Abstract
A system of two dual-band dual-polarized antennas is proposed. It operates in two bands, 700 to 862 MHz and 2.5 to 2.69 GHz, thereby making it suitable for LTE applications. The design is composed of two compact orthogonal monopoles printed close to each other to perform diversity in mobile terminals such as tablets or laptops. For each band, two orthogonal polarizations are available and an isolation higher than 15 dB is achieved between the two monopoles spaced by λ0/10 (where λ0 the central wavelength in free space of the lower band). A good agreement is observed between simulated and experimental results. The antenna diversity capability is highlighted with the calculation of envelope correlation and mean effective gain for several antennas' positions in different environment scenarios.
1. Introduction
Deployment of existing and emerging wireless communication systems require a high-data-rate transmission, in order to satisfy the needs of multimedia applications on terminals. Multiple Input Multiple Output (MIMO) applications have been suggested as an effective way to increase the channel capacity by exploiting multipath scattering effects.
MIMO technology is present in many recent wireless standards, such as Long Term Evolution (LTE), and will be implemented in mobile devices [1]. Several research works have proven the efficiency of two-antenna diversity on mobile terminals [2, 3]. However, when the available space is limited, the use of a dual-polarized antenna is more suitable than two separated antennas [4]. A variety of dual-polarized antennas have been reported recently in which good dual-polarized radiation over a wide bandwidth [5] and high isolation between the feeding ports [6] have been achieved. However, these antennas are mainly designed for single-band operation [7] or for frequencies above 800 MHz [8]. Most of the dual-band dual-polarized antennas proposed in literature exploit harmonics frequencies [9] or use techniques to generate additional resonances such as insertion slot [10]. But generally, it leads to a ratio between frequency bands below or equal to 2 and implies a dependence between the two frequency bands. Today, very few designs are reported for dual-band dual-polarized operations for the following bands: 700–862 MHz and 2.5–2.69 GHz. In this paper, we firstly present the design of a dual-band antenna which can provide a dual-polarization for each band, for LTE devices such as a tablet or a laptop. Then, we introduce an enhanced design in which the lower bandwidth has been increased and the mutual coupling between ports has been reduced in the two bands. The lower band is extended towards TV White Space (TVWS) band to provide radio-cognitive capabilities to the terminal [11].
Finally, the diversity performances of the proposed dual-band dual-polarized antennas are evaluated through the envelope correlation and the mean effective gain in isotropic, indoor, and outdoor environments.
2. Antenna Design
As shown in Figure 1, the proposed structure is composed of two orthogonal monopoles with dimensions of mm2. The two monopoles are identical and chosen for their omni-directional radiations pattern, enabling them to receive signals whatever their orientation. They are printed on a mm2 low cost substrate (FR4: , , thickness of 0.7 mm). Each monopole is connected to two bend endings: one bend ending is a meander line whose length is mm operating at 790–862 MHz and the small one whose length is mm operates at 2.5–2.69 GHz. The distance between the two bend endings is mm. This design allows to obtain independent frequency bands. The two monopoles are spaced by mm which corresponds to for the lower band and λ02/3 for the higher band, where is the free-space wavelength of the lower band central frequency ( MHz) and the free-space wavelength of the higher band central frequency ( GHz).
(a)
(b)
The monopoles are fed by two 50 ohms coplanar waveguides (CPW), directly etched in the ground plane, as shown in Figure 1(b), in order to distance the connectors and to avoid perturbations on the measured radiation patterns. Each CPW has a line width of 1.8 mm and a gap of 0.33 mm with the ground plane. Monopoles are connected to CPW thanks to metallic via holes located as the extremity of each monopole as shown in Figure 1(a).
2.1. Ground Plane Geometry
Since the small bend ending is close to the ground plane extremities, it is sensitive to the path taken by currents along the ground plane. Therefore, a study of the upper part of the ground plane geometry is relevant. It is found that removing corners (shaded part in Figure 1(b)) provides an improvement of higher band matching, leading to optimized dimensions mm, mm, and mm.
Coupling between the two antennas occurs via currents flowing from one antenna to the other one through the ground plane. It can be reduced by altering the ground plane to modify currents’ path. Thus, the ground plane is extended without increasing the overall structure size, by adding on the lower part of the substrate, two rectangular shapes on either side with dimensions of each one mm and mm (framed part in Figure 1(b)). Simulations have been performed with Transient Solver of CST Microwave Studio. Figure 2 shows a comparison between parameters for designs without slot, with and without added ground plane in each band. Because of the structure’s symmetry, only |S11| and |S21| are plotted. The matching bandwidth criterion is taken for a return loss less than −10 dB. With added ground plane, a shift of the lower band towards lower frequencies (from 0.9 to 0.85 GHz) is observed in Figure 2(a) without increasing the structure size. The bandwidths of the structure without added ground plane are: 837–957 MHz (13.4%), 2.35–2.86 GHz (19.6%), and for the structure with added ground plane are: 796–914 MHz (13.8%), 2.38–2.78 GHz (15.5%). Regarding the isolation, it is largely reduced thanks to the added ground plane: |S21| becomes below −20 dB in the lower band. Indeed, a resonance has been introduced at the frequency where coupling occurs. However, the coupling remains high (|S21| < −7 dB) in the higher band as shown in Figure 2(b).
(a)
(b)
To improve isolation between ports in the higher band, a slot is etched in the ground plane while keeping the same distance between ports (d), as shown in Figure 1(a). The introduction of the slot produces an open circuit which stops the circulation of current from one radiating element to the other one [12]. The optimized structure has a length mm and a width mm. Figure 3 shows a comparison of simulated parameters of one meander bend ending antennas with added ground plane, with and without slot in the higher band. The introduction of the slot achieves an isolation improvement of 10 dB in the higher band, while it has no effect in the lower band. The bandwidth is slightly reduced but still covers the desired band. Thus, optimization of the two degrees of freedom which are the slot dimensions and rectangular shapes ground plane dimensions leads to a high isolation in the two frequency bands.
2.2. Radiating Element
In order to increase the bandwidth of the lower band towards the TVWS band, two bend endings are added below the initial meander line to provide additional resonances close to each other. These two meanders are out of sync to provide a single wide band. Moreover, the three lines are connected to each other to extend the bandwidth towards lower frequencies. After optimization with Transient Solver of CST Microwave Studio, the distance between each meander is mm as shown in Figure 5 and the overall size of three bend endings antennas with added ground plane and slot becomes 150 × 90 mm2.
Figure 4 shows the comparison between S-parameters of one and three bend endings antennas with added ground plane and slot. Matching bandwidth criterion is taken for |S11| < −10 dB. It is seen that the bandwidth is enhanced towards lower frequencies. Indeed, the relative bandwidth for the structure with one bend ending is 9.8% (786–867 MHz) and 21.9% (692–862 MHz) for the structure with 3 bend endings. While keeping almost the same electrical length of the structure, the relative bandwidth has been improved by 12%. Indeed, the overall size is for three meander bend ending antennas (: the free space wavelength at 692 MHz) when it is 0.37 × 0.22 for one meander bend ending antennas (: the free space wavelength at 786 MHz).
3. Prototype and Measurement
A prototype of three bend endings antennas with added ground plane and slot described previously has been realized. Monopoles and the ground plane with CPW are located on opposite sides of the same substrate and can be seen simultaneously on Figure 5 because of the transparency of the FR4 substrate. Simulated and measured S-parameters are compared in Figure 6. Simulations results are in good agreement with measurement. The measurement results show that the antenna operates in two bands (|S11| < −10 dB): the lower band extends from 700–880 MHz (21.9%) and the higher one from 2.51–2.72 GHz (8%). In these two bands, the two monopoles are satisfactorily uncoupled with an isolation |S21| below −15 dB within the higher band and from 770 to 880 MHz. At the beginning of the lower band, the isolation remains acceptable and is below −10 dB. The simulated total efficiency of the structure, which takes into account all losses, has been evaluated: it varies from 83 to 97% in the lower band and from 74 to 87% in the higher band as shown in Figures 7 and 8.
Figure 9 compares the simulated and measured copolar and cross-polar radiation patterns in the E plane (YZ plane) and H plane (XZ plane), respectively. Because both ports are symmetrical, we only represent radiation patterns for port number 1 while port 2 is loaded by 50 ohms. For both planes and both bands, it is found that the simulated and the measured co-polar radiation patterns are in good agreement. The maximum simulated realized gain is 2.5 dB at 778 MHz and 5 dB at 2.6 GHz. The measured cross-polar level is about 10 dB lower than the copolar level in the lower band but in the higher one, the polarization purity is deteriorated. It is probably due to the proximity of the meander bend endings to the small one.
(a)
(b)
(c)
(d)
To further investigate the diversity, the simulated radiation patterns of each radiating element in the XY plane for the two bands are plotted in Figure 10 (one port is excited while the other one is loaded by 50 ohms). Thanks to a good agreement observed in Figure 9 between simulations and measurement, only simulations results are presented. As it can be observed for the lower band, the directions of the pattern maxima are close to orthogonal, leading to good pattern diversity. Each antenna presents monopole-like radiation patterns. Indeed surface currents are weak on the bend endings. For the higher band, even if patterns are not orthogonal, one monopole presents minimum gain directions where the other one has a maximum gain, except for the directions . This is well-suited to provide high diversity capabilities.
(a)
(b)
4. Evaluation of the Diversity Performance
The diversity performance of a mobile’s antenna system can be affected by the environment in which the device is located [13]. Therefore, in this section, we evaluate the diversity performance of the proposed three bend endings antennas with added ground plane and slot, by calculating the envelope correlation coefficient () and the mean effective gain (MEG) taking into account the propagation environment.
The envelope correlation quantifies the similarity between the radiation patterns of the two monopoles. The lower the correlation, the better the diversity performance. Vaughan and Andersen show in [13] that the coefficient can be expressed by
, , , are simulated complex electric fields along θ and φ radiated by the antenna fed by two different ports. The solid angle Ω is defined by in elevation and in azimuth. and are the Angle-of-Arrival (AoA) distributions of incoming waves. The parameter XPD is the cross-polarization discrimination of the incident field and is defined as (where and represent the average power along the spherical coordinates θ and φ).
The environment depends strongly on the angles of arrival distribution and on XPD. The most common distributions proven by measurements are Gaussian (G) and Laplacian (L) distributions [14]. Thus, we consider different distributions in elevation, while in azimuth plane (XY plane) the distribution is uniform, as demonstrated by two measurement campaigns in the literature [14, 15].
To obtain more realistic results, different environments are considered. Each environment is characterized by typical values of XPD, mean angle of incident wave distribution , and standard deviation of wave distribution (σ) [16]. These values were deduced from several measurements [14–16] for different environments: isotropic, indoor, and outdoor. The isotropic environment is defined by dB, , the indoor (In) environment by dB, , , and the outdoor (Out) environment by dB, , .
As antennas will be implemented on a mobile terminal, a study of the effect of the antennas orientation on the correlation has been done. Three configurations of rotations have been studied: rotation of antenna around axis A, and around axis B for two initial positions: horizontal and vertical, as shown in Figure 11.
(a)
(b)
(c)
For each configuration, the envelope correlation coefficient for the three meander bend endings antennas with added ground plane and slot has been calculated from simulated radiation patterns. Minimum and maximum values at center frequencies of the two bands 777 MHz and 2.6 GHz are reported in Table 1.
For isotropic environment, a very low correlation is observed, in the two bands as a result of good matching (|S11| < −10 dB), a high isolation level (|S21| < −10 dB), and orthogonality between radiation patterns especially in the lower band. In addition, polarization diversity is naturally achieved because of the orthogonal positions of both antennas.
For the other cases, maximum values of the correlation envelope coefficient are close to 0.5 for outdoor environment, whatever the distribution. Indeed, the incoming waves are mainly along which implies less diversity in some antenna’s position.
When XPD gets close to 0 dB (indoor environment: dB), and values are almost the same. Because these two components are uncorrelated by definition and because each antenna receives preferentially one of each component, the correlation is getting low.
For rotation around axis A, minimum values of are obtained for position at which one antenna receives only component of the incoming waves while the other one only component.
For rotation around axis B, for both configurations (b and c), minimum values are obtained when the two radiating elements are positioned on AB plane. Indeed, at these positions, the radiation diversity is exploited as shown in Figure 10, and thus a low correlation is obtained.
Finally, for most configurations, envelope correlation coefficient is less than 0.5 which provides high diversity capabilities [13]. This result has been achieved thanks to the two orthogonal and identical antennas which are spatially separated. It can provide for either or both spatial and pattern diversity. In addition, polarization diversity is available in the Z-direction.
In the following part, we evaluate the MEG which was introduced by Taga [17]. It is defined as the ratio between the mean received power of antennas over the random route and the total mean incident power. When each monopole receives the same quantity of power, the MEG ratio (R) of the two antennas is equal to one, which means that no performance deterioration is expected due to some power imbalance [18].
The mathematical expression is given by the following equation: where and are the θ and φ components of the antenna power gain pattern, respectively. The calculated mean effective gains of the monopoles from simulated radiation patterns at 777 MHz and 2.6 GHz are presented in Table 2.
The Maximum values of the ratio (R) of MEG1, determined at port 1, over MEG2, determined at port 2, are equal to 1, which satisfy an equal contribution of the two monopoles to receive the same quantity of power. The proposed structure is completely symmetric, and the Gaussian and Laplacian angular distributions are taken only along the elevation as presented in [15]. In addition, the incident power in the outdoor environment (or indoor) is concentrated around 10° (or 20°) above the horizon with an aperture of 30° (or 60°), and for these directions both antennas receive an equal amount of power.
Minimum values of ratio (R) are obtained for positions at which the (or ) components of the two antennas have different levels in the directions of incident power. For example, if antenna 1 presents a low component where antenna 2 a high one, an unbalanced power is obtained.
For most configurations, ratio (R) is greater than 0.5 which is acceptable to provide high diversity capabilities [18].
5. Conclusion
In this paper, a compact dual-band, dual-polarized antenna for LTE applications is proposed, with an extension of the lower band towards TV White Space band, to provide radio-cognitive capabilities to the terminal. A design provides dual polarizations in both of the bands: 700–862 MHz and 2.5–2.69 GHz with good impedance matching (|S11|< −10 dB).
Measurement results are in good agreement with simulated ones. In addition, good performances are obtained by calculating the envelope correlation coefficient and the MEG ratio for several antennas’ positions in different environments: isotropic, indoor, and outdoor. For most configurations, it is found that the system satisfies the condition and . Thus, the presented design is suitable for MIMO communication applications, and thus enables the SNR value at the terminal side to be maximized.
Acknowledgment
The research leading to these results has received funding from the European Community’s Seventh Framework Program (FP7/2007–2013) under Grant agreement SACRA no. 249060.