Landmines are buried and forgotten, so often that an estimated total of 110 million landmines are buried in over 60 countries around the world. Statistics show that hidden and forgotten landmines kill or injure approximately 70 people each day, resulting in over 25,000 deaths or injuries per year. Unfortunately, there is little international effort to detect and clear the landmines due to shortage of funding. Before they can be removed, they must be located and ultrawideband (UWB) ground-penetrating-radar (GPR) technology is ideal for finding the location of these deadly devices.

Newer landmines are built without metal, rendering conventional metal detectors useless for locating these landmines. But several methods can be used for surveying buried landmines, including the use of nuclear quadrupole resonance, electromagnetic (EM) induction spectroscopy, thermal neutron activation, infrared imaging, biological detection, and UWB GPR systems. GPR has been contributed by microwave engineers for this humanitarian purpose, although the technology also has many other commercial applications. A GPR system works by transmitting EM pulses into the ground and detecting the backscattered EM waves that indicate changes in permittivity, permeability, and conductivity in the ground.1 If interpreted correctly, the received signals can be used to indicate the presence of a landmine beneath the surface.

One of the most critical subsystems in a GPR system is the antenna. Because of its use of short EM pulses, a GPR system must process an extremely wide bandwidth. The antenna must provide consistent performance over an UWB frequency range. But in contrast to antennas used for UWB wireless communications, antennas for GPR systems must have the following characteristics:

1. Efficient coupling of EM waves into the ground, i.e., good impedance matching at the antenna/ground interface;
2. Short ringing times (ringing occurs due to reflection within the antenna as well as the reflections at the antenna/ground interface, and these reflections can mask target return signals);
3. High power-handling capability so that the EM waves can achieve adequate ground penetration and reach a target with sufficient amplitude to generate a receivable return signal;
4. Adequate bandwidth to benefit from both larger penetration depth of low-frequency signals, as well as the high range resolution of higher-frequency signals;
5. Provide nondispersive propagation, radiating all frequencies at the same speed (or as a linear function of frequency) and have a constant phase center;
6. Have impedance that is well matched across the entire bandwidth in the actual operating setup where the antenna is near to the ground;
7. Provide stable performance for different ground properties and at different elevation levels;
8. Provide high directivity, concentrating EM energy into a narrow solid angle into the ground;
9. Provide an antenna pattern that does not change significantly over the operating bandwidth;
10. Exhibit constant polarization over the operating bandwidth;
11. Provide weak coupling between the transmit and receive antennas in a bistatic configuration;
12. Be compact enough to occupy a reasonably small volume; and
13. Be inexpensive to produce.

Element-based antennas, such as biconical and printed dipole antennas, and frequency-independent antennas, such as spiral, Vivaldi, and log-periodic antennas, can only meet some of these requirements. While a TEM horn antenna can meet most of the requirements,2-12 some modifications must be made to satisfy the remaining requirements.

The medium into which the antenna must transmitthe groundis generally lossy, inhomogeneous, sometimes anistropic, and exhibits frequency-dependent attenuation. Some media with high resistivity, such as gravel, sand, dry rock, and fresh water, present low signal attenuation and are relatively easy to probe. Media with low resistivity, such as clay soil, ground with conductive ore and minerals, and saltwater, present large signal attenuation, making it difficult for EM waves to penetrate. The average soil has a relative permittivity, er, of 2 to 9 and attenuation factor, e, of 5 to 10 dB/m.13

In order to improve the impedance match at the antenna/ ground interface, a dielectric with relative permittivity, er, that is close to the ground medium is chosen for the design of the TEM horn antenna. At the same time, establishing a good impedance match between the antenna and the ground also reduces antenna coupling between the transmitter and the receiver and decreases the physical dimensions of the antenna. (The length of a basic TEM horn antenna is typically three times the pulse duration multiplied by c, the speed of light.)

The plates of the horn antenna can be linearly tapered or exponentially tapered. Linearly tapered plates are easier to construct. Exponentially tapered plates have the advantage of lower internal reflection and larger bandwidth. If the plates of the antenna are considered to be a transmission line, then the parameters that determine the impedance at a point along the antenna are the plate width and the spacing between the plates. Generally, the further apart the plates are, the greater the impedance will be. Thus, it is possible to gradually increase the impedance of the antenna from around 50 ohms at the feed point to the wave impedance of the propagation medium (377 ohms for air and lower impedances for other media depending on their dielectric constants) at the aperture to match both the input and output impedances and hence reduce unwanted reflection at these two points. However, there are finite reflections along the antenna plate as its separation distance changes. The idea here is to vary the plate width and separation such that these internal reflections are minimal. Huang et al. have shown that the optimum performance can be achieved with a nonlinear impedance profile from feed point to the TEM horn aperture which means that the resulting antenna plate width would also be nonlinear.11

A wedge-shaped dielectric material was used (as opposed to pyramid-shaped dielectric) in the design of the TEM horn antenna. Apart from the electrical advantages described above, it also serves as substrate to hold the plates at the desired separation. The design steps for the TEM horn antenna can be summarized as follows:

1. Specify the maximum input reflection in the passband as well as the lowerfrequency limit:
|R(0)|max = the maximum input reflection in the passband and fmin = the lower-frequency limit of the passband

2. Specify the characteristic impedance of the antenna at the feed point and at the aperture, where
Z0(0) = the characteristic impedance at the feed point and
Z,0(d) = the characteristic impedance at the aperture.

3. Calculate parameter B in Eq. 1

Parameter B will be used to determine the exponential taper in the antenna plates. Larger values of B will result in a greater curve i the taper and lower reflection at the input. While this is good, as will be seen in the next step, larger values of B result in longer antennas 4. Calculate the length of the antenna, d, using Eqs. 2 and 3.

where
d = the length of the antenna and
v = the velocity of the EM wave in the dielectric medium filling the void between the antenna plates. It is clear from these equations that having a lower minimum frequency, fmin, will result in a longer antenna. In addition, a larger value of B will increase the length of the antenna.

5. Calculate the optimum characteristic impedance profile of the antenna to minimize internal reflections, as shown by Eq. 4.14

where
Z0(z) = the characteristic impedance of the antenna at position z and parameter G(B, ?) is given by Eq. 5.

Continue to page 2

Page Title

6. Calculate the separation of the plates at the aperture. The reflection of the aperture can be found by using Eq. 6.

equ

where
b = the separation of the plates at the aperture and fb = v/b.

7. Calculate the plate width to plate separation profile of the antenna, using Eq. 7.

equ

where
? = the wave impedance of the medium filling the void between the antenna plates and
w = x(z)/y(z) = the ratio of the plate width to the plate separation.

8. Select a separation profile for the antenna. The antenna plate separation can be either linear or exponential. For linear separation, apply Eq. 8a and for exponential separation, use Eq. 8b.

equ

where
bline = the plate separation at the feed point.

9. Calculate the width profile of the antenna, using Eq. 9.

equ

A TEM horn section was designed using the parameters in the table. The calculated width profile is shown in Fig. 1. The TEM horn antenna is balanced, having at least one plane of symmetry (Fig. 2). It is important to feed the two plates with equal and opposite currents at the feed point. If the feed current is not balanced, the antenna will transmit an asymmetric E-plane radiation pattern. Furthermore, due to an unbalanced current, the feed line will also radiate and become part of the radiating structure. For these reasons, unbalanced current at the feed point should be avoided. There are a number of ways to feed the antenna with balanced current15:

feed the antenna with a balanced source precisely at the feed point;
feed it with a balanced source through a balanced transmission line; or
feed the antenna through an unbalanced transmission line (e.g., coaxial cable) and a balanced-to-unbalanced transformer (balun).

Since it is much more convenient to feed the antenna through a coaxial cable, it is necessary to design a suitable balun for the task, customized to the properties of the antenna. The balun must operate over the ultrawideband frequency range of the TEM horn section. This is achieved by extending the metal plates at the feed end and optimizing the shape through simulation using the High-Frequency Structure Simulator (HFSS) EM simulation software from Ansoft Corp.. The final design is shown in Fig. 3.

Computer simulations were performed to optimize the width of the dielectric wedge. The widest part of the plate is 100 mm. If the wedge width is set at 100 mm, the radiation pattern will exhibit a broad beamwidth but many undesirable sidelobes (Fig. 4). By increasing the wedge width to 150 mm, the beamwidth becomes much narrower, hence with higher gain, and the sidelobes are eliminated (Fig. 5). A further increase in the width of the dielectric wedge does not provide significant improvement in antenna gain. The input VSWR of the TEM horn antenna was predicted through computer simulation with and without the balun. The balun was fnd to improve the VSWR significantly at higher frequencies.

Computer simulations were also performed to study the effect of ground proximity. Assuming the ground medium is a dry loamy soil with dielectric constant of 2.5 and bulk conductivity of 10-3 S/m, simulations were performed of the antenna input VSWR when the aperture is 10 mm, 50 mm, and 100 mm above the ground. Better performance is achieved at low frequencies when the antenna aperture is placed nearer to the ground.

With the antenna 10 mm above the ground, computer simulations were performed and compared to the measured VSWR performance for different soil dielectric constants. The results show that efficient coupling of EM waves into the ground, i.e., good matching at the antenna/ ground interface, can be achieved for soil dielectric constants above 2.5.

Figure 6 shows a picture of the fabricated TEM horn antenna supported on polystyrene foam. The measured VSWR is below 2.0:1 over the entire UWB bandwidth from 1 GHz to above 10 GHz (Fig. 7). Due to space constraints in this article, only a few plots of the measured radiation patterns are shown here (Figs. 8 to 11). A good match between the simulation and measurement results is obtained. The radiation patterns are adequately directional over the operating frequency range, with antenna gain oscillating around 13 dB.

A time-domain-reflectometry (TDR) experiment using a microwave vector network analyzer (VNA) was conducted with a small iron block (10.5 x 9.0 x 2.5 cm) buried 14-cm under the ground. Although not shown here, those measurements compared the performance of the antenna with TDR pulses directed to the sky as well as with TDR traces directed to the ground. The measured signals with the VNA represented reflections at the antenna input terminal, which was the largest of the measured signals, and a reflection due to the antenna/medium (in this case, the ground) interface. A third pulse measured with the microwave VNA represents the backscatter signal from the target, measured as a lower, return signal as commonly done with military radar systems. The experiment verifies that the TEM horn antenna is suitable for the GPR land-mine-siting application with no ringing problem. With the addition of approriate signal processing and software, any backscatter signals can be appropriately identified and labeled to simplify identification.

REFERENCES
1. B. Scheers, M. Piette, and A. Vander Vorst, "The Detection of AP Mines Using UWB GPR," 2nd Intnl. Conf. on the Detection of Abandoned Land Mines, 1998, Conf. Pub. No. 458, Oct. 12-14, 1998, pp. 50-54.
2. D. A. Kolokotronis, Y. Huang, and J. T. Zhang, "Design of TEM Horn Antennas for Impulse Radar," High Frequency Postgraduate Student Colloquium, 1999, pp. 120-126.
3. R. T. Lee and G. S. Smith, "A Design Study for the Basic TEM Horn Antenna," IEEE Antennas & Propagation Mag., Vol. 46, No. 1, February 2004, pp. 86-92.
4. R. T. Lee and G. S. Smith, "On the Characteristic Impedance of TEM Horn Antennas," IEEE Transactions on Antennas & Propagation, Vol. 52, No. 1, January 2004, pp. 315-318.
5. A. G. Yaravoy, A. D. Shukin, I. V. Kaploun, and L. P. Ligthart, "The Dielectric Wedge Antenna," IEEE Transactions on Antennas & Propagation, Vol. 50, No. 10, October 2002, pp. 1460-1472.
6. Kyung-Ho Chung, S. Pyun, and Jae-Hoon Choi, "Design of an Ultrawideband TEM Horn Antenna with a Microstrip-Type Balun," IEEE Transactions on Antennas & Propagation, Vol. 53, No. 10, October 2005, pp. 3410-3413.
7. W. Burnside and C. Chuang, "An Aperture- Matched Horn Design," IEEE Transactions on Antennas & Propagation, Vol. 30, No. 4, July 1982.
8. M. J. Ahmed, "Impedance Transformation Equations for Exponential, Cosine-Squared, and Parabolic Tapered Transmission Lines," IEEE Transactions on Microwave Theory & Techniques, Vol. 29, No. 1, January 1981, pp. 67-68.
9. Li Jing, X. Y. Zhu, M. X. Wang, and Jen Lang, "A New Design of TEM Horn Antennas for Pulse Radiation," Asia-Pacific Microwave Conference Proceedings, Vol. 2, Dec. 2-5, 1997, pp. 629-631.
10. R. V. de Jongh, A. G. Yarovoy, L. P. Ligthart, I. V. Kaploun, and A. D. Schukin, "Design and analysis of new GPR antenna concepts," 7th Intnl. Conference on Ground Penetrating Radar, Vol. 1, Issue, May 27-30, 1998, pp. 81-86.
11. Y. Huang, M. Nakhkash, and J. T. Zhang, "A dielectric material loaded TEM horn antenna," 12th Intnl. Conference on Antennas and Propagation, Vol. 2, March 31 - April 2003, pp. 489-492.
12. Jeong Hwan Kim and Jeong Il Park, "TEM Horn Antenna for the Time Domain Shielding Effectiveness Measurement," Intnl. Symposium on Electromagnetic Compatibility Proceedings,
21-23 May 1997, pp. 265-269. 13. D. J. Daniels, Ground Penetrating Radar, 2nd Ed., IEE Radar, Sonar and Navigation Series 15, 2004.
14. R. P. Hecken, "A Near-Optimum Matching Section without Discontinuities," IEEE Transactions on Microwave Theory and Techniques, Vol. 20, No. 11, November 1972, pp. 734739.
15. M. Manteghi and Y. Rahmat-Samii, "A Novel UWB Feeding Mechanism for the TEM Horn Antenna, Reflector IRA, and the Vivaldi Antenna," IEEE Antennas and Propagation Mag., Vol. 46, No. 5, Oct. 2004, pp. 8187.