04558297
04558297
04558297
7, JULY 2008
1875
I. INTRODUCTION
N RECENT YEARS there has been increased interest in ultrawideband (UWB) technology for different applications such as communications, radar or localization. Nowadays, the capabilities of UWB systems are well known but some specic challenges have to be solved before expected performance levels can be reached. The study of a suitable radiating structure is an important task, especially for pulsed devices that need minimum signal distortion. Ideally, the antenna should faithfully replicate the transmitted pulse on reception. But in practice, the change of radiation characteristics with frequency alters the antenna response. Thus the UWB antenna is a much more important component than its narrowband counterpart. Besides standard specications (impedance matching, efciency, and integration), the UWB antenna must maintain almost constant radiation patterns and phase center position over a large bandwidth. Specic measurement procedures are also necessary to validate UWB antenna performance in terms of transfer functions [1], [2]. Characterization of an antennas behavior enables its ability to radiate pulses efciently depending on their spectrum
and bandwidth to be forecast. Measurements may be performed either in the frequency or the time domain to obtain the cascaded transfer function (or impulse response) of transmitting and receiving antennas in a particular direction. In order to characterize omnidirectional antennas, the procedure must be repeated in different directions for each radiation plane. In this paper, we present the design and optimization of a coplanar UWB printed antenna. Its shape is designed to provide impedance matching at 50 from 3.1 to 10.6 GHz, an omnidirectional radiation pattern, and a limited size at low cost. The rst part concerns the design of the triangular monopole printed , , on FR4 epoxy substrate ( ). We describe the insertion of slots to confer better UWB matching to the antenna. The ground plane shape is also optimized to obtain more constant radiation patterns versus frequency, and decrease directivity. The second part deals with the time domain measurements that have been performed to characterize the antennas. The measurement procedure is presented and the calculated transmission transfer functions are plotted versus angle and frequency in 3D format. This characterization in several planes and directions is used to calculate the distortion of a reference pulse radiated by the antenna. The dispersion of this pulse is calculated and compared for different antennas and different directions of propagation. II. ANTENNA DESIGN A. Antenna Structure The bow-tie antenna [3] is a widely-used structure that offers good radiation and matching properties with low prole. However, this balanced antenna presents a high input impedance ) depending on its are angle [4]. The level (typically proposed antenna [see Fig. 1(a)] is a triangular monopole with printed ground plane fed by a coplanar waveguide. Its input impedance is divided by two compared to the bow-tie. The are angle can be xed at 118 to confer 50 matching characteristics and the ground plane is designed to obtain a good impedance stability over the frequency band of interest [5]. B. Monopole Design This study was conducted to minimize the return loss and size of the initial antenna. An inverted triangular aperture, inspired by fractal technology, was cut in the monopole in order to create more resonances in the same frequency band and obtain better matching (Fig. 1). Compared to the fractal approach, the two upper triangles are not connected to the fed one; they can be considered as two parasitic elements coupled to an active triangular antenna. This aperture gives the fractal antenna the same resonant frequencies as the other two structures,
Manuscript received May 4, 2007; revised January 24, 2008. Published July 7, 2008 (projected). N. Fortino is with the Telecommunications and Networks Department, Institute of Technology and also with the Laboratoire dElectronique Antennes et Tlcommunications (LEAT) CNRS, UMR 6071, Sophia-Antipolis, France. J.-Y. Dauvignac and G. Kossiavas are with the Laboratoire dElectronique Antennes et Tlcommunications, Universit de Nice-Sophia Antipolis-UMR CNRS 6071, 06560 Valbonne, France. R. Staraj is with the Ecole Polytechnique de lUniversit de Nice-Sophia Antipolis, Laboratoire dElectronique Antennes et Tlcommunications, Universit de Nice-Sophia Antipolis-UMR CNRS 6071, 06560 Valbonne, France. Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TAP.2008.924704
1876
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 7, JULY 2008
Fig. 1. (a) Initial antenna structure, (b) antenna with fractal aperture, (c) active triangle of fractal antenna, and compared S .
but with reduced impedance variations. Fig. 1 shows the of the fractal antenna [see Fig. 1(b)], compared with the initial antenna [Fig. 1(a)] and with the active triangle alone [see below 10 Fig. 1(c)]. This leads to a matched antenna with dB from 3 to 7 GHz (Fig. 1). To compensate for the rise in the lower cutoff frequency, slots were created between the triangular elements of the fractal antenna by adding metallic extensions to the upper elements (Fig. 2). Thereby, capacitive coupling effects between the active and parasitic triangles were enhanced. The aim was to increase the electrical length of the antenna and consequently decrease the lower cutoff frequency. A study was conducted to measure of the slots on antenna matching the inuence of the length (Fig. 2). The results shown in Fig. 2 conrm that lengthening the slots from 1 to 5 mm increases antenna bandwidth at low frequencies. Return losses are also decreased at high frequency remain above 10 dB. The slots are not lengthened but beyond 6 mm: in this case low-frequency cutoff is not decreased while impedance matching is altered at high frequency. Previous studies show that the best results are obtained with . However, this is insufcient to match the antenna between 7 and 9 GHz. An additional study indicates that above 8 GHz current concentrates on parasitic elements around the aperof the aperture (Fig. 2) is thus optimized ture. The dimension to obtain better matching at high frequencies without changing antenna properties below 6 GHz. The 3 to 10 GHz frequency band can thus be covered by an antenna with and optimized aperture , resized overall to (Fig. 3). During the measurement, coaxial feeding cables were circled by ferrite chokes to avoid return current.
Fig. 3. S
C. Ground Plane Shape Secondly, we sought to obtain more stable radiation patterns and lower directivity across the frequency range. The antenna in Fig. 2 is omnidirectional in the plane but the diagrams plane. Furthermore, the raare more directive in the diation properties change with frequency in this plane between 3 and 10 GHz, especially around 6 GHz. This behavior can be expected to alter the impulse response of the antenna in particular directions. Consequently, it is important to obtain more uniform radiation patterns and to minimize their variations with frequency.
FORTINO et al.: DESIGN OPTIMIZATION OF UWB PRINTED ANTENNA FOR OMNIDIRECTIONAL PULSE RADIATION
1877
Fig. 5.
(a) Radiation pattern of the slotted antenna and (b) the antenna with double folded ground plane in the = 90 plane.
The ground plane shape is an important parameter for control plane. Different shapes of the radiation patterns in the were investigated to obtain the best compromise between diagram stability, impedance matching and size. The total length of the ground plane must remain constant to maintain good impedance matching at low frequencies. An optimized antenna with double folded arms is presented (Fig. 4). This structure is matched with from 3.2 to more than 11 GHz (Fig. 4). The radiation patterns are still omnidirecplane, and are less directive in the tional in the plane (Fig. 5). Typically, the antenna with standard ground plane presents a 3 dB aperture of 40 to 50 in this frequency band. In contrast, the gain of the antenna with double folded ground plane varies between 3 and 2 dB for apertures of 90 to 120 in this band. Moreover, the measured diagrams are almost constant from 3 to 8 GHz. The measurements conrm the simulated results (performed by ADS and HFSS software tools). All these design improvements lead to a size reduction of 70% compared with the initial structure (Table I). III. TIME DOMAIN MEASUREMENT A. Measurement Procedure Time domain measurements are also necessary to complement xed frequency performance assessments and to conrm the compatibility of antennas with pulse systems. These measurements were performed on an UWB Time Domain facility composed of a pulse generator (Picosecond 4050B) delivering
a 50 ps front, and a fast sampling scope (Tektronix CSA8000). The ltered signal of [212 GHz] bandwidth was applied to the transmitting antenna. In reception, the pulse was rst amplied by a LNA and then digitized by the scope (Fig. 6). These time domain measurements require a pre-calibrating technique in order to deconvolute the reception components. This was done in the frequency domain [6] using MATLAB. Different measurement procedures are possible in order to characterize the antenna under test. The rst method involves placing two identical antennas in the same position in transmission and reception. After elimination of the impact of the channel, we obtain the product of the transmit and receive transfer functions of this antenna in this direction. This method is accurate because only one measurement is necessary, but the antennas have to be precisely positioned. In addition, both have to be moved for measurements in another direction. For the second method, a pre-calibrating measurement is used to characterize a reference antenna in the working frequency band. This reference antenna, which must possess a phase center position that is stable with respect to frequency in order to minimize phase errors, is used to illuminate the antenna under test. In these conditions the accuracy of this measurement procedure is similar to that of the other, and it is easier to implement in different directions. Consequently, this method was chosen and a tapered slot antenna [7] was used as a transmitting reference antenna in this work. This antenna was kept xed while the position of the other was controlled by a turntable. Both antennas
1878
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 7, JULY 2008
were placed in an anechoic chamber to avoid multipath effects and simplify propagation channel modeling. All data processing was performed in the frequency domain using Fourier Transforms. The transfer function of the propagation channel was characterized using the Friis formula and a plane-wave propagation approach. The cascaded transfer functions of the transmit and receive antenna were extracted in all measurement directions. These functions were separated using the derivation effect described in various publications [8][10] between the transmit and receive transfer functions of an antenna. Therefore, a reference antenna can be calibrated and used to measure the antenna under test in different directions. It is im-
portant to note that these computed transfer functions take into account the antenna mismatch and efciency losses. Moreover, the use of a low-cross-polarization tapered-slot antenna as reference implies that only one component of the radiated eld is measured in these characterizations. B. Compared Results The three antennas described in Table I were measured in the and planes. Computed transmit transfer functions (magnitude and group delay) of the initial structure and the optimized antenna with double folded ground plane were plotted versus plane (Fig. 7). These frequency and radiation direction in the
FORTINO et al.: DESIGN OPTIMIZATION OF UWB PRINTED ANTENNA FOR OMNIDIRECTIONAL PULSE RADIATION
1879
Fig. 7.
(a) Transfer function of the initial antenna and (b) the optimized antenna with double folded ground plane in the xy plane.
antennas are omnidirectional with almost constant transfer funcplane, but have a more frequency-dependent diation in the gram in the plane. Attention was thus focused on the study of this latter plane. The results obtained for the initial antenna indicated that its plane [see distortion depended on radiation direction in the Fig. 7(a)]. This is due to the fact that its radiation patterns change with frequency and are slightly directive in this plane, with a maximum gain around 5 dB. Typically, this antenna illuminates under 6 GHz and presents more dipolethe direction like diagrams at higher frequencies. These changes in antenna radiation involve some phase distortions at certain frequencies. Note that zeros of radiation that can occur [for example at 3.5 in Fig. 7(a)] imply strong group-delay variGHz for ations that are partly due to phase measurement precision at low level. However, these uctuations alter the transmitted pulse shape in these directions. Consequently, study of these transmit transfer function diagrams indicates that the best results, in term in this plane. of pulse distortion, will be obtained for The calculated transmit transfer functions of the optimized antenna with double folded ground plane are presented in Fig. 7(b). Due to more constant radiation patterns, optimized
antenna transfer function and group delay are stable from 3 to and 50 [see Fig. 7(b)] in the 12 GHz for between plane. The antenna maximum gain in this plane is lower (3 dB) but the 3 dB aperture is larger. Moreover, this antenna avoids severe decrease of radiation in this plane. These results suggest that the improved antenna will be able to radiate pulses more plane. In the plane, where the radiation efciently in the patterns are almost constant and omnidirectional, this antenna exhibits stable transmit transfer functions varying between 3 and 3 dB (except a zero of radiation at 7.5 GHz in the direction), while the gain of the initial antenna is lower. C. Pulse Dispersion Computation of the dispersion that occurs when the antenna radiates a pulse signal is also of interest. The transmit transfer functions of the antennas were used to compute the radiated pulse in different directions when a reference signal was applied at the antenna input. This signal should present an UWB spectrum covering the antenna bandwidth and particularly the FCC mask from 3.1 to 10.6 GHz. It is shown in [12] that an acceptable approximation to a FCC mask compliant pulse can be obtained
1880
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 56, NO. 7, JULY 2008
Fig. 8. Reference pulse applied to the antenna and its power spectral density compared to FCC mask.
Fig. 9. Time to radiate 90% of pulse power in xy and xz planes for different antenna designs.
with a Gaussian second derivative. This pulse is represented in the time domain by:
(1) with xed at 37.5 ps to cover the desired frequency band. This signal and its spectrum are represented in Fig. 8. The pulse bandwidth is slightly larger than desired, but it is nevertheless an adequate criterion for estimation of antenna performance in the time domain. The radiated pulses were computed in all directions ( and planes) using the transfer functions of the transmitting antennas. To collect the large amount of data and compare the antenna behavior in a concise format, the time required to radiate 90% of the pulse power in each direction was chosen as the disand planes in persion criterion. Comparative results in the each characterized direction (Table I) are presented for the three antennas in Fig. 9. This gure shows antenna dispersion versus direction in comparison with the reference pulse applied at the antenna input. This pulse contains 90% of power in 156 ps. plane indicates that its Study of the initial antenna in the , with 90% of pulse power dispersion is minimal for reached in 250 ps. Results in this plane conrm that this andirections, due to its directivity. tenna is inefcient for
As predicted from the transfer functions in the plane, the performance of the antenna in this plane is quite constant due to its direction. omnidirectional characteristics, except in the The insertion of the triangular aperture and slots alters the plane antennas behavior but it is more dispersive only in the . Fig. 9 shows that in other directions the antenna performs as well as or better than the initial design. However, these structures are still quite dispersive in the principal planes. The antenna with double folded ground plane is optimized to obtain more constant radiation pattern versus frequency and consequently decrease pulse distortion. The results shown in Fig. 9 validate these changes: the antenna radiates with low disfrom to 130 in the plane, persion plane. Beyond these directions, and in a 60 aperture in the the antenna presents limited pulse spreading and radiates 90% of the pulse power in less than 300 ps. This antenna is thus less dispersive than the initial one over a larger beamwidth. IV. CONCLUSION A CPW-fed UWB triangular antenna matched to 50 from 3 to 11 GHz is presented. Geometrical modications to reduce return losses and size are described. Special attention is given to radiation pattern stability and a specic ground-plane shape is proposed. A time domain characterization procedure is used to verify experimentally the effect of antenna improvements on
FORTINO et al.: DESIGN OPTIMIZATION OF UWB PRINTED ANTENNA FOR OMNIDIRECTIONAL PULSE RADIATION
1881
phase distortion. Transmit and receive antenna transfer functions are separated using the derivation effect. These characterizations allow us to compute the radiated pulses when a reference synthesized pulse is applied to the input of the transmitting antenna. Dispersion for the different antennas is also compared, showing that antenna design modications make it possible to decrease pulse distortion. ACKNOWLEDGMENT The authors would like to thank C. Delaveaud, J. Keignart, and N. Daniele for their fruitful collaboration and for providing the Time Domain Facility of CEA-LETI that was of fundamental importance in this work. REFERENCES
[1] A. H. Mohammadian, A. Rajkotia, and S. S. Soliman, Characterization of UWB transmit-receive antenna system, in Proc. IEEE Conf. on Ultra Wideband Systems and Technologies, Reston, VA, 2003, pp. 157161. [2] X. Qing, Z. N. Chen, and M. Y. W. Chia, Characterization of ultrawideband antennas using transfer functions, Radio Sci., vol. 41, 2006. [3] K. L. Shlager, G. S. Smith, and J. G. Maloney, Optimization of Bow-tie antennas for pulse radiation, IEEE Trans. Antennas Propag., vol. 42, no. 7, pp. 975982, 1994. [4] G. H. Brown and O. M. Woodward, Experimentally determined radiation characteristics of conical and triangular antennas, RCA Review, pp. 425452, Dec. 1952. [5] N. Fortino, G. Kossiavas, J.-Y. Dauvignac, and R. Staraj, Novel antennas for ultra-wideband communications, Microw. Opt. Technol. Lett., vol. 41, no. 3, pp. 166169, May 2004. [6] B. Denis and J. Keignart, Post-processing framework for enhanced UWB channel modeling from band-limited measurements, in Proc. IEEE UWBST 2003, Reston, Nov. 2003, pp. 260264. [7] E. Guillanton, J.-Y. Dauvignac, C. Pichot, and J. Cashman, A new design tapered slot antenna for ultrawide-band communications, Microw. Opt. Technol. Lett., vol. 19, no. 4, pp. 286289, 1998. [8] C. E. Baum, General properties of antennas, Sensor and Simulation NotesNote 330, Jul. 23, 1991. [9] J. Kunish and J. Pamp, Consideration regarding the correlation between UWB antenna transmit and receive responses, presented at the Proc. URSI Int. Symp. on Electromagnetic Theory, Pisa, Italy, May 2327, 2004. [10] A. Sibille, About the rle of antennas in UWB impulse radio, in Proc. Euro COST, Athens, Greece, Jan. 2628, 2004. [11] H. G. Schantz, Introduction to ultra-wideband antennas, presented at the Proc. IEEE UWBST, Reston, VA, Nov. 1619, 2003. [12] G. Lu, P. Spasojevic, and L. Greenstein, Antenna and pulse design for meeting UWB spectrum density requirements, presented at the Proc. UWBST, Reston, VA, Nov. 1619, 2003. Nicolas Fortino was born in Nice, France. He received the D.E.A. degree in propagation, remote sensing, and telecommunications and the Ph.D. degree in electrical engineering from the University of Nice, Sophia Antipolis (UNSA), France, in 2002 and 2006, respectively. Since 2007, he has been an Assistant Professor at the Telecommunications and Networks Department, Institute of Technology, UNSA. He also with the Laboratoire dElectronique, Antennes et Tlcommunications, CNRS-UMR, Sophia Antipolis. His research interests include design and characterization of ultra wide band antennas, planar UWB array, but also time and frequency domain measurements for UWB detection and imaging microwave.
Jean-Yves Dauvignac is currently a Professor at the University of Nice, Sophia Antipolis (UNSA), France. During his Ph.D. studies (1993), his research activities were concerned the modelization of microwave antennas using nite-surface-element-method to solve integral equations of the electromagnetic eld. Since 1996, he is involved in microwave imaging for the detection of buried objects and in the study of dielectric resonator antennas. He developed a new class of UWB antennas (ETS Antennas) in planar printed technology for GPR and radar synthetic impulse microwave imaging system (SIMIS) for surface penetrating radar applications. During this period, he has also contributed in several campaigns of measurements road survey, pipes and mines detection in collaboration with LCPC and ONERA. In 2002, he was involved in the design of UWB antennas for telecommunications applications and the measurements of UWB antennas in time domain. More recently he has worked on the design of high gain antennas for radar systems in millimeter waves, and he continues his research activities on subsurface radar systems, UWB antennas for telecommunications and high gain millimeter-waves-antenna. Since 2000, he has also worked at the Laboratoire dElectronique Antennes et Tlcommunications (LEAT), CNRS, Sophia-Antipolis, where he is the Head of the Team High gain antennas and UWB detection and imaging microwave and millimeter-wave systems.
Georges Kossiavas received the Ph.D. degree in electronics from the University of Limoges, Limoges, France, in 1981. He is currently a Professor of electrical engineering and the Director of the Master Telecommunications RF and Microelectronic (TRFM) at the University of Nice, Sophia-Antipolis, France. He also works at the Laboratoire dElectronique Antennes et Tlcommunications (LEAT) CNRS, Sophia-Antipolis. His research interests are in microstrip antennas, especially large bandwidth and high purity polarisation radiating elements, compact microstrip multiband antennas for the mobile communications systems and design optimization of compact printed UWB antennas for lower pulse distortion.
Robert Staraj received the Ph.D. degree in electrical engineering from the University of Nice-Sophia Antipolis, France, in 1992. In 1993, he integrated the electrical engineering school Ecole Suprieure dIngnieurs de Nice-Sophia Antipolis (ESINSA) and the Laboratoire Electronique, Antennes et Tlcommunications-CNRS UMR 6071 (LEAT) as Assistant Professor. Since 2003, he is Professor in the new poly-technical engineering school PolytechNice-Sophia (Ecole Polytechnique de lUniversit de Nice-Sophia Antipolis). In charge of the research group Integrated antennas and Active antennas of the LEAT, his research focuses on printed antennas, microstrip planar and conformal arrays, active, integrated or auto-adaptative antennas, miniature antennas for wireless communications, ultra wide band radiating elements, RFID and MIMO systems.