TEM HORN ANTENNA FOR NEAR-FIELD MICROWAVE IMAGING

Transcription

1 Figure 11 E-Plane and H-plane the radiation pattern at 1 and 20 GHz. [Color figure can be viewed in the online issue, which is available at discontinuities at different operating band. EBG structures are used to enhance the antenna performance and decrease bandwidth discontinuity. 2D-EBG as etched dumb-bell shape on feed line to improve impedance matching and using embedded AMC as four arms spiral to reduce the cross polarization and ultimately improve the bandwidth and increase the numbering of 0 reflection angle, hence, increase the antenna gain. There is a good agreement between simulated and measured results for the proposed antennas. Further more acceptable E- and H-plane radiation pattern at different frequencies with average antenna gain13 dbi are achieved. REFERENCES 1. G. Kumar and K.C. Gupta, Directly coupled multiple resonator wide-band microstrip antenna, IEEE Trans Antennas Propag 33 (1985), F. Yang, X.-X. Zhang, X. Ye, and Y. Rahmat-Samii,Wide-band E- shaped patch antennas for wireless communications, IEEE Trans Antennas Propag 49 (2001), S. Weig, G.H. Huff, K.H. Pan, and J.T. Bernard, Analysis and design of broadband single layer U-slot microstrip patch antennas, IEEE Trans Antennas Propag 51 (2003), K.L. Wong and Y.F. Lin, Small broadband rectangular microstrip antenna with chip-resistor loading, Electron Lett 33 (1997), S. Cheng, P.H. Jorner, and A. Ryberg, Printed slot planar inverted cone antenna for ultra wideband applications, IEEE antennas wireless propag 7 (2008). 6. Artificial magnetic conductors, soft/hard surfaces and other complex surfaces (special issue), IEEE Trans Antennas Propag 53 (2005). 7. G. Cakir and L. Sevgi, Design of a novel microstrip electromagnetic band-gap (EBG) structure, Microwave Opt Technol Lett 46 (2005), M. Gujral, T. Yuan, C.-W. Qiu, L.-W. Li, and K. Takei, Bandwidth increment of microstrip patch antenna array with opposite double-e VC EBG structure for different feed positions, Int Symp Antennas Propag (ISAP) (2006). 9. D. Nashaat, H.A. Elsadek, E. Abdallah, H. Elhenawy, and M.F. Iskander, Enhancement of ultra-wide bandwidth of microstrip monopole antenna by using metamaterial structures, IEEE Int Symp Antennas Propag, Charleston, SC (2009) Wiley Periodicals, Inc. TEM HORN ANTENNA FOR NEAR-FIELD MICROWAVE IMAGING Mark A. Campbell, Michal Okoniewski, and Elise C. Fear Department of Electrical Engineering, Schulich School of Engineering, University of Calgary, 2500 University Drive NW, Calgary AB Canada T2N 1N4; Corresponding author: fear@ucalgary.ca Received 22 July 2009 ABSTRACT: Antennas capable of sending and receiving ultra-wideband pulses are required for radar-based microwave breast imaging. This article describes a TEM horn antenna designed to operate over a 2 12 GHz band with specific radiated near-field characteristics. Simulations and experimental measurements are presented, including detection of objects representing tumors. VC 2010 Wiley Periodicals, Inc. Microwave Opt Technol Lett 52: , 2010; Published online in Wiley InterScience ( DOI /mop Key words: antenna; ultra-wideband; radar; microwave breast imaging 1. INTRODUCTION Microwave techniques for breast imaging have recently generated a great deal of interest [1]. One approach, radar-based imaging, illuminates the breast with short-time pulses and collects reflections at the same or multiple antenna locations MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May 2010 DOI /mop

2 Radar-based methods rely on differences in the electrical properties of tissues in the breast to generate reflections which are then processed to create images. A recent study shows that, as there is a 1:10 contrast between adipose tissue (fat) and malignant tissue, the contrast between glandular/fibroconnective tissue in the breast and malignant tissue is on the order of 10:11 [2]. The presence of high permittivity tissue near a tumor increases the challenges of tumor detection with radar-based microwave breast imaging. Antennas with minimal reflections from the antenna structure and predictable field patterns facilitate interpretation of reflections from complex distributions of tissues in the breast. This implies that the antenna is a critical component of radar-based microwave imaging systems. Numerous antenna designs for radar-based microwave breast imaging are evident in the literature. For example, real-aperture synthetically organized radar is a multistatic system operating from 4 to 9 GHz and incorporating an array of low-profile, stacked patched antennas distributed on a hemisphere [3]. A second example is microwave imaging via space-time beamforming (MIST) which is a monostatic system [4]. MIST incorporates a pyramidal horn antenna that transmits and receives ultra-wideband (UWB) pulses (1 11 GHz) as the antenna is scanned over or around the breast. Although this antenna exhibits excellent performance, resistors incorporated in the design reduce efficiency and a small beamwidth is expected close to the antenna, implying that data may need to be collected at a greater number of spatial locations as the antenna scans the breast. A printed monopole antenna fed with a co-planar waveguide has been designed, implemented, and used to demonstrate tumor sensing [5]. A tapered slot antenna has been investigated in [6]. However, none of the proposed designs provides an efficient antenna operating over the range from 2 to 12 GHz, and exhibiting low late-time reflections, predictable radiated near-field energy and compatibility with the imaging system under development at the University of Calgary [7]. Specifically, the patient lies in the prone position with breast extending through a hole in the examination table and immersed in canola oil. The antenna is immersed in the tank of canola oil and scanned around the breast at an offset distance of 1 2 cm, implying that the antenna must function in the immersion medium and fit within the tank, as described in more detail in Section 2. To date, we have developed several specific antennas for our imaging system. For example, a tapered slot antenna defined with linear, exponential, and elliptical sections over various portions of its profile has been tested [8]. This design operates over the frequency range of GHz and has a directional radiation pattern, however, late time reflections from within the antenna increase the difficulty of interpreting reflections from the breast. In parallel to the research reported in this article, a balanced antipodal Vivaldi antenna is under development [9]. In this article, we explore the transverse electromagnetic (TEM) horn antenna, a known broadband design with many desirable features. As far as the authors are aware, this antenna has not been investigated previously for near-field, radar-based breast cancer detection. The basic TEM horn antenna consists of two simple triangles separated by an acute angle (e.g., [10]). The antenna is often described by three parameters: angles a (plate flare) and b (plate separation) and length S. If the length of this antenna is longer than several wavelengths, its impedance can be approximated by the impedance of an infinite pyramidal waveguide described by angles a and b [10]. Several approaches to improving the performance of TEM horns have also been proposed. For example, resistive loading has been suggested to reduce reflections from the structure [11]. Changing the separation and width of the plates along the structure has also been proposed in order to obtain a desired impedance profile [12, 13]. We refer to this design approach as parallel plate (PP) to reflect the calculation of the impedance at a specific point along the antenna utilizing a parallel plate line of a given width and separation. Various methods for specifying the gradual change in impedance have been used, such as exponential [13] and optimum [12] tapers. The variation in impedance may also be combined with resistive loading [12]. Although a design that includes variations of impedance along the antenna length without resistive loading has been reported [13], the performance of the reported design has not been assessed for near-field imaging. In this article, we develop a TEM horn antenna suitable for radar-based microwave imaging. The required performance characteristics are obtained utilizing the PP approach to design and without resistive loading. In Section 2, antenna design methods and simulation results are outlined. Section 3 presents measured antenna results, demonstrating the ability to detect small objects. Finally, Section 4 presents conclusions. 2. ANTENNA DESIGN AND SIMULATIONS 2.1. Design The TEM horn antenna is designed in order to satisfy a number of criteria: (1) The antenna must operate in an immersion medium of canola oil (e r 2.5). This liquid provides reduced reflections from the skin, the ability to reduce antenna size, and has advantages for imaging [14]. (2) System dimensions require the antenna to fit inside a volume of 6 cm 6cm 8 cm. (3) It is desired to have the antenna operate over a frequency range of 2 12 GHz with voltage standing wave ratio (VSWR) less than 2 or, equivalently, reflection coefficient (S 11 ) better than 10 db. This bandwidth represents a trade-off between resolution and tissue penetration. (4) A directional radiation pattern is desired with a half-energy beamwidth of at a distance of 2 3 cm from the aperture. A half-energy beamwidth of 15 would illuminate approximately half of a breast phantom of 10 cm diameter, whereas a width of 40 would illuminate the majority of the phantom. The shape of the radiated energy pattern in the boresight direction would ideally be circular (cross section of a single beam) and not change greatly for pulses of different frequency content. Finally, a close match between the radiated pulse time signature and the derivative of the excitation signal is desired. This is evaluated via fidelity [15], which is defined as: Fðx; y; zþ ¼max s Z 1 1 ^Eðx; y; z; t þ sþ^rðtþdt where ^Eðx; y; z; tþ is the dominant electric field observed at spatial location (x,y,z) and ^rðtþ is a reference signal. Both of these signals are normalized (hat notation) in order to compute fidelity. Fidelity values over 0.90 are targeted. A TEM horn is a balanced antenna, however, in typical use it is fed from a coaxial cable. Therefore, it is necessary to also design a balun. The microstrip-to-parallel strip balun developed for this antenna has been previously reported [16]. Simulations showed VSWR to be less than 31 db over the frequency range of 2 13 GHz and the output currents to have a balance of 96% when considering peak-to-peak current pulse values in each conductor. The balun output impedance and the antenna input impedance, Zo min, are fixed at 50 X. (1) DOI /mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May

3 Figure 1 Coordinate system used to define plate separation (y) and profile (x). The width of the plates (z) also varies along the antenna structure To describe the antenna, the coordinate system shown in Figure 1 is used. The maximum length of the antenna (X max )is initially fixed at 80 mm, whereas the separation profile [Y(x)], impedance profile [Zo(x)], maximum aperture (Y max ), and impedance at the aperture (Zo max ) are all varied. Figures 2 and 3 show the various separation and impedance profiles that are considered. After selecting impedance and separation profiles to test, the plate widths [W(x)] that give the desired impedance profile are calculated. The antenna is assumed to be immersed in and surrounded by oil. Wheeler introduced a method to approximate the characteristic impedance of a PP transmission line in air with a dielectric substrate between the conductors [19]. This technique is modified to allow for the calculation of impedance in a homogeneous medium of arbitrary permittivity. To verify the modified equations, full wave simulations are also performed and computed impedances agree to within 2% of calculations. The modified equations are used to calculate the plate widths for the impedance and separation profiles tested. Figure 3 Impedance profiles considered. The exponential-mirrored separation profile is similar to another TEM horn antenna design [17]. The near-optimum impedance profile is a variation on the Klopfenstein taper [18] 2.2. Simulations To evaluate potential designs, we utilize the method of moments (MoM) solver FEKO (EM Software & Systems, Stellenbosch, South Africa). Electric and magnetic symmetry are used so that only one quadrant of space is specified in the model. To evaluate antenna time domain characteristics, it is necessary to generate pulses synthetically. The frequency domain electric field at a location is multiplied by the spectrum of the desired input pulse and then an inverse chirp Z transform is performed to obtain the desired time domain signal. Two different pulses are used to evaluate the radiated energy pattern and signal fidelity in order to give a sense of the robustness of the designs to different time-domain excitations. These are the differentiated Gaussian pulse (1) and the Gaussian modulated sine pulse (2), given in Table 1. Three different metrics are used to evaluate the performance of antenna designs, namely, (1) input VSWR, (2) radiated energy pattern, and (3) radiated signal fidelity. The first metric is evaluated in the frequency domain, whereas the other two are evaluated in the time domain. To obtain the radiated energy pattern, which is pulse specific, we calculate the relative energy flow density of the radiated pulse at specified points in front of the antenna. These locations are determined using a spherical coordinate system and grid of points separated by 2 and ranging from 0 < u < 26 and 64 < h < 90 (for reference, the boresight direction is u ¼ 0 and h ¼ 90 ; i.e., the x-axis). Fields are evaluated at radial distances of 1, 2, and 3 cm from the antenna aperture. The square of the time-domain electric field is integrated at the selected locations, and normalized to the maximum value recorded at a given radial distance. Once the radiated energy flow density values are calculated, it is then possible to determine the pattern shape and the half-energy beamwidth. The radiated signal fidelity is also pulse specific. For a TEM horn antenna, the radiated pulse is the time derivative of the current flowing on the antenna; the time-derivative of the excitation pulse is used as the reference signal in Eq. (1). When evaluating the radiated pulse fidelity, the same locations as the energy pattern are used. However, only locations with energy greater than 50% of the maximum energy are considered. Figure 2 Candidate separation profiles 2.3. Simulation Results VSWR Results. First, the influence of the impedance at the antenna aperture (Zo max ) is investigated. Simulations are conducted with a near-optimum impedance profile (see Fig. 2), a circular separation profile and Y max ¼ 24 mm (k/4 at 1166 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May 2010 DOI /mop

4 TABLE 1 Equations and Parameters Used to Define Two Pulses Expression Variables Frequency Range (GHz) mðtþ ¼V o ðt t o Þexp ðt toþ2 12 s (1) s ¼ Peak: 3.6 mðtþ ¼V o sin½2pðf cen Þðt t o ÞŠ exp ðt toþ2 2r 2 ð2þ f cen ¼ 6.75 GHz r ¼ Peak: 6.75 Both pulses have t o ¼ 0.35 ns. Frequency range is defined by identifying content greater than 10% of the peak value. 2 GHz). Results obtained with various Zo max values are presented in Figure 4. The minimum VSWR does not occur at the intrinsic impedance of canola oil (238 X), but continues to improve as Zo max is lowered to 50 X. The idea of matching to the intrinsic impedance of medium appears not to be correct, as noted similarly by Shlager et al. [20]. To maintain bandwidth at lower frequencies, values of Zo max between 50 and 150 X appear desirable. With Zo max ¼ 50 X, the width of the antenna at the aperture is 15 cm, which is too large to be practical for our prototype system. As a compromise between antenna size and performance in the 2 GHz range, 115 X is selected for Zo max. Next, the effect of aperture width (Y max ) is investigated. For several values of Zo max, Y max is varied between 24 and 38 mm. VSWR improved slightly at lower frequencies for larger Y max values, but the effect is not strong. Because of the dimension specifications, the value of Y max is set to 31.6 mm (quarterwavelength at 1.5 GHz). With Zo max ¼ 115 X and Y max ¼ 31.6 mm, the plate width is 60.4 mm at the aperture. With the aperture size fixed, a parametric study is performed with the various combinations of separation and impedance profiles [Y(x) and Zo(x), as shown in Figs. 2 and 3]. All designs have VSWR of less than 2 above 2.5 GHz and only subtle differences in VSWR are observed Near-Field Results. For the antenna designs with fixed apertures of 63.2 mm 60.4 mm and length of 80 mm, the radiated near fields are evaluated using the two pulses described in Table 1. We aim for a single beam with a circular cross-section when examined on boresight, as a circular beam is expected to provide a more compact tumor response in images. Table 2 shows the patterns for various combinations of impedance and separation profiles, indicating that the energy patterns for the two pulses are generally different. The best radiated energy patterns are obtained with the Y(x) ¼ exponential profile, and tend to be round, slightly oval or figure eight shaped. The rest of the designs produced poor radiated energy patterns. Fidelity values for a few of the TEM horn antenna designs at 3 cm are given in Table 3. When evaluating the radiated pulse fidelity, results are found to be consistently better for the Gaussian modulated sine pulse than the differentiated Gaussian pulse. The maximum fidelity tends to increase with distance from the antenna, however, the range of fidelity values found over the beamwidth often increases. Of the Y(x) ¼ exponential designs, the Zo(x) ¼ linear profile has slightly better fidelity results overall, though any differences may not be significant for practical application. The initial design targets an antenna length of 80 mm, however, it is of interest to determine whether similar performance is achieved with a more compact sensor. We reduce Y max to 25 mm and X max to 70 mm, and note limited impact on the performance of the antenna. The final design consists of Y max ¼ 25 mm, X max ¼ 70 mm, Zo max ¼ 115 X, Zo(x) ¼ linear, and Y(x) ¼ exponential. VSWR is below 2 from 2.1 to over 12 GHz. The half-energy beamwidth is found to be (Y axis/z axis) for the differentiated Gaussian pulse and for the Gaussian modulated sine pulse. At a distance of 3 cm, the minimum and maximum fidelity values for the differentiated Gaussian pulse are found to be 0.88 and 0.93, respectively. For the Gaussian modulated sine pulse, the corresponding values are 0.89 and EXPERIMENTAL RESULTS 3.1. Implementation The supporting structure to hold the balun and antenna arms in position is constructed from acrylic, polyvinyl chloride (PVC) pipe, nylon, and epoxy glue. These materials are chosen because of the similarity of their relative permittivities to that of oil (Table 4). Figure 5 shows the completed TEM horn antenna and balun structure. The acrylic parts are shaped using a milling machine. The metal arms are made from mm ( ) thick brass shim stock and glued in place with epoxy. Three nylon screws pinch the two large curved pieces of acrylic together at the feed point of the antenna. In this way, good contact is made with the copper at the output of the balun and no soldering is required. The balun is glued to the acrylic. A piece of threaded PVC piping surrounds an SMA connector at the input of the balun and is used for attachment to the microwave imaging system. Figure 4 Zo max VSWR results when varying impedance at the aperture, 3.2. Measurements The antenna is first evaluated by measuring reflection coefficient (equivalently S 11 ). Next, radiated fields are assessed with frequency domain measurements. Finally, reflections from small objects representing tumors are presented. DOI /mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May

5 TABLE 2 Generalized Radiated Energy Patterns Zo(x) Profile Y(x) Profile Linear Pulse Linear Exponential Circular Scaled Circular Circular Exponential Exponential mirrored TABLE 3 Fidelity Ranges in Half-Energy Beam for Several TEM Horn Designs at 3 cm From the Antenna Aperture Design Y(x) ¼ exponential and Zo(x) ¼ linear Y(x) ¼ linear and Zo(x) ¼ circular Differentiated Gaussian Pulse Fidelity Range Gaussian Modulated Sine Pulse The connected balun and antenna structures are immersed in a tank of canola oil and S 11 is measured. Figure 6 compares theoretical and measured S 11 for two implementations of the final design. There is good agreement between the two measured antennas to about 10 GHz and the goal of 10 db was achieved TABLE 4 Electrical Properties of Antenna Support Structure Materials [21] Material Frequency Relative Permittivity, e r Loss Acrylic (3 10 GHz) 2.6 tan d ¼ PVC (3 10 GHz) 2.8 tan d ¼ Nylon (3 GHz) tan d ¼ Epoxy glue (3 10 GHz) 3.0 tan d ¼ Canola oil (2 15 GHz) r ¼ S/m or tan d Substrate (10 GHz) 2.33 tan d ¼ Figure 5 Manufactured TEM horn antenna with balun. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com] 1168 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May 2010 DOI /mop

6 Figure 6 balun Simulated and measured S 11 for final TEM horn design with Figure 7 Measured radiation pattern for final TEM horn design at 4 GHz. [Color figure can be viewed in the online issue, which is available at Figure 8 Reflected pulses from tumor phantom located at 40 mm from the aperture on boresight. Note that two measurements have been offset in the vertical direction for clarity for frequencies of 1.65 GHz and above. Differences between measurements of the two antennas likely reflect small variations in the manufacturing process. Although the measured resonant points do not align well with the calculated values, the overall trend is in agreement. Differences between measurements and simulations are attributed to reflections from the SMA connector and the boundaries of the tank, which are not included in simulations. Approximations to the losses resulting from oil and metals are also incorporated into simulations. Near field measurements are taken at several frequencies and compared with simulated near field values at the same frequencies. A DASY system (SPEAG, Zurich, Switzerland) is used for measurements. The measurement probe is not calibrated for oil, so relative values are obtained. At 4 GHz, the measured pattern (over a planar grid) for the total electric field is oval in shape and has good symmetry (see Fig. 7). The measured half-power beamwidth is 56 mm by 38 mm at a distance of 3 cm from the antenna aperture, whereas the simulated half-power beamwidth at 4 GHz is found to be 60 mm by 40 mm at the same distance. Pulses reflected from a small phantom tumor placed at various locations in front of the antenna are measured next. The phantom [7] has total diameter of 8 mm, comprising a core of 6 mm diameter coated with an epoxy layer of 1 mm thickness. The core material has a permittivity of 44 and a conductivity of 7 S/m. The epoxy coating has a relative permittivity of 10 and a loss tangent of The epoxy is required to keep the core material from dispersing into the oil. The phantom tumor is attached to the end of an acrylic rod using epoxy. S 11 is first recorded (as a reference) without a tumor phantom present, but with an identically positioned acrylic rod. A second measurement is performed with the tumor present. The reference signal is then subtracted, and the result is converted to the time domain. Both pulses described in Table 1 are considered with their respective energies normalized to 1. Figure 8 shows the pulses reflected from the phantom tumor on boresight and positioned 40 mm from the aperture. The reflected pulses exhibit the expected time signature with minimal late time response. Simple time analysis of the results, whereby the time of the reflection from the antenna aperture is subtracted from the time of the tumor reflection, allows for approximate calculation of the tumor location. For the results in Figure 8, the tumor location is estimated as 40.3 mm with the differentiated Gaussian pulse and 39.2 mm with the Gaussian modulated sine pulse. Additional measurements are taken with the tumor phantom positioned at six other locations at the same distance of 40 mm from the aperture: 640 mm along the Z axis, 640 mm along the Y axis, and 620 mm along the Y axis. The 640 mm locations in Y correspond to an angle of 20, whereas 620 mm corresponds to 10. Table 5 gives the relative energy and fidelity values obtained. The energy is highest on boresight, as expected. The values found at 640 mm on the Z axis are larger than those found on the Y axis. This corresponds with the expected oval shape of the radiated energy beam. The fidelity is found to be slightly below the target of 0.90 but above 0.86 at all locations and for both pulses except at 640 mm on the Y axis for the differentiated Gaussian pulse. Analysis of the simulations indicates that the lower frequency components of this pulse are not radiated in this direction as effectively as the higher frequencies. The 620 mm locations on the Y axis exhibit increased energy and fidelity values for both pulses. The larger pulse energy at DOI /mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May

7 TABLE 5 Energy and Fidelity Values Obtained From Phantom Tumor Reflections Location Differentiated Gaussian Pulse Relative Energy Fidelity Relative Energy Gaussian Modulated Sine Pulse Fidelity Boresight at x ¼ 40 mm þ40 mm on Z axis mm on Z axis þ40 mm on Y axis mm on Y axis þ20 mm on Y axis mm on Y axis þ20 mm for the Gaussian modulated sine pulse shows that the radiated energy pattern may not be perfectly symmetric. Overall, these results demonstrate the detection of small objects and reflections that are consistent with expected radiated field patterns. 4. CONCLUSIONS This article presents a TEM horn antenna appropriate for use in near-field, radar-based microwave breast imaging. The PP design approach provides an antenna that operates over an ultra-wideband and has predictable and appropriate radiated field characteristics. Moreover, the performance specifications are achieved without the use of resistive loading. Good agreement is obtained between measured and simulated results, and the antenna demonstrates the capability to detect small objects representing tumors in a homogeneous background. Comparison of the TEM horn antenna with other candidate designs in the context of imaging is ongoing. REFERENCES 1. E.C. Fear, Microwave imaging of the breast, (invited), Technol Cancer Res Treat 4 (2005), M. Lazebnik, et al., A large-scale study of the ultrawideband microwave dielectric properties of normal, benign and malignant breast tissues obtained from cancer surgeries, Phys Med Biol 52 (2007), M. Klemm, et al., Radar-based breast cancer detection using a hemispherical antenna array Experimental results, IEEE Trans Antennas Propag 57 (2009), X. Li, et al., An overview of ultra-wideband microwave imaging via space-time beamforming for early-stage breast cancer detection, IEEE Antennas Propag Mag 47 (2005), H.M. Jafari, et al., A study of ultra-wideband antenna operating in a biological medium, IEEE Trans Antennas Propag 55 (2007), W.C. Khor and M.E Bialkowski, Investigations into an UWB microwave radar system for breast cancer detection, In: Proceedings of IEEE Antennas Propagation International Symposium, Honolulu, HI, 2007, pp J.M. Sill and E.C. Fear, Tissue sensing adaptive radar for breast cancer detection Experimental investigation of simple tumor models, IEEE Trans Microwave Theory Tech 53 (2005), M.H. Shenouda and E.C. Fear, Design of dielectric immersed tapered slotline antenna for radar-based microwave breast imaging, Microwave Opt Technol Lett 51 (2009), J. Bourqui, et al., Balanced antipodal Vivaldi antenna for breast cancer detection, In: Proceedings of Second European Conference Antennas Propagation, 2007, 5 pp. 10. R.T. Lee and G.S. Smith, A design study for the basic TEM horn antenna, IEEE Antennas Propag Mag 46 (2004), M. Kanda, The effects of resistive loading of TEM horns, IEEE Trans EMC 24 (1982), E.A. Theodorou, et al., Broadband pulse-optimised antenna, IEE Proc H 128 (1981), K. Chung, et al., Design of an ultrawide-band TEM horn antenna with a microstrip-type balun, IEEE Trans Antennas Propag 53 (2005), J.M. Sill and E. Fear, Tissue sensing adaptive radar for breast cancer detection: Study of immersion liquids, Electron Lett 41 (2005), T.P. Montoya and G.S. Smith, A study of pulse radiation from several broad-band loaded monopoles, IEEE Trans Antennas Propag 44 (1996), M. Campbell, et al., Ultra-wideband microstrip to parallel strip balun with constant characteristic impedance, In: EMTS 2007 International URSI Commission B Electromagnetic Theory Symposium, 2007, 4 pp. 17. C. Nguyen, et al., Ultra-wideband microstrip quasi-horn antenna, Electron Lett 37 (2001), R.P. Hecken, A near-optimum matching section without discontinuities, IEEE Trans Microwave Theory Tech 20 (1972), H.A. Wheeler, Transmission-line properties of parallel strips separated by a dielectric sheet, IEEE Trans Microwave Theory Tech 13 (1965), K.L. Shlager, et al., Accurate analysis of TEM horn antennas for pulse radiation, IEEE Trans Electromagn Compat 38 (1996), A.R. Von Hippel (Ed.), Dielectric material and applications, Wiley, New York, VC 2010 Wiley Periodicals, Inc. DUAL-BAND ANTENNA DESIGN USING GENETIC ALGORITHM-GENERATED TOPOLOGY Jae Hee Kim and Wee Sang Park Department of Electronic and Electrical Engineering, POSTECH, Pohang, South Korea; Corresponding author: wsp@postech.ac.kr Received 22 July 2009 ABSTRACT: This article represents an antenna topology selection for dual-band mobile phone using an optimization technique that applies a genetic algorithm (GA) in the early stage of the design. Commercial software (CST MWS) is used to predict the performance of the antenna. Then, the fitness function is evaluated by integrating the CST with the GA using visual basic language. After the antenna topology was selected, a dual-band antenna for GSM and DCS applications was designed by adding a parasitic line. VC 2010 Wiley Periodicals, Inc. Microwave Opt Technol Lett 52: , 2010; Published online in Wiley InterScience ( DOI / mop Key words: inverted F antenna; genetic algorithm; dual-band; mobile phone 1. INTRODUCTION Many types of antenna for mobile phones have been reported especially for small size and large bandwidth [1 3]. One of the most popular and widely used prototypes is the inverted F antenna (IFA) [4]. The IFA resonates a quarter-wavelength and it is easy to match the input impedance by adjusting the position of shoring stub, but the original IFA has only one resonance. Therefore, antenna variation or additional stubs are required to adopt the antenna for dual-band applications MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 52, No. 5 May 2010 DOI /mop