Adaptive wire bow-tie antenna for GPR applications

(1)

Adaptive Wire Bow-Tie Antenna for GPR

Applications

Andrian Andaya Lestari, Alexander G. Yarovoy, Senior Member, IEEE, and Leo P. Ligthart, Fellow, IEEE

Abstract—In this paper, the basic design of an adaptive ground penetrating radar antenna is introduced. The antenna is able to adapt its input impedance to a variation in the antenna elevation and soil type to keep reflections at the antenna’s terminal min-imum. As a result, energy transfer from the generator to the an-tenna is maximized, which in turn maximizes the energy radiated by the antenna into the ground for different antenna elevations and soil types. The antenna is based on a wire bow-tie structure with variable flare angle for adjusting the antenna’s input impedance. The flare angle variation is realized by short-circuiting the gaps separating the wires from the feed point of the antenna, for which electronic switching devices such as PIN diodes could be used to allow fast and convenient control of the antenna’s flare angle.

Index Terms—Adaptive antenna, bow-tie antenna, ground pen-etrating radar (GPR) antenna, wire antenna.

I. INTRODUCTION

I

N this paper we address a problem related to improvement of the performance of an impulse ground penetrating radar (GPR) from the antenna aspect. In particular, the problem per-tains to an antenna matching problem. We have learned that when an antenna is situated in a proximity to the ground its input impedance varies significantly with the antenna elevation and the type of the ground [1], [2]. From the system point of view, such input impedance variation is disadvantageous as it leads to difficulty in maintaining a matched condition at the an-tenna’s terminal. When maximum power transfer from the gen-erator to the antenna (via a transmission line) is considered crit-ical, one should provide a variable matching device which is capable of coping with such impedance variation to achieve a matched termination of the antenna for different antenna ele-vations and ground types. While such a matching network can easily be devised for time-harmonic applications, it is not the case for time-domain applications. A variable ultrawide-band matching network needed by transient applications is difficult to produce and commercially available ones are expensive.

It has been demonstrated that the input impedance of typ-ical GPR antennas (i.e., dipoles and bow ties) fluctuates more with antenna elevation for a larger dielectric permittivity of the ground [2]. Moreover, it has been found that the fluctuation is significant only for very small distances (i.e., a fraction of the

Manuscript received April 7, 2003; revised April 23, 2004. This work was supported by the Dutch Technology Foundation (STW) under the projects “Improved Ground Penetrating Radar Technology” (1999–2000) and “Ad-vanced Re-Locatable MultiSensor System for Buried Landmine Detection” (2001–2002).

The authors are with the International Research Centre for Telecommunica-tions and Radar (IRCTR), Delft University of Technology, 2628 CD Delft, The Netherlands (e-mail: a.lestari@irctr.tudelft.nl).

Digital Object Identifier 10.1109/TAP.2005.846726

wavelength corresponding to the central frequency of the ex-citing pulse) from the ground [2]. With varying antenna eleva-tion it is difficult to keep refleceleva-tions at the antenna’s terminal minimum and this problem usually occurs during a practical GPR survey, in which the distance between the antenna and the ground varies due to nonflatness of the ground surface and/or the operator’s movement (in the case of a hand-held system).

To deal with the above-mentioned problem one needs an an-tenna which should be adaptive with respect to the anan-tenna ele-vation and soil types. It has been shown that the input impedance of a bow-tie antenna varies with flare angle [3], offering the pos-sibility of antenna matching by flare angle variation. When the input impedance of the antenna is modified by the presence of the ground, one may be able to compensate for it by adjusting the flare angle to keep reflections at the antenna’s terminal min-imum. However, varying the flare angle of a solid bow-tie an-tenna is not easily realizable and therefore solid bow ties are here approximated by wire bow ties. It has been demonstrated that the proposed wire bow-tie structure is a good approximation for a solid bow-tie antenna [4]. In this work, the main advantage of such a wire bow tie is that it offers the possibility for control-ling the flare angle electronically. By virtue of the wire structure, it would be possible to implement electronic switching devices (e.g., PIN diodes) for deactivating certain wire elements of the antenna to form the desired flare angle. This approach allows a fast and convenient way for controlling the bow-tie flare angle required in a real GPR survey, as the antenna should promptly adapt to any changes in antenna elevation and soil type.

II. NUMERICALANALYSIS

A. Numerical Model

The geometry of the proposed adaptive wire bow-tie antenna is depicted in Fig. 1(a). It can be seen that the antenna is basi-cally an array of identical dipole elements having a common feed point and equal angular separation between two neigh-boring elements. The antenna comprises 16 wire elements in each of its arms with 10 angular separation between two ad-jacent elements. In Fig. 1(b) the model of the antenna’s feed region is shown. It can be observed that the wires are separated from the feed point by 2 mm gaps and to obtain the desired flare angle, the corresponding gaps are short-circuited. Hence, the an-tenna is capable of forming eight effective flare angles, namely 10 , 30 , 50 , 70 , 90 , 110 , 130 , and 150 . The term “ef-fective flare angle” is here used to designate a flare angle ob-tained in this way [with inactive (disconnected) wires remain physically present in the antenna] in order to distinguish it from real flare angles, which are obtained when the inactive wires are

(2)

1746 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 53, NO. 5, MAY 2005

Fig. 1. (a) Geometry of the proposed adaptive wire bow-tie antenna used in simulations, and (b) its feed region model with 2-mm gaps separating the wires from the feed point. An effective flare angle is formed by short-circuiting the corresponding gaps with a wire of one segment long. The maximum effective flare angle is 150 and two neighboring wires are separated by 10 . The wire numbers are indicated in (b). The antennalength = 50 cm and the wire diameter = 2 mm.

physically removed from the antenna. In Table I the possible effective flare angles of the proposed adaptive wire bow-tie an-tenna are listed together with the corresponding wires indicated in Fig. 1(b) needed to form the effective flare angles. In sim-ulations an effective flare angle is formed by short-circuiting the corresponding gaps using a wire of one segment long. Since the antenna consists only of wires, the Numerical Electromag-netics Code (NEC-2) [5] has been selected as the main numer-ical tool due to its proven accuracy and efficiency for analyzing wire antennas. However, since NEC-2 renders inaccurate when modeling antennas touching the ground, for zero elevations the homemade code developed in [2] and [6] has been used. This code is based on the method of moments (MoM) with trian-gular-patch methodology using Green’s functions for layered media which are valid at the air-ground interface. For wire struc-tures this code is however much more time consuming than NEC-2 and therefore in this paper it is employed only for zero elevations. In addition, as NEC-2 is not capable of computing

TABLE I

EFFECTIVEFLAREANGLES AND THECORESSPONDINGWIRECONNECTIONS

subsurface fields, another code has been developed in [4]. This code incorporates Green’s functions for a lossy half-space to compute transmit waveforms in the subsurface using the an-tenna currents computed by NEC-2.

B. Free-Space Input Impedance

Assuming the positive time dependence , in Fig. 2 we present the computed input impedance of the proposed antenna (in free space) for all effective flare angles listed in Table I. It is interesting to compare the results with the input impedance of the antenna due to real flare angles, given in Fig. 3. One observes in Fig. 3 that reducing the real flare angle generally leads to an increase of the impedance level with stronger oscillation of the impedance curves, which is a typical behavior of solid bow-tie antennas [7]. For the input impedance due to the effective flare angles in Fig. 2, such a behavior is however observable only for frequencies below 1 GHz. At higher frequencies the impedance behaves oppositely, for which reducing the real flare angle gen-erally leads to a decrease of the resistance level, a more induc-tive behavior of the reactance, and flatter impedance curves. The only conceivable explanation for the occurrence of this induc-tive behavior is the presence of the inacinduc-tive wires which intro-duces an inductive loading in the antenna. We notice that a re-duction of the effective flare angle is followed by an increase in the number of inactive wires, which in turn results in a larger value of the inductance, causing the more pronounced inductive behavior for smaller flare angles shown in Fig. 2. In the figure it can also be seen that the inductive behavior is more pronounced at higher frequencies as the significance of an inductive loading increases with frequency. Moreover, due to the presence of the inductive loading the oscillation of the impedance curves, which is caused by reflections at the antenna ends, is also reduced. From the physical point of view this means that the currents re-flected at the ends induce the inactive wires, so that they contain less energy when returning to the feed point. In effect, especially

(3)

Fig. 2. Computed input impedance of the proposed adaptive wire bow-tie antenna for all possible effective flare angles in free space: (a) resistance and (b) reactance.

at high frequencies this reduces the oscillation of the impedance curves for small flare angles, as seen in Fig. 2. Furthermore, it is worth mentioning that the discontinuities in the form of the peaks observed in Figs. 2 and 3 [e.g., the one that can clearly be seen at about 1 GHz in Figs. 2(b) and 3(b)] have been found to be caused by spurious modes generated in the antenna. These spurious modes can be remedied by using additional wires to connect the ends of neighboring wires [8].

In addition, it can be seen in Fig. 2 that for low frequencies (below 1 GHz) decreasing the effective flare angle generally shifts the reactance down, indicating the more capacitive be-havior of the antenna attributed to the presence of more gaps. As frequency increases, the influence of the capacitive loading introduced by the gaps decreases due to the increase in the elec-trical length of the gaps’ width. Especially for small effective flare angles it is obvious that at high frequencies (above 1 GHz) the role of the capacitive loading is taken over by the inductive loading, indicated by the increase of the reactance.

Fig. 3. Computed input impedance of the proposed adaptive wire bow-tie antenna for all possible real flare angles in free space: (a) resistance and (b) reactance.

C. Reflection Coefficient

From the practical point of view it is imperative to observe the reflections experienced by the exciting pulse at the terminal of the antenna due to mismatch between the antenna and the feed line. In transient applications such as impulse GPR those flections correspond to the input impedance due to the initial re-sponse of the antenna [9]. This is the input impedance that would be “seen” by the exciting pulse at the moment of excitation. In this work the antenna’s initial response is obtained by removing reflections at the antenna ends using a time-window technique [2], [6]. Using the method outlined in [2] the initial-response input impedance of the proposed antenna is then computed as functions of the antenna elevation above the ground with dif-ferent electrical properties. The types of soil considered for the simulations are dry sand ( S/m), dry clay

(4)

Fig. 4. Computed input impedance of the proposed adaptive wire bow-tie antenna as functions of antenna elevation for the eight possible effective flare angles. The ground is (a) dry sand (" = 2:5; = 0:004 S/m), (b) dry clay (" = 16; = 0:03 S/m), and (c) wet clay (" = 25; = 0:06 S/m).

S/m), modeled as a half-space. Furthermore, we employ an ex-citation impulse which is a monocycle with duration of 0.8 ns [2].

The input impedance of the proposed antenna due to the 0.8-ns monocycle has been computed using NEC-2 for a number of elevations from 1 to 6 cm and using the above-men-tioned homemade code for zero elevations. The computed results have been interpolated and are given in Fig. 4 for all possible effective flare angles. The results clearly demonstrate the possibility to adjust the input impedance of the antenna by varying the effective flare angle, which could be useful for matching purposes. However, Fig. 4 also indicates that when the antenna is fed by a traditional 50-Ohm feed line, minimal reflections at the antenna’s terminal would be achieved only by the largest possible effective flare angle (150 ). Thus, in this case varying the flare angle will not be useful for the purpose of antenna matching for different elevations and ground types. This problem can be tackled by feeding the antenna using a feed line with larger characteristic impedance to enlarge the dynamic range of the effective flare angle variation with respect to the variation of the reflections at the antenna’s terminal. To optimize the dynamic range of the effective flare angle variation, we propose to increase the characteristic impedance

of the feed line to the value of the characteristic impedance of a solid bow tie with 90 flare angle in free space, which according to [3] is 188 . However here we take a value of 200 to make it more easily realizable in practice. The resulting reflection coefficients due to the 200-Ohm feed line were computed using the technique described in [4], which is similar to that introduced in [9]. The computed reflection coefficients as functions of elevation for the three soil types are plotted in Fig. 5. As a consequence of this approach, an ultrawide-band impedance transformer is needed to increase the characteristic impedance of the feed line to 200 . Such a device will however not be discussed here, as it lies outside the scope of this paper. It is evident in Fig. 5 that for the three soil types, the ground significantly influences the reflection coefficient only for small distances from the interface. In all cases minimum reflection coefficients are achieved for elevations higher than about 3 cm by an effective flare angle of 110 . For dry sandy soil in Fig. 5(a), when the antenna is very close to the interface (less than about 3 cm from the interface) minimum reflections at the antenna’s terminal can be maintained by switching the effective flare angle to 90 . While for clayey soil in Fig. 5(b) and (c) minimum reflections for very low elevations can generally be achieved by switching the flare angle to 90 and 70 for

(5)

Fig. 5. Computed reflection coefficients of the proposed adaptive wire bow-tie antenna as functions of antenna elevation for the eight possible effective flare angles. The ground is (a) dry sand (" = 2:5; = 0:004 S/m), (b) dry clay (" = 16; = 0:03 S/m), and (c) wet clay (" = 25; = 0:06 S/m). The assumed characteristic impedance of the feed line is 200.

elevations between about 3 and 1.5 cm, and between 1.5 and 0 cm, respectively. It can be seen that for elevations lower than 0.5 cm minimum reflections can also be achieved by an effective flare angle of 50 .

The above results demonstrate the adaptation capability with respect to the reflections at the terminal of the proposed antenna for different elevations and soil types. However, this adapta-tion capability would not be evident when the antenna is fed by a traditional 50- feed line. By increasing the characteristic impedance of the feed line to an appropriate value, one improves the dynamic range of the effective flare angle variation with re-spect to the variation of the reflections at the antenna’s terminal. In this way, minimum reflections can be obtained for different elevations and soil types by adjusting the effective flare angle accordingly. Note that when the characteristic impedance of the feed line is too large, it would also make the adaptation capa-bility ineffective. Thus, the characteristic impedance of the feed line should not be increased excessively. Intuitively, one may choose the optimal value of the characteristic impedance of the feed line to be the value of the characteristic impedance of a 90 bow tie in free space. This seems to be a good choice as it is shown above that minimum reflections at the antenna’s terminal

for different antenna elevations and soil types can be achieved by a relatively wide range of effective flare angles, i.e., 50 , 70 , 90 , and 110 .

D. Radiated Fields

In impulse GPR applications the waveform of the transient fields transmitted into the ground should accurately be known to improve imaging of the subsurface. In particular, for the pro-posed antenna it is important to examine the influence of the gaps and the corresponding inactive wires on the waveform. For this purpose, in the following paragraphs we compare the forms of the antenna for each effective flare angle with the wave-forms due to real flare angles in free space.

It is important first to observe the shape of the transmitted main pulse (the pulse which is radiated directly from the feed point) because it mainly determines the waveforms transmitted by the antenna into the subsurface. To this end, a series of com-putations by NEC-2 has been carried out, for which the 0.8-ns monocycle was used for excitation and the observation point was chosen to be 25 cm in the broadside direction of the an-tenna in free space. The computed transmit waveforms due to effective and real flare angles are shown in Fig. 6, where it can

(6)

Fig. 6. Computed free-space transmit waveforms of the proposed adaptive wire bow-tie antenna at a distance of 25 cm in the broadside direction of the antenna for different (a) effective flare angles and (b) real flare angles.

be seen that the main pulse is followed by strong end reflections for all flare angles. It is important to notice that the shape of the main pulse preserves the shape of the 0.8-ns monocycle for all effective flare angles and is not much changed by the presence of the gaps. In the practical implementation of the antenna one could suppress the end reflections by applying resistive loading, as the wire structure of the antenna allows relatively simple re-alization of such loading.

Furthermore, in Fig. 6(a) we observe that the shape of the end reflections resembles the shape of the first derivative of the main pulse. This behavior can be explained by the inductive property of the antenna introduced by the narrow spacing between the wires (especially near the feed point). Because of this, in frequency domain the current along the wires experiences an inductive impedance proportional to (with being the angular fre-quency and the value of the inductance), which is equivalent to differentiation in time domain. In Fig. 6(b) it can be seen that in the case of real flare angles the end reflections are differentiated more pronouncedly with a larger number of wires as it corresponds

Fig. 7. Computed subsurface transmit waveform of the proposed adaptive wire bow-tie antenna with effective flare angle of 130 . The observation point is located at a 16-cm depth in the broadside direction of the antenna which is elevated 1 cm above dry clay (" = 16; = 0:03 S/m).

to a larger value of . Such a behavior does not occur with a dipole, which consists only of a single wire in each of its arms. It has also been shown in [4], theoretically and experimentally, that the end reflections of a solid bow-tie antenna experience a similar behavior, although less pronounced than in the case of the wire bow-tie antenna shown here. Moreover, we observe that the reduction of the amplitude of the main pulse with decreasing effective flare angle in Fig. 6(a) is less than with decreasing real flare angle in Fig. 6(b). It is found that the amplitude of the main pulse is reduced by 30% and 43% in the former and the latter situation, respectively. This indicates that a portion of the currents is induced on the inactive wires, which in turn contributes to the favorable increase in the amplitude of the main pulse.

To analyze the fields transmitted into the ground, the subsur-face transmit waveforms of the antenna have been computed using NEC-2 and the code for subsurface field analysis men-tioned above. As an example, the transmit waveform of the an-tenna with an effective flare angle of 130 is plotted in Fig. 7. The observation point is located at a 16-cm depth in the broad-side direction of the antenna which is elevated 1 cm above dry clay having and S/m. It obvious that the am-plitude of the end reflections relative to the main pulse is smaller in comparison with the free-space case in Fig. 6(a) due to the presence of the lossy soil in the vicinity of the antenna. Never-theless, it is shown that the main pulse preserves the shape of the 0.8-ns monocycle used for excitation.

Moreover, for impulse GPR it is important to examine the am-plitude of subsurface transmit waveforms since it indicates the amount of the energy transmitted by the antenna into the ground. In fact, the amount of the energy transmitted into the ground depends not only on mismatch losses at the antenna’s terminal but also on the antenna’s footprint. An antenna footprint is de-fined as a distribution of the peak values of transmit waveforms within a horizontal plane on the ground surface or subsurface, which indicates the shape and size of the spot illuminated by the antenna [6]. In Fig. 8, we present the normalized peak-to-peak values of the subsurface waveforms transmitted in the broadside

(7)

Fig. 8. Typical peak-to-peak values (normalized) of the subsurface waveforms transmitted by the proposed antenna in the broadside direction for different effective flare angles. Mismatch losses between the antenna and the feed line are not taken into account.

Fig. 9. Computed footprint of the proposed antenna with 130 effective flare angle (in dB). The footprint is situated horizontally at a depth of 16 cm in dry clay (" = 16; = 0:03 S/m). The antenna is elevated 2 cm above the interface.

direction of the proposed antenna, computed for different effec-tive flare angles. Note that in this figure mismatch losses are not taken into account (the antenna is assumed to be always matched to the feed line), and hence the curve is determined mainly by the antenna footprint which varies with the flare angle [6]. This curve has been found to be typical for different depths, antenna elevations and soil types considered in this paper. It is shown that the largest values correspond to effective flare angles of 110 , 130 and 150 , because for these angles the footprints are the smallest and therefore give the highest radiation intensity on the illuminated spot. As an example, the computed footprint of the antenna is plotted for an effective flare angle of 130 in Fig. 9, where it can be seen that on the dB level the radiated fields concentrate on a nearly circular spot.

Fig. 10. Normalized peak-to-peak values of the subsurface waveforms transmitted by the proposed antenna with 1-cm and 6-cm elevations above the ground for different effective flare angles. Mismatch losses between the antenna and the feed line are taken into account. The assumed characteristic impedance of the feed line is 200. The ground is (a) dry sand (" = 2:5; = 0:004 S/m), and (b) dry clay (" = 16; = 0:03 S/m).

In Fig. 10 the normalized peak-to-peak values of the subsur-face waveforms are plotted by taking the mismatch losses be-tween the antenna and the feed line into account. The antenna is assumed to be fed by a feed line with 200-Ohm characteristic impedance, and elevated 1 and 6 cm above the ground which is assumed to be dry sand in Fig. 10(a) and dry clay in Fig. 10(b). Since the radiated electric field is linearly proportional to the transmission coefficient at the antenna’s terminal, the curve in Fig. 10 is obtained as the product of the curve in Fig. 8 and the transmission coefficients derived from the reflection coefficients presented in Fig. 5. It is demonstrated in Fig. 10 that the energy radiated into the ground can be optimized by varying the ef-fective flare angle for different antenna elevations and ground types. It can be seen that the amplitude of subsurface wave-forms is maximized by changing the effective flare angle to 70 , 90 or 110 for the given elevations and ground types. Note

(8)

Fig. 11. (a) Experimental adaptive wire bow-tie antenna and (b) its feed region. The wires are separated from the bow-tie plates by 2-mm gaps. The wires are hold together by a PVC support.

that those effective flare angles correspond to the minimum re-flection coefficients in Fig. 5 for the respective elevations and ground types. It means that by selecting the proper effective flare angle for a certain elevation and ground type one can optimize both the energy transfer from the generator to the antenna and the energy radiated by the antenna into the ground.

III. EXPERIMENTS

The model of the proposed antenna shown in Fig. 1 has been constructed for experimental investigations. The experimental antenna is shown in Fig. 11. The antenna is 50 cm long and the diameter of the wires is 2 mm. The wires are hold together by a PVC support as can be seen in the figure. The PVC support has been used for this purpose due to practical considerations. However, its presence in the feed region substantially modifies the antenna’s input impedance and hence this leads to a discrep-ancy between the computed and measured input impedance of the antenna [4].

As can be seen in Fig. 11(b), the feed region of the exper-imental antenna consists of the mentioned PVC support, the feed point, two small bow-tie plates, the wires, and the gaps. The small bow-tie plates form the wire junctions at which the wires meet. The 2-mm gaps modeled in Fig. 1(b) have been con-structed in the experimental antenna between the wire ends and

Fig. 12. Measured input impedance of the experimental antenna as functions of elevation above dry sand for all possible effective flare angles.

the bow-tie plates, as shown in the figure. In the measurements, to form a certain effective flare angle the corresponding gaps listed in Table I are manually short-circuited.

The input impedance with respect to the exciting 0.8-ns monocycle has been measured by means of the measurement setup and technique described in [2], which do not require the use of any balun. The antenna is situated above dry sand with relative dielectric permittivity in the 500 MHz–3 GHz range and with very small conductivity. The result as functions of elevation above the dry sand is presented in Fig. 12 for all possible effective flare angles. It is apparent that the antenna has different levels of input impedance for different effective flare angles, which offers an antenna matching possibility by flare angle variation. One also observes that the antenna becomes more capacitive as the effective flare angle is reduced due to the presence of more gaps in the antenna. Such a behavior has been theoretically predicted by the simulation results shown in Fig. 4(a). We observe that in comparison with the computed input impedance in Fig. 4(a), the largest absolute value of the measured input impedance is reduced by about 100 due to the presence of the PVC support in the feed region of the exper-imental antenna. NEC-2 is unfortunately not able to model such a dielectric material in the antenna. Nevertheless, the reduction of the antenna’s input impedance due to the PVC support has been predicted by a simulation using the commercial code FEKO which is capable of modeling a metallic antenna with a dielectric part [4]. This 100-Ohm drop of the input impedance is disadvantageous for the antenna’s adaptation capability be-cause it reduces the dynamic range of the flare angle variation significantly. It means that the possibility for antenna matching by varying the effective flare angle for different antenna eleva-tions and soil types becomes more limited.

Making use of the measurement technique outlined in [2] for avoiding using a balun, the antenna has been fed by a twin semi-rigid cable with 100-Ohm characteristic impedance. The reflection coefficients due to mismatch between the feed line and the antenna are given in Fig. 13 as functions of elevation for all possible effective flare angles. It is shown in the figure

(9)

Fig. 13. Measured reflection coefficients of the experimental antenna as functions of elevation above dry sand for all possible effective flare angles. The characteristic impedance of the feed line is 100.

that the effective flare angle of 130 provides the best-matched condition for all elevations above the sand. This fact can easily be deduced from Fig. 12 where at 130 for most of elevations the resistance and the reactance assume a value of nearly 100 and 0 , respectively. However, in this case the adaptation capability of the antenna is not really evident since the best matched condi-tion is achieved only by a single flare angle for all antenna eleva-tions. In Section II we have proposed to increase the character-istic impedance of the feed line to enlarge the dynamic range of the flare angle variation. By inspection of Fig. 12 one observes that unfortunately increasing the characteristic impedance of the feed line in this case would degrade the reflection coefficient for very small elevations (few millimeters) as the values of the re-sistance for these elevations are already smaller than 100 . It is found that to some extent this problem can be remedied by slightly lowering the characteristic impedance of the feed line to a proper value. In Fig. 14 the reflection coefficients due to a feed line with 90-Ohm characteristic impedance is plotted for all possible effective flare angles. In this case the adaptation ca-pability of the antenna is more evident as minimum reflections are obtained by two different effective flare angles, i.e., 130 for very low elevations and 150 for higher elevations. Hence, one can obtain the best matched condition for any antenna elevation by switching the effective flare angle to either 130 or 150 . In Fig. 4 it is indicated that generally when the permittivity of the ground changes from small to large (e.g., the ground changes from sand to clay), the absolute value of the input impedance drops by approximately 30 or more for elevations lower than 5 mm. This suggests that the adaptation capability of the antenna with 90-Ohm feed line would be more pronounced for a ground with larger permittivity since the minimum reflection coefficient for very low elevations would be shifted to a smaller effective flare angle, which in effect enlarging the dynamic range of the flare angle variation. In such a situation minimal reflection coef-ficients for different elevations could be obtained by two or more effective flare angles, as demonstrated theoretically in Fig. 5. In

Fig. 14. Measured reflection coefficients of the experimental antenna as functions of elevation above dry sand for all possible effective flare angles. The characteristic impedance of the feed line is assumed to be 90.

addition, Fig. 5 shows that the adaptation capability of the an-tenna would be much better without the use of the PVC sup-port which significantly reduces the absolute value of the input impedance. Hence, when the use of such a support is inevitable, one should select a dielectric material with minimal impact on the input impedance.

In this paper, the effective flare angle variation is performed manually by short-circuiting the gaps between the wires and the feed point. As this approach is adequate for a laboratory evalua-tion, it is not acceptable for a practical implementation since the antenna should be able to adapt to ground and elevation changes promptly and therefore the short-circuiting of the gaps should be performed rapidly. To realize the effective flare angle variation in the most convenient way one should be able to short-circuit the gaps electronically, for which electronic switching devices (e.g., circuits involving PIN diodes) are needed. The implemen-tation of such switching devices in the proposed antenna will be reported in a future paper.

IV. CONCLUSION

An adaptive antenna for impulse GPR applications has been proposed. The antenna is capable of adapting its input impedance against a variation in the antenna elevation and soil type for maintaining a minimum reflection at the antenna’s ter-minal to obtain the best matched condition between the antenna and the feed line. The antenna is based on a wire bow-tie struc-ture and the antenna adaptation is achieved by varying the flare angle of the antenna, for which the wires are separated from the feed point by narrow gaps. It has been shown that the flare angle variation can be done by short-circuiting the gaps, which would allow one to control the flare angle electronically using electronic switching devices such as PIN diodes. It has been demonstrated theoretically that the best matched condition can be achieved by at least three different effective flare angles for different antenna elevations and soil types. As a result, energy transfer from the generator to the antenna is maximized, which

(10)

in turn maximizes the energy radiated by the antenna into the ground. The adaptation capability of the proposed antenna has been confirmed experimentally.

ACKNOWLEDGMENT

The authors thank P. Hakkaart for his assistance in the con-struction of the experimental antenna and J. Zijderveld for his assistance in the measurements.

REFERENCES

[1] G. Tuner, “The Influence of subsurface properties on ground penetrating radar pulses,” Ph.D. dissertation, Macquarie University, Sydney, NSW, Australia, 1993.

[2] A. A. Lestari, A. G. Yarovoy, and L. P. Ligthart, “Ground influence on the input impedance of transient dipole and bow-tie antennas,” IEEE Trans. Antennas Propag., vol. 52, no. 8, pp. 1970–1975, Aug. 2004. [3] R. L. Carrel, “The characteristic impedance of two infinite cones of

arbi-trary cross section,” IRE Trans. Antennas Propag., vol. 6, pp. 197–201, Apr. 1958.

[4] A. A. Lestari, “Antennas for improved ground penetrating radar: mod-eling tools, analysis, and design,” Ph.D. dissertation, Delft Univ. Tech-nology, The Netherlands, 2003.

[5] G. J. Burke and A. J. Poggio, Numerical Electromagnetic Code (NEC)—Method of Moments. San Diego, CA: NOSC TD-116, Naval Ocean Syst. Center, Jan. 1980.

[6] A. A. Lestari, A. G. Yarovoy, and L. P. Ligthart, “Numerical and ex-perimental analysis of circular-end wire bow-tie antennas over a lossy ground,” IEEE Trans. Antennas Propag., vol. 52, no. 1, pp. 26–35, Jan. 2004.

[7] G. H. Brown and O. M. Woodward, “Experimentally determined radi-ation characteristics of conical and triangular antennas,” RCA Rev., pp. 425–452, Dec. 1952.

[8] K. D. Palmer, “Private communication,” unpublished, 2002.

[9] R. W. P. King and H. J. Schmitt, “The transient response of linear an-tennas and loops,” IRE Trans. Anan-tennas Propag., vol. 10, pp. 222–228, May 1962.

Andrian Andaya Lestari was born in Bogor,

Indonesia. He received the ingenieur and Ph.D. de-grees in electrical engineering from Delft University of Technology, The Netherlands, in 1993 and 2003, respectively.

From 1993 to 1998 he was with a government research agency in Jakarta, Indonesia. He joined the International Research Centre for Telecommu-nications and Radar (IRCTR), Delft University of Technology, as a Researcher in 1998. His work at IRCTR has resulted in over 30 publications, which include patents, scientific reports, and articles presented in journals and inter-national conferences. Currently he works on development of ultrawide-band antennas for ground penetrating radar applications and numerical tools for analysis of transient antennas near a lossy ground. He is IRCTR’s representa-tive in Indonesia and the coordinator of joint research projects in the field of telecommunication and radar between IRCTR and Indonesian organizations.

Alexander G. Yarovoy (M’96–SM’04) received

the Diploma (with honors) in radiophysics and electronics and the Cand. Phys. & Math. Sci. and Dr. Phys. & Math. Sci. degrees in radiophysics, from Kharkov State University, Kharkov, Ukraine, in 1984, 1987, and 1994, respectively.

In 1987, he joined the Department of Radio-physics, Kharkov State University, as a Researcher and became a Professor in 1997. From September 1994 through 1996, he was with the Technical University of Ilmenau, Germany, as a Visiting Researcher. Since 1999, he has been with the International Research Centre for Telecommunications-Transmission and Radar (IRCTR), Delft University of Technology, The Netherlands, where he coordinates all GPR-related projects. His main research interests are in ultra wideband electromagnetics, wave scattering from statistically rough surfaces and penetrable obstacles and computational methods in electromagnetics.

Leo P. Ligthart (M’94–SM’95–F’02) was born in

Rotterdam, the Netherlands, on September 15, 1946. He received the Engineer’s degree (cum laude) and the Doctor of Technology degree from Delft University of Technology, Delft, The Netherlands, in 1969 and 1985, respectively, the Doctorates (honoris causa) from Moscow State Technical University of Civil Aviation, Moscow, Russia, in 1999, and the Doctorates (honoris causa) from Tomsk State University of Control Systems and Radioelectronics, Tomsk, Russia, in 2001. He is an academician of the Russian Academy of Transport.

Since 1992, he has held the Chair of Microwave Transmission, Radar and Remote Sensing in the Department of Information Technology and Systems, Delft University of Technology, where in 1994, he became Director of the In-ternational Research Centre for Telecommunications-Transmission and Radar. He has published over 300 papers. His principal areas of specialization include antennas and propagation, radar and remote sensing, but he has also been active in satellite, mobile, and radio communications.