As shown in Fig. 1, the reference antenna is an 8 × 10 asymmetrical ridged waveguide slot antenna array (WGSAA), which is widely used in radar and communication systems due to its simple structure, good mechanical strength, high power capacity, and low side-lobe.^[21] The reference antenna operates at 3.20 GHz with a distance between adjacent slots along the x-axis of 0.494λ_3.2GHz. Apparently, the WGSAA is subjected to strong structural scattering due to its large planar metallic ground.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Schematic diagram of (a) bare WGSAA and (b) M × N MS array.

The working principle of ultra-wideband RCSR that isproposed in this paper is the backscattering cancellation, which further depends on the phase difference among different AMC tiles. Unlike the implementation in traditional chessboard configuration, the phase differences in this paper are created by three different AMC tiles resonant at different frequencies. Since the phase difference no longer relies on resonance, effective energy cancellation can be obtained over an ultra-wideband. To gain an insight into the RCSR mechanism, we choose a general MS array consisting of M × N elements as depicted in Fig. 2. When a plane wave normally impinges on such an array, the total scattering field is the superposition of the reflected energy from all M × N elements. Assuming that each of all elements presents an equal reflection pattern, on the basis of standard array theory,^[4] then the total reflection can be determined by 1 where P is the element pattern, approximate to cosθ, F_A is the array factor, k is the wavenumber, Δx and Δy are the distances between adjacent elements along x- and y-directions, respectively, ϕ(m,n) is the phase of (m, n) element, and θ, φ are the elevation and azimuth angles of an incidence.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) (a) Geometry and dimensions of AMC1, AMC2, and AMC3. (b) Schematic view of quadruple-triangle arrangement. P = 9 mm, a = 1.4 mm, a1 = 2.45 mm, b = 0.6 mm, b1 = 2.4 mm, b2 = 0.86 mm, W = 0.4 mm, W1 = 0.7 mm, W2 = 1 mm, W3 = 1 mm, L1 = 7.6 mm, L2 = 2.44 mm, and L3 = 1.1 mm.

For normal incidence with (θ, φ) = (0, 0), once the reflection phase difference between adjacent elements is ±180°, the reflection can be totally canceled out. However, since the reflection phase varies with frequency, the 180° phase difference cannot be maintained over an ultra-wideband. Usually, a 10-dB RCSR is set to be a criterion to compare with the same-sized PEC surface, that is, 2 Hence, the effective reflection phase difference is deduced to be 3

Likewise, we set 180° ± 30° as a criterion for the following analysis. Note that the backscattering cancellation is dependent on a dynamic variation of phase difference instead of a fixed value, and the RCSR bandwidth is consequently expected to expand to a large margin.

In 2007, Paguay et al.^[3] first combined PEC and AMC in a chessboard arrangement to redistribute the reflected energy into four diagonal directions based on scattering cancellation. To broaden the RCSR bandwidth, two different AMC units were utilized in a chessboard configuration by Zhao et al.^[8] However, the RCSR bandwidth was still limited. In 2014, Cui et al. proposed the concept “coding metamaterials”.^[22] When combined with antenna array, the restricted space between array elements cannot accommodate so many kinds of basic units. Consequently, the closely-spaced WGSAA resorts to an alternative method to address the problem of ultra-wideband RCSR.

Here, we choose three different AMC tiles resonate at different frequencies to achieve ultra-wideband RCSR. Each single sub-unit of the three different AMC tiles with detailed dimensions is depicted in Fig. 2(a), denoted as AMC1, AMC2, and AMC3, respectively. Each of the three sub-units is composed of two metallic layers separated by an exactly same-sized substrate with a dielectric constant of 2.65 and a loss tangent of 0.002. Moreover, despite of the top metallic layers with different shapes, all of the three sub-units are each backed by a full metallic ground to ensure no plane wave penetration. To create destructive interference in the largest bandwidth, full-wave numerical simulations have been carried out to optimize the parameters by using commercial simulation software Ansys HFSS. Floquet port excitations and master/slave boundaries are performed on each single sub-unit to simulate the infinite two-dimensional periodic boundary condition.

Figure 3 shows the reflection magnitudes and phases under the x- and y-polarized incidence for each sub-unit. As can clearly be observed, the magnitudes maintain 0.98 from 2 GHz to 21.5 GHz for both polarizations, indicating that the energy is almost reflected without absorption. The simulated results meet the requirement for phase cancellation principle. One can observe that AMC1 exhibits only one 0° reflection phase point at 10.9 GHz, while AMC2 and AMC3 demonstrate dual 0° reflection phase points at 7.15 GHz, 18 GHz and 3.98 GHz, 19.18 GHz, respectively. The different resonant statuses yield the destructive phase difference ranging over an ultra-wide band as depicted in Fig. 4. For normal incidence, the phase difference covering a range of 180° ± 30° between every two of the three sub-units nearly ranges from 3.98 GHz to 18.84 GHz except from 4.2 GHz to 6.9 GHz.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) Reflection characteristics of AMC1, AMC2, and AMC3 under (a) x-polarized incidence and (b) y-polarized incidence.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Reflection phase differences between every two of the three AMC units under (a) x-polarized incidence and (b) y-polarized incidence.

Considering their application in the WGSAA, the three different AMC sub-units are arranged in a quadruple-triangle-type chessboard configuration to adapt itself to the limited room between array elements. To balance the RCSR performance in the low and high frequency bands and maximize the RCSR performance, on the basis of the simulated reflection phase difference in Fig. 4, the AMC1 is chosen as a mediator, which accounts for two-quarters of the quadruple-triangle configuration. In this way, the AMC1 tiles can be always surrounded by AMC2 and AMC3 tiles to yield required phase difference in frequency band as wide as possible as depicted in Fig. 2(b). Moreover, the proposed arrangement can significantly diminish EM echo in the specular direction and redistribute backscattered energy more uniformly in the rear hemisphere space to the benefit of the bi-static detection case. It is worth pointing out that the three AMC tiles can be arranged in other alternative configurations.

To verify the ultra-wideband low RCS capability of the proposed design, the three AMC tiles are arranged in a quadruple-triangle-type configuration to form a finite composite planar MS array with a size of 432 mm × 432 mm × 3 mm. Each AMC tile consists of sixteen optimized sub-units in a triangle-type arrangement. The simulated mono-static RCSR under normal incidence is normalized by a same-sized PEC surface as illustrated in Fig. 5. An ultra-wideband RCSR over –6 dB is observed from 4.52 GHz to 4.86 GHz (7.2% bandwidth) and from 5.9 GHz to 19.9 GHz (108.5% bandwidth). Moreover, a –3 dB RCSR is obtained nearly from 4.4 GHz to 22 GHz with a relative bandwidth of 133.3% for both x- and y-polarized incidences. The consistency between mono-static RCSR performances under the x-incidence and under the y-polarized incidence is attributed to the symmetric layout. The three-dimensional (3D) scattering patterns under normal incidence at 4.7 GHz, 10.4 GHz, and 16.8 GHz are also demonstrated in the inset. It is clearly shown that the scattered energy is redistributed into several lobes, forming a low scattering space at normal direction as a consequence. The simulated results of the finite MS array verify the capability and potentiality of our design method for ultra-wideband RCSR and bi-static RCSR to some extent.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Monostatic RCSR versus frequency of the composite planar MS array normalized by the same-sized PEC surface. Inset: 3D scattering patterns under normal incidence at 4.7 GHz, 10.4 GHz, and 16.8 GHz.

After elaborately designed and judiciously arranging the three AMC tiles, we exploit its application in WGSAA performances. In contrast to common objects, effective radiation must be assured to be a necessary condition in terms of RCSR for antenna. Therefore, we seek to maintain or even improve the radiation performance first and foremost after employing the MS. Considering the limited space to accommodate the MS, each AMC tile contains four sub-units when applied to the WGSAA. The fabricated prototype is shown in Fig. 6 with seven identical MS bars fabricated and mounted on the bare WGSAA. Each bar consists of 4 × 68 sub-units with a size of 36 mm × 615 mm. The vector network analyzer (VNA) Agilent N5230C is utilized to measure the reflection coefficient. With respect to bare WGSAA, the measured S₁₁ of the MS-based antenna array is 3.165–3.297 GHz, 24 MHz broader than that of bare WGSAA, as depicted in Fig. 7. The radiation patterns of the two antennas at 3.2 GHz (the best impedance matching at this frequency for both configurations) in Fig. 8

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Photograph of (a) fabricated WGSAA loaded with MS and (b) measurement setup.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. (color online) Comparisons between measured and simulated S11 parameters.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. (color online) Comparisons between measured and simulated radiation patterns in (a) E-plane (yoz-plane) and (b) H-plane (xoz-plane).

show that the loading of MS makes antenna gain increased by 0.77 dBi and 0.29 dBi in E-plane and H-plane, respectively. The measured results agree well with the simulation’s results, which further verifies the loading of MS has little influence on antenna radiation performances.

To experimentally evaluate the application of MS in scattering performance, the fabricated antenna array is measured in a microwave anechoic chamber as shown in Fig. 6(b). Two identical linear-polarized pyramidal horn antennas working at 1–18 GHz are placed adjacently as transmitter and receiver, respectively. The centers of the prototype and the two horns are at the same height and the distance between them is far enough to satisfy the far-field condition. The gate-reflect-line calibration in time-domain analysis kit of VNA is employed to further eliminate undesirable signals. The VNA Agilent N5230C is utilized to evaluate mono-static RCS value of transmission coefficients. Owing to the restriction of experimental equipment, the measured mono-static RCSR is only demonstrated in a range from 3 GHz to 18 GHz.

As shown in Fig. 9, a continuous 6-dB RCSR is achieved from 10.12 GHz to 18 GHz and from 7.2 GHz to 18 GHz for cross- (E-field along y-axis) and co-polarized (E-field along x-axis) incidences, respectively. Meanwhile, in-band RCSR is achieved for co-polarized incidence with maximum RCSR reaching up to –5.8 dB. This is believed to be due to the co-polarized incidence inducing the slot voltage, and then leading to RCSR based on phase cancellation. Taking fabrication and measurement errors into consideration, reasonable agreement between the measured and simulated results can be found, which confirms the effectiveness of applying MS to WGSAA for ultra-wideband RCSR.

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. (color online) Measured and simulated RCSR in comparison with bare WGSAA for (a) cross-polarization and (b) co-polarization.