**4.6. Section II: The tapered transition**

166 Ultra Wideband – Current Status and Future Trends

distorted till 50GHz as shown in Figure 8.

**4.5. Section I: The feed section** 

discussed in details.

summarized as follows:

**Prototype 4**: is obtained by the combined optimization of the FSD-identified parameters WT and WC, By keeping the optimized parameters of the previous optimization steps (LT of prototype 3, all others of prototype 1) fixed, and doing 2 SVO sequences, we obtain the SWB impedance bandwidth of BR greater than 30:1, the result is plotted in Figure 9, the corresponding design parameters are listed in table 2. We noted here that by controlling radii-separation distance WC we are able to keep the radiation pattern of this prototype less

The created parameters, their functions, their effects and their usages are discussed in greater details in the next sub-sections. The FSD is detailed in §4.5-4.7, the resulted SWBperformances are given in §4.8 and in §4.9, the optimizations of all the prototypes are

Planar antennas and arrays have been used for micro-wave and millimeter-wave applications for decades, especially in mobile communications where system design requires low profile, lightweight, and high directivity. The two most used feeding methods are micro-strip line (MS) and coplanar waveguide (CPW), they both carried signal excellently in narrow-band and UWB antennas and devices. Many planar antenna arrays have been designed by using MS, however, until recently, only a few works so far have used CPW to feed the array. The CPW has gained increasing popularity in recent years, since it has several advantages over the MS, such as low radiation losses, less dispersion, easier integration with solid-state active devices, and the possibility of connecting series and shunt elements, and suitable for SMD-technology, also for SWB- antennas/devices CPW feeding

In search for the SWB radiator, both radiator and the feed must be super wide band. Since the SWB-signal first must able to pass through the feeding-line before reaching the antenna, obviously that the feed must be considered first in advance of other sections, we conduct the work with bottom-up approach, i.e., the feed is consider first, because if the feeding mechanism fail to be SWB, then there is no SWB radiator exists no matter how good the radiator will be. The coplanar waveguide is the first choice for feeding the signal to the radiator, because the CPW's effective dielectric is constant, (this property is a key feature in wide band matching the antenna), over a wider BW than micro-strip line, another advantage is, in contrast with MS line, one of the parameter pair ( WS, WG) can be varied in size and shape, whilst the other is correspondingly changes to keep the characteristic impedance stays unchanged, furthermore CPW is low-loss, and the signal width can be chosen width enough to support characteristic impedance from 30Ohm and higher (Simons, 2001, p.52),

The CPW would be a better choice for SWB-feeding because of it considered features,

• **SWB behavior**: the effective dielectric constant is almost independent of frequency (Simon op cit.), this feature is a priori condition for SWB feeding and matching.

provides better match and performs better than the MS line (Simons, 2001).

The tapered transition has been inserted between the CPW-feed and the radiating patch, this section responsible for a smooth transition between the feed and the antenna, and because the current distribution is denser in this region than the others, this property indicated that this section must have strongest influence in matching the impedance bandwidth. This section has two parameters (LT, WT), which are described in details in §4.6.1 and §4.6.2 below.

### *4.6.1. The internal Transition Length LT*

The length LT of the tapered transition (section II) is responsible for the smooth transition between the feed and antenna, and proved to be the most sensitive parameter in the design of our prototypes. Anticipation from the theory and design of micro-strip antenna (MSA), it is well-known that the length of the MSA determines the resonance frequency (by lengthen or shorten this parameter, one can shift the resonance frequency to lower or higher band, respectively).

The antenna's resonance is affected by its length, this length is composed by LT + LR1 + LR2, when this composed length is changed, and the resonance will presumably change accordingly.

It is observed, from the results plotted in Fig.9, that when LT is longer the resonance will shift to lower frequency (as shown by prototype 1, LT = 3.64, Fig.9), and when LT is shorter the radiator's resonance will shift to higher frequency (prototype 2, LT= 1.64, Fig.9)

## *4.6.2. The internal Transition Width WT*

The width WT of the taper transition section (section II), is also a "share-parameter" with lower radiating section (section IIIA); this width provides, as similar role as the width in microstrip patch antenna, a fine-tune mechanism for impedance matching as its nominal value varies. This enhanced matching mechanism are numerically demonstrated with the reflection coefficients of the prototypes 3 and 4 as plotted in Fig. 10, in which they shown a lower reflection coefficient, i.e., a better match.

#### **4.7. Section III: The radiation section**

This section comprise of six parameters LR1, LR2, WT, WR, WC, and RC, in which WT is the share parameter described in §4.6.2, we divided this section into sub-sections IIA and IIB

as sectioned in Fig. 5b. The parameter set of the internal sub-section (IIIA) and the external parameter sub-section (IIIB) are {WT, WR, LR2} and {WR, LR1, WC, RC}, respectively.

Architecture and Design Procedure of a Generic SWB Antenna with Superb Performances for Tactical Commands and Ubiquitous Communications 169

**Figure 7.** Impedance bandwidth performance of the proposed prototype 4.

we may have to look for other dielectric material.

patterns from 5 to 50 GHz with increment step of 5GHz.

computational bandwidth.

The numerical results were computed with the following practical assumptions:

• The relative dielectric constant and the loss are assumed constant over the

• The commercial Duroid RT5870 high frequency laminated material we used, in deed could practically support up to 77 GHz only (Huang, 2008, p.64); beyond this frequency

**Figure 8.** Pattern performance of prototype 4. Complete spectra of 3D-farfield co-polar radiation

Fig.8 shows the simulated 3D radiation patterns of prototypes 4; the plot indicated that the radiator exhibits a super wide band pattern characteristics, the usable spherical patterns are sustained in a bandwidth wider than 10:1. The SWB properties displayed in fig.7 and fig.8 shown that prototype 4 is a true SWB radiator in both impedance and patterns aspects, and its SWB-behavior is superior to antennas reviewed by (Massey, 2007, pp.163-196). We emphasize here that this radiator is termed as quasi-magnetic antenna, because it is clearly seen that this antenna possessed radiation patterns similar to that of a dipole, therefore calling it monopole reflects wrongly the EM-characteristics that it possesses and exhibits.

#### *4.7.1. The internal radiating matching section*

Two parameters, which identified to be key player for this subsection, are {WT, LR2} (the WR is not touched because it is share parameter of these two sub-sections), by first optimize the width parameter WT, and by keeping this optimized parameter fixed, and continuing to optimize the other parameter LR2. By this token, (Tanyer et al., 2009b, op cit.) obtained a huge enhancement in ratio bandwidth reported as 9:1. The design parameters are listed in table.1, the result is plotted in Fig.9. More detailed works and measurement results, the reader should refer to (Tanyer et al., op cit.).

### *4.7.2. The external radiating transition section*

This sub-section was often neglected by the designers due to the fact that the current distribution is weak along the edges of this section. However, we observed that it plays an important role in maintaining the shape of radiation pattern in a wide range of frequencies, as will be numerically proved in §4.8

The sub-section IIIB consists of a set of parameters {WR, LR1, WC, RC}; WR is a share parameter so we keep it intact. From fig.1 it is seen that WR1 is suppressed and covered by varying the radii-distance WC, we can also single LR1 out because it contribution to the length of the radiator can already be economized by LT and LR2, so WC is the only parameter left that we may use to fine-tune the radiator for both SWB performance and radiation pattern characteristics. Prototype 4 utilized this philosophy by varying LR2 (instead of LR1) and WC to obtain the super wideband performance plotted in Fig.7. It is observed that variation of WC had no significant impact on reflection coefficient (current distribution along the antenna circle edges are rather week compared with those close to the tapering transition region). Nevertheless, by properly controlling WC we are able to maintain the usable shape of the radiation patterns up to 50 GHz as shown in Fig.8. So, WC is clearly to be the parameter to control the interference of the edge/corner scattering and diffraction of the radiator. In transforming the vertex-diffraction to edge-diffraction, we advocate the use of circular shape; nevertheless, other researchers suggested the shapes (elliptical, football cape, etc.). To answer the question which shape would serve best, we need a further in depth study about all possible pattern sensitive shapes before providing a final conclusive appraisal.

#### **4.8. The prototype 4 and its super wide band performances**

Fig. 7 shows the simulated result of the magnitude of the reflection coefficient of our SWB prototype 4. We computed and shown here only up to 150GHz. By close inspection of the reflection coefficient, the reader could observe that the prototype 4 shows a trend and exhibits the behavior of an all-high-pass antenna; its impedance bandwidth could be much wider than shown here.

Architecture and Design Procedure of a Generic SWB Antenna with Superb Performances for Tactical Commands and Ubiquitous Communications 169

**Figure 7.** Impedance bandwidth performance of the proposed prototype 4.

168 Ultra Wideband – Current Status and Future Trends

*4.7.1. The internal radiating matching section* 

reader should refer to (Tanyer et al., op cit.).

as will be numerically proved in §4.8

final conclusive appraisal.

wider than shown here.

**4.8. The prototype 4 and its super wide band performances** 

*4.7.2. The external radiating transition section* 

as sectioned in Fig. 5b. The parameter set of the internal sub-section (IIIA) and the external parameter sub-section (IIIB) are {WT, WR, LR2} and {WR, LR1, WC, RC}, respectively.

Two parameters, which identified to be key player for this subsection, are {WT, LR2} (the WR is not touched because it is share parameter of these two sub-sections), by first optimize the width parameter WT, and by keeping this optimized parameter fixed, and continuing to optimize the other parameter LR2. By this token, (Tanyer et al., 2009b, op cit.) obtained a huge enhancement in ratio bandwidth reported as 9:1. The design parameters are listed in table.1, the result is plotted in Fig.9. More detailed works and measurement results, the

This sub-section was often neglected by the designers due to the fact that the current distribution is weak along the edges of this section. However, we observed that it plays an important role in maintaining the shape of radiation pattern in a wide range of frequencies,

The sub-section IIIB consists of a set of parameters {WR, LR1, WC, RC}; WR is a share parameter so we keep it intact. From fig.1 it is seen that WR1 is suppressed and covered by varying the radii-distance WC, we can also single LR1 out because it contribution to the length of the radiator can already be economized by LT and LR2, so WC is the only parameter left that we may use to fine-tune the radiator for both SWB performance and radiation pattern characteristics. Prototype 4 utilized this philosophy by varying LR2 (instead of LR1) and WC to obtain the super wideband performance plotted in Fig.7. It is observed that variation of WC had no significant impact on reflection coefficient (current distribution along the antenna circle edges are rather week compared with those close to the tapering transition region). Nevertheless, by properly controlling WC we are able to maintain the usable shape of the radiation patterns up to 50 GHz as shown in Fig.8. So, WC is clearly to be the parameter to control the interference of the edge/corner scattering and diffraction of the radiator. In transforming the vertex-diffraction to edge-diffraction, we advocate the use of circular shape; nevertheless, other researchers suggested the shapes (elliptical, football cape, etc.). To answer the question which shape would serve best, we need a further in depth study about all possible pattern sensitive shapes before providing a

Fig. 7 shows the simulated result of the magnitude of the reflection coefficient of our SWB prototype 4. We computed and shown here only up to 150GHz. By close inspection of the reflection coefficient, the reader could observe that the prototype 4 shows a trend and exhibits the behavior of an all-high-pass antenna; its impedance bandwidth could be much The numerical results were computed with the following practical assumptions:


**Figure 8.** Pattern performance of prototype 4. Complete spectra of 3D-farfield co-polar radiation patterns from 5 to 50 GHz with increment step of 5GHz.

Fig.8 shows the simulated 3D radiation patterns of prototypes 4; the plot indicated that the radiator exhibits a super wide band pattern characteristics, the usable spherical patterns are sustained in a bandwidth wider than 10:1. The SWB properties displayed in fig.7 and fig.8 shown that prototype 4 is a true SWB radiator in both impedance and patterns aspects, and its SWB-behavior is superior to antennas reviewed by (Massey, 2007, pp.163-196). We emphasize here that this radiator is termed as quasi-magnetic antenna, because it is clearly seen that this antenna possessed radiation patterns similar to that of a dipole, therefore calling it monopole reflects wrongly the EM-characteristics that it possesses and exhibits.

#### **4.9. Optimization process and development of prototypes**

Although the proposed topology and architecture is simple, however with a total of 14 parameters it would be an impossible task for the multivariate optimization process.

Architecture and Design Procedure of a Generic SWB Antenna with Superb Performances for Tactical Commands and Ubiquitous Communications 171

transformation from coax's TEM to quasi-TEM of the CPW line, so the feed's parameters

can ruled out for optimization, and set to be fixed with values as listed in table.1. **Step 6.** Start values: This step initiated the start values for the set of parameter listed in step 1. The initialization of the start values was elaborately detailed in (Tran et al., 2007). Prototype 1 (Fig.4) with the start values as listed in table.1, obtained a BW of 10GHz (fig.9). Detailed discussions and simulated, measured results can be found in (Tran et

**Step 7.** Simplified process: The 7th step is the simplification of the optimization process by breaking the MVO process down into series of SVO one. It was observed that the current intensity is mainly distributed along the edges of the transition region, this suggested that : a) small variation of the transition parameter in this section (LT, WT) would have a strong influence on the flow of the signal current and hence play a large impact on the impedance bandwidth, and exploiting the fact, as similarly in microstrip antenna, that b) the length of the radiator defines the shifting-property of the resonance, whilst c) the width of the radiator effectuates the matching-property, and d) another

So by keeping WT, the share-parameter between sections II and III.A (fig.5b), fixed and took just a single-variable-optimization (LT), the impedance bandwidth was double to 20GHz as obtained by prototype 2, which is plotted in fig.9. The parameters of prototype 2 are listed in table.2, more details of the electromagnetic properties and measurements regarding the performances of this prototype can be found in (Tanyer et

**Step 8.** BW enhancement: Prototype 3 (fig.4) was obtained by analyzing and optimization the lower part of radiation section IIIA (fig.1b). This section has three parameters (WT, LR2, WR); first, following the rules discussed in step 2 we left out the parameter WR since it is the share-parameter between section IIIA and IIIB; next, in order to avoid multivariate-optimization, anticipating the shifting-property and matching-properties discussed as (b) and (c), respectively, in step 7; then two SVO sweepings were carried out, first WT (property c) and then LR1(property b); the order of WT or LR1 can be chosen freely according to independent-property (d, in step 7). The design parameters of prototype 3 are listed in table.1 in which the parameters of the previous optimized prototype are kept fixed, only the two parameters (WT and LR1) belonged to section

**Step 9.** The design of prototype 4 is aimed at two SWB-compliances: 1) SWB impedance bandwidth, and 2) SWB radiation pattern. The radiation section consists of the following set of parameters (WT; LR2, WR; LR1, WC, R ), the length parameters of this set can be keep fixed, because LR2 and LR1 are resonance-shifting parameters, and the antenna architecture allows us to use other length of the antenna to control the resonance, this was already done by LT of the lower section, so these two parameters can be singled out of the optimization process; the share parameter WR can also be neglected because it matching-property is covered by the set {WC, R}, so by keeping R fixed the parameter set of section III was left with only two parameters left {WC, WT}. The procedure followed: fist, keeping the parameters of prototype 1 with the optimized

independent-property is that they can be separately optimized.

al., op. cit.).

al., op. cit.).

IIIA are investigated and optimized.

This section reports in details of how to delimit the variables and how to reduce the number of variables for simplifying the MVO process to a SVO one. In addition, some pragmatic rules are also given for identifying the key parameters, and for weighting the priority of those parameters in the sequentially SVO processes.

To keep the optimization process controllable en less entangled in multivariate-optimization process, efforts have been done in the following steps:


transformation from coax's TEM to quasi-TEM of the CPW line, so the feed's parameters can ruled out for optimization, and set to be fixed with values as listed in table.1.

**Step 6.** Start values: This step initiated the start values for the set of parameter listed in step 1. The initialization of the start values was elaborately detailed in (Tran et al., 2007). Prototype 1 (Fig.4) with the start values as listed in table.1, obtained a BW of 10GHz (fig.9). Detailed discussions and simulated, measured results can be found in (Tran et al., op. cit.).

170 Ultra Wideband – Current Status and Future Trends

**4.9. Optimization process and development of prototypes** 

those parameters in the sequentially SVO processes.

process, efforts have been done in the following steps:

priority on taking the share parameter.

in the optimization process.

Although the proposed topology and architecture is simple, however with a total of 14

This section reports in details of how to delimit the variables and how to reduce the number of variables for simplifying the MVO process to a SVO one. In addition, some pragmatic rules are also given for identifying the key parameters, and for weighting the priority of

To keep the optimization process controllable en less entangled in multivariate-optimization

**Step 1.** Topology and architecture: this step is important in that it had to form a basis for the FSD, and had to keep the antenna topology as simple as possible, but not simplest as (Al Sharkawy et al., 2004), Although all efforts has been carried out to ensure a minimum amount of the created parameters, the structure (fig.1a) still have a considerable set of parameters {εr, *h*, *t; LF, W*S*, W*GND*, W*G; *L*T*, W*T*; L*R2*, W*R*; L*R1*, W*C*, R* }. **Step 2.** The FSD: dividing the radiator into sections depending on their main function. Inspection of the radiator's current distribution and the radiator's topology shown in fig. 1a revealed that the radiator could presumably be divided into functional sections as depicted in fig. 1b. The analyzing and optimizing process are conducted following the bottom-up approach that always started from the feed (section I) and ended at the last radiation-section (section III.B). The analysis, optimization and development of the prototypes all should start and avoid as much as possible the *share-parameters* between sections (*W*T *, W*R); if impasse is met then, as a thumb-rule, the section on top has

**Step 3.** Excluding of fixed parameters: The numbers of parameters of the radiator have been reduced by singling out the non-optimizable parameters. The first three materialparameters {εr, *h*, *t*}, because their nominal values were already fixed by the manufacturers, are not quite suitable for optimization process as continuousparameters, so the parameters of the feed section (section I) are kept fixed and excluded

**Step 4.** Setting boundaries: in order to accelerate the optimization process and avoiding the problem of unbounded optimization, we put geometrical restrictions on the total width (2WGND + 2WG + WS) and length (LF + LT + LR2 + LR1) of the antennas fixed to λb/2, where λb is wavelength at the design-frequency, and force all the internal parameters

and their combination to be constrained inside this antenna's boundary λb/2. **Step 5.** Reduction of parameters: in this step, we did further reduction of the number of parameters involved. Exploiting the fact that the feed-section's parameters have no significant added values to the total performance once its optimum values are founded, and the following prior measures have been set, 1) for impedance-matching it is fixedly set to 50 Ω and, 2) for field-matching the impedance-parameter-pair (WS, WG) has been chosen such that it is wide enough to support the currents to separately flow along the edges of the signal line WS, and the feed length LF must be long enough to support the

parameters it would be an impossible task for the multivariate optimization process.

**Step 7.** Simplified process: The 7th step is the simplification of the optimization process by breaking the MVO process down into series of SVO one. It was observed that the current intensity is mainly distributed along the edges of the transition region, this suggested that : a) small variation of the transition parameter in this section (LT, WT) would have a strong influence on the flow of the signal current and hence play a large impact on the impedance bandwidth, and exploiting the fact, as similarly in microstrip antenna, that b) the length of the radiator defines the shifting-property of the resonance, whilst c) the width of the radiator effectuates the matching-property, and d) another independent-property is that they can be separately optimized.

So by keeping WT, the share-parameter between sections II and III.A (fig.5b), fixed and took just a single-variable-optimization (LT), the impedance bandwidth was double to 20GHz as obtained by prototype 2, which is plotted in fig.9. The parameters of prototype 2 are listed in table.2, more details of the electromagnetic properties and measurements regarding the performances of this prototype can be found in (Tanyer et al., op. cit.).


LT = 1.64mm, then two SVOs were carried out first for WT and then WC. The reason that WT was chosen first is twofold, 1) WT is more sensitive on matching because the current distribution is denser at the lower part, 2) WC is the share parameter that mainly located in section IIIA where the matching effect is week, and it is purposely inserted to control the radiation patterns instead of wideband matching the radiator. The optimized parameters of prototype 4 are listed in table.2. The impedance bandwidth of the prototype 4 is plotted in fig.9, and fig.7; also, its SWB radiation patterns were in fig.8.

Architecture and Design Procedure of a Generic SWB Antenna with Superb Performances for Tactical Commands and Ubiquitous Communications 173

**Table 2.** Parameters of the prototypes; the alphabetical order A, B, C indicates the priority-order of

All prototypes depicted in Fig.4, with their design dimensions listed in table.2, have been fabricated on Duroid RT 5880 high frequency laminate with substrate height h=0.787mm, copper cladding thickness t=17μm, relative dielectric constant εr=2.2, electric and magnetic loss tangents are given by tan δE=0.0027 and tan δH=0, respectively. The foremost reason of choosing this material is that it could relatively afford SWB frequency range up to 77 GHz (Huang et al., 2008, p.64). Other reasons are assessments related to temperature, moisture,

Since the dimension of the SMA connector's flange is considerably large in comparison with the antenna dimension (see Fig.10a), this comparable size exerts a huge impact on the antenna's electromagnetic-properties in particularly to the transmission, scattering and

corrosion and stability, which were investigated in details by (Brown et al., 1980).

parameters in the SVO process.

**5.2. Feed elongation** 

**5.1. Design** 

**5. Design and fabrication** 

Prototypes 1, 2, 3, 4 have been design, fabricated, measured and evaluated, photographs of them are shown in fig 4. At first sign, they seemingly looked different, however they all shared the same topology and architecture as depicted in fig.5a, only one or two parameters is slightly changed to obtained difference desired performances. For comparison, their correspond impedance bandwidths are plotted together in fig.9. The impedance bandwidth enhancement is improved from 10 GHz, to 20 GHz, 40 GHz and beyond 150 GHz as shown in fig.9 and fig.7

#### **4.10. Comparison of the prototypes**

The design parameters of prototype 4 and all other prototypes are listed in table.2, so that, the reader, could independently recheck, or reproduce them without much difficulty.

These are results of the FSD methodology and SVO steps described in previous sub-section (§4.9). Note that the SVO should be carried out orderly by A, B and C. ( bold and capitalized in table.2), A is dedicated as the first to be optimized, keep that optimized parameter fixed, and goes on with B, then continue with C. For example, the prototype 4, (A) first fixing the taper's height LR2 to 4.335mm, then (B) optimizing the taper width WT, and then (C) adjust the WC for the radiation-characteristics. The optimized results showed an SWB impedance bandwidth of at least over 150GHz. In fact the result of prototype 4 (with parameters listed in column 4 of Table.2) shown the downtrend of reflection coefficient for increasing frequency (Fig.7), we expect that prototype 4 will well-behave beyond 150GHz as well.

**Figure 9.** Impedance bandwidth of the developed prototypes. Ordinate: magnitude of the reflection coefficient [dB]; Abscissa: frequency [GHz].

Architecture and Design Procedure of a Generic SWB Antenna with Superb Performances for Tactical Commands and Ubiquitous Communications 173


**Table 2.** Parameters of the prototypes; the alphabetical order A, B, C indicates the priority-order of parameters in the SVO process.
