 , and C d1 ). The corners and rectangular slots can be expressed as a parallel RLC resonator circuit (R s1 , L s1 , and C s1 ). When one loop dipole is excited, the other acts as a coupled loop resonator. Consequently, they are configured in parallel within the circuit. The adjacent gaps between two dipoles are modeled as a π-shaped network (C g2 , C g21 , and C g22 ).
By utilizing the proposed equivalent circuit and circuit analysis software, we can expedite the design process of the proposed antenna, particularly when attempting to bring two resonant modes into proximity. The values of circuit parameters can be obtained by using a curve-fitting method. The extracted results are L f = 1.73 nH, C f = 0.10 pF, C g1 = 3.48 pF, C g11 = 0.43 pF, C g12 = 4.17 pF, R d1 = 99.9 Ω, L d1 = 1.07 nH, C d1 = 0.36 pF, R s1 = 5.02 Ω, L d1 = 1.07 nH, C s1 = 2.23 pF, C g2 = 2.45 pF, C g21 = 0.43 pF, C g22 = 4.17 pF, R d2 = 5.24 Ω, L d2 = 0.35 nH, C d2 = 3.18 pF, R s2 = 5.24 Ω, L d2 = 0.35 nH, C s2 = 2.93 pF. Figure 10b shows the reflection coefficient derived from the equivalent circuit model and that from the HFSS. Dipole III is obtained by adding rectangular slots on Dipole II. The introduced rectangular slots effectively make two different resonant modes close (seen in Figure 8c), resulting in a wide-impedance passband of 55.4% (1.58-2.79 GHz). Figure 9 also shows the Smith charts of antennas with different dipoles. The Smith chart of Dipole I has no resonant loop, while the Smith charts of Dipole II and Dipole III have 1 and 2 resonant loops, respectively. Dipole III achieves a wider impedance bandwidth by cutting arm corners and centers and adding rectangular slots. The equivalent circuit model of Dipole III is shown in Figure 10a. According to previous studies [25,26], the feeding structure is modeled as a parallel LC resonator circuit (Lf and Cf) and a π-shaped network (Cg1, Cg11, and Cg12). Dipole III is obtained by cutting corners and opening rectangular slots in Dipole I. Dipole I can be expressed as a parallel RLC resonator circuit (Rd1, Ld1, and Cd1). The corners and rectangular slots can be expressed as a parallel RLC resonator circuit (Rs1, Ls1, and Cs1). When one loop dipole is excited, the other acts as a coupled loop resonator. Consequently, they are configured in parallel within the circuit. The adjacent gaps between two dipoles are modeled as a π-shaped network (Cg2, Cg21, and Cg22).
By utilizing the proposed equivalent circuit and circuit analysis software, we can expedite the design process of the proposed antenna, particularly when attempting to bring two resonant modes into proximity. The values of circuit parameters can be obtained by AMC is employed to replace the traditional metal ground, thereby increasing the antenna gain and maintaining a low profile. Therefore, it becomes crucial to select an appropriate quantity of AMC units. This strategic selection aims to ensure that the antenna exhibits a broad, high-gain bandwidth, adequately covering the frequency ranges of 2G (1.71-1.92 GHz), 3G (1.88-2.17 GHz), and LTE (2.3-2.4 GHz and 2.5-2.69 GHz) networks. Figure 11 shows the simulated realized gains of the proposed antenna with different numbers (m × m) of AMC units. With reference to Figure 11, when the number of elements is 11 × 11, the proposed antenna has the highest average gain across the bandwidth of 1.71-2.69 GHz. Therefore, the number 11 × 11 is chosen for AMC design.  Figure 12 shows the comparison of simulated radiation patterns between the crosseddipole antenna with/without four parasitic strips. With reference to the figure, significant improvements in both E-and H-plane radiation patterns can be obtained by the loading AMC is employed to replace the traditional metal ground, thereby increasing the antenna gain and maintaining a low profile. Therefore, it becomes crucial to select an appropriate quantity of AMC units. This strategic selection aims to ensure that the antenna exhibits a broad, high-gain bandwidth, adequately covering the frequency ranges of 2G (1.71-1.92 GHz), 3G (1.88-2.17 GHz), and LTE (2.3-2.4 GHz and 2.5-2.69 GHz) networks. Figure 11 shows the simulated realized gains of the proposed antenna with different numbers (m × m) of AMC units. With reference to Figure 11, when the number of elements is 11 × 11, the proposed antenna has the highest average gain across the bandwidth of 1.71-2.69 GHz. Therefore, the number 11 × 11 is chosen for AMC design. AMC is employed to replace the traditional metal ground, thereby increasing the antenna gain and maintaining a low profile. Therefore, it becomes crucial to select an appropriate quantity of AMC units. This strategic selection aims to ensure that the antenna exhibits a broad, high-gain bandwidth, adequately covering the frequency ranges of 2G (1.71-1.92 GHz), 3G (1.88-2.17 GHz), and LTE (2.3-2.4 GHz and 2.5-2.69 GHz) networks. Figure 11 shows the simulated realized gains of the proposed antenna with different numbers (m × m) of AMC units. With reference to Figure 11, when the number of elements is 11 × 11, the proposed antenna has the highest average gain across the bandwidth of 1.71-2.69 GHz. Therefore, the number 11 × 11 is chosen for AMC design.  Figure 12 shows the comparison of simulated radiation patterns between the crosseddipole antenna with/without four parasitic strips. With reference to the figure, significant improvements in both E-and H-plane radiation patterns can be obtained by the loading Figure 11. Simulated realized gain for proposed antenna under different numbers (m × m) of AMC units. The blue region is the commercial bands for 2G, 3G, and LTE. Figure 12 shows the comparison of simulated radiation patterns between the crosseddipole antenna with/without four parasitic strips. With reference to the figure, significant improvements in both Eand H-plane radiation patterns can be obtained by the loading strips. It is worth mentioning that, using this method, the cross-polar fields in the two principal cutting planes decrease by more than 12 dB. strips. It is worth mentioning that, using this method, the cross-polar fields in the two principal cutting planes decrease by more than 12 dB.

Results
To verify the idea, a prototype of the proposed antenna was fabricated, as shown in Figure 13. Figure 14 shows the measured and simulated S-parameter of our antenna. With reference to Figure 14a, the measured and simulated −10-dB impedance bandwidths (|S11| ≤ −10 dB) are 54.05% (1.62-2.82 GHz) and 55.37% (1.58-2.79 GHz), respectively. As shown in Figure 14b, the isolations in both the simulation and measurement are almost greater than 30 dB across the impedance-matching passband. As can be observed from Figure 14, reasonable agreement between the measured and simulated results is observed. Acceptable errors are caused by slight deformation due to uneven structural support. The designed antenna exhibits good broadband characteristics and high isolation.

Results
To verify the idea, a prototype of the proposed antenna was fabricated, as shown in Figure 13. Figure 14 shows the measured and simulated S-parameter of our antenna. With reference to Figure 14a, the measured and simulated −10-dB impedance bandwidths (|S11| ≤ −10 dB) are 54.05% (1.62-2.82 GHz) and 55.37% (1.58-2.79 GHz), respectively. As shown in Figure 14b, the isolations in both the simulation and measurement are almost greater than 30 dB across the impedance-matching passband. As can be observed from Figure 14, reasonable agreement between the measured and simulated results is observed. Acceptable errors are caused by slight deformation due to uneven structural support. The designed antenna exhibits good broadband characteristics and high isolation. strips. It is worth mentioning that, using this method, the cross-polar fields in the two principal cutting planes decrease by more than 12 dB.

Results
To verify the idea, a prototype of the proposed antenna was fabricated, as shown in Figure 13. Figure 14 shows the measured and simulated S-parameter of our antenna. With reference to Figure 14a, the measured and simulated −10-dB impedance bandwidths (|S11| ≤ −10 dB) are 54.05% (1.62-2.82 GHz) and 55.37% (1.58-2.79 GHz), respectively. As shown in Figure 14b, the isolations in both the simulation and measurement are almost greater than 30 dB across the impedance-matching passband. As can be observed from Figure 14, reasonable agreement between the measured and simulated results is observed. Acceptable errors are caused by slight deformation due to uneven structural support. The designed antenna exhibits good broadband characteristics and high isolation.    Figure 15a. Figure 15b shows the measured antenna efficiency of the low-profile crossed-dipole antenna. As can be observed from Figure 15b, the measured antenna efficiency is higher than 90% across the entire operating band.   Figure 16 shows the measured and simulated normalized radiation patterns in two principal cutting planes. Reasonable agreement can be found between simulations and measurements. As shown in the figure, stable radiation patterns can be found at three different frequencies. The simulated cross-polar field is almost lower than −30 dB. It can be observed from the figure that the co-polar field is 10 dB higher than the cross-polar field over a wide range of angles (−60° < θ < 60°) in the measurements. Results of Port 1 are only shown due to the geometric symmetry of our crossed-dipole antenna.    Figure 16 shows the measured and simulated normalized radiation patterns in two principal cutting planes. Reasonable agreement can be found between simulations and measurements. As shown in the figure, stable radiation patterns can be found at three different frequencies. The simulated cross-polar field is almost lower than −30 dB. It can be observed from the figure that the co-polar field is 10 dB higher than the cross-polar field over a wide range of angles (−60° < θ < 60°) in the measurements. Results of Port 1 are only shown due to the geometric symmetry of our crossed-dipole antenna.  Figure 16 shows the measured and simulated normalized radiation patterns in two principal cutting planes. Reasonable agreement can be found between simulations and measurements. As shown in the figure, stable radiation patterns can be found at three different frequencies. The simulated cross-polar field is almost lower than −30 dB. It can be observed from the figure that the co-polar field is 10 dB higher than the cross-polar field over a wide range of angles (−60 • < θ < 60 • ) in the measurements. Results of Port 1 are only shown due to the geometric symmetry of our crossed-dipole antenna. Table 1 compares our antenna with the reported crossed-dipole antennas. With reference to the table, the antennas in [8,9] have an extremely low antenna height but a narrow bandwidth and low gain. The designs in references [11][12][13][14] have a higher profile. The antennas in references [14,21] have similar profiles, but their antenna gains are lower than the proposed antenna. Our wideband dual-polarized antenna has the highest antenna efficiency due to the proposed AMC. Compared with listed base-station antennas, the proposed antenna outperforms in overall performance. It is a good candidate for base-station applications.

Conclusions
A wideband low-profile crossed-dipole antenna with a novel AMC has been investigated. By introducing the AMC as antenna ground, the profile can be effectively decreased to 0.12λ 0 . In addition, four parasitic metal strips have been introduced to reduce the crosspolar field and, thus, increase the antenna gain in its high-frequency impedance passband. To verify the idea, a prototype has been simulated, fabricated, and measured. A measured overlapping bandwidth of 55.4% (1.58-2.79 GHz) has been obtained. Its maximum measured realized gain is 10.8 dB at 2.6 GHz. Additionally, a stable gain and radiation pattern have been observed. The prototype has a measured isolation of~30 dB. It should be mentioned that its cross-polarization level is more than −10 dB over a wide range of angles (−60 • < θ < 60 • ). Finally, these advantages enable the proposed antenna to be potentially applied to the integrated design of 2G/3G/LTE base station antennas.