a through-dielectric ultrawideband (uwb)

7/28/2019 A Through-Dielectric Ultrawideband (UWB)

http://slidepdf.com/reader/full/a-through-dielectric-ultrawideband-uwb 1/6

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 60, NO. 11, NOVEMBER 2012 5495

A Through-Dielectric Ultrawideband (UWB)Switched-Antenna-Array Radar Imaging System

Gregory L. Charvat, Leo C. Kempel, Edward J. Rothwell,Christopher M. Coleman, and Eric L. Mokole

Abstract— A through-dielectric switched-antenna-array radar imagingsystem is shown that produces near real-time imagery of targets on the op-posite side of a lossy dielectric slab. This system operates at S-band, pro-vides a frame rate of 0.5 Hz, and operates at a stand-off range of 6 m orgreater. The antennaarray synthesizes 44 effective phase centersin a lineararray providing element-to-element spacing by time division multi-plexing the radar’s transmit and receive ports between 8 receive elementsand 13 transmit elements, producing 2D (range vs. cross-range) imagery of what is behind a slab. Laboratory measurements agree with simulations,the air-slab interface is range gated out of the image, and target scenes con-sisting of cylinders and soda cans are imaged through the slab. A 2D modelof a slab, a cylinder, and phase centers shows that blurring due to the slaband bistatic phase centers on the array is negligible when the radar sensoris located at stand-off ranges of 6 m or greater.

Index Terms— Synthetic aperture radar, real time systems, dielectric

slab, radar imaging, ultrawideband radar.

I. I NTRODUCTION

In this communication, the performance (simulated and measured)of an ultrawideband (UWB) radar for imaging through walls is dis-cussed. This system features a hardware architecture that yields a sig-ni cantly better acquisition time (2 s) than the 20 minutes of an earlier linear rail SAR [1], while retaining the same quality of imaging. Thisimprovement is effected by integrating the spatial frequency range-gated frequency modulated continuous wave (FMCW) architecture (asdescribed in [1]) with a switched antenna array.

The original rail SAR, capable of imaging targets through a

lossy dielectric slab, used an FMCW architecture that chirped from1.926–4.096 GHz in 2.5–10 ms and provided a range gate to rejectthe air-slab and slab-air scattering by signi cantly band-limitingthe IF stages. The modeling agreed with measurements for 15.2 cmdiameter cylinders located behind the slab. In fact, the rail SAR iscapable of imaging targets as small as soda cans through a 10 cm thick lossy dielectric slab (made of concrete) at a stand-off range of 9 m.Unfortunately, the radar required 20 minutes to acquire a completedata set across the length of the rail, which does not support practicalapplications.

The modi ed radar architecture of this communication providesnearly the same length aperture and resolution but reduces the acqui-sition time from 20 minutes to less than 2 seconds. This array consists

Manuscript received July 11, 2011; revisedFebruary15, 2012, May29, 2012;accepted June 18, 2012. Date of publication July 10, 2012; date of current ver-sion October 26, 2012. This work was supported by the Naval Research Lab-oratory with funding from the Of ce of Naval Research, ONR Code 30, theExpeditionary Maneuver Warfare & Combating Terrorism Department.

G. L. Charvat, L. C. Kempel, and E. J. Rothwell are with the Department of Electrical and Computer Engineering, Michigan State University, East Lansing,MI 48824 USA (e-mail: [email protected]; [email protected]; [email protected]).

C. Coleman is with Integrity Applications Incorporated, Chantilly, VA 20151USA (e-mail: [email protected]).

E. L. Mokole is with the Naval Research Laboratory, Washington, DC 20375USA (e-mail: [email protected]).

Color versions of one or more of the gures in this communication are avail-able online at http://ieeexplore.ieee.org.

Digital Object Identi er 10.1109/TAP.2012.2207663

of 8 receive elements and 13 transmit elements that are time-divi-sion-multiplexed (TDM) to the radar receive and transmit ports. Theradar acquires bi-static range pro les between a subset of all possible bi-static transmitter and receiver element combinations, providing a 44virtual element linear array that is 2.24 m in length with approximately

spacing, producing 2D (range vs. cross-range) imagery of what is behind a slab. Given the large array size, this system would be used

in applications where it could be mounted on a vehicle. Simulationsand measurements show that this array is effectively the same as arail SAR of equal length when imaging targets in free space [2]. Theair-slab-air model [1] is applied to the switched-antenna-array [2]to show that the bi-static combinations have a negligible eff ect onimagery through a lossy slab at stand-off ranges.

Similar arrays have been used for short-range free-space radar imaging [3]–[5] and for through-slab applications [6]–[15]. The arrayin this communication is different from previous work because itfacilitates element spacing and is interfaced to the range-gatedFMCW radar system [1], thereby enabling through-s lab imaging ata stand-off range of 6 m or greater. A more recent paper [16] uses aswitched array but is different than the work discussed here becauseits array facilitates 3D imaging but is sign i cantly smaller providinglower resolution and does not use the range-gated FMCW architecture.

A description of the radar imaging system is presented in Section IIand simulated and measured results of a cylinder through the lossy slabare compared in Section III.

II. R ADAR SYSTEM IMPLEMENTATION

Range-to-target information from a de-correlated FMCW radar signal is in the form of low-frequency beat tones representingranges-to-targets in the spatial frequency domain. The more distant thetarget the higher the spatial frequency [18]. It is possible to implementa range gate with a FMCW radar system by placing a band-pass lter (BPF) at the output of the video ampli er ltering out only the spatialfrequency bandwidth that corresponds to the range swath of interest,however this implementation is challenging because it is dif cult todesign practical high-circuit-Q BPFs at baseband. Higher performanceBPFs are available as IF communication lters which operate at highfrequencies and are found in two-way radios and communicationreceivers. Examples include crystal, ceramic, surface acoustic wave,and mechanical lters. The radar architecture discussed here [1] useshigh-Q IF lters as an analog range gate to reject spatial frequenciesdue to scattering from the air-slab boundary and pass those corre-sponding to the desired range swath behind the slab. This narrow IF bandwidth provides a short-duration spatial frequency range gate for long duration linear frequency modulated (LFM) chirp waveforms.For example, this radar is capable of chirping from 1.926 GHz to4.069 GHz in 2.5 ms, 5 ms, and 10 ms providing chirp rates of

857 GHz/s, 428 GHz/s, and 214 GHz/s, respectively. The IF lter used in this radar system has a center frequency of (10.7 MHz ) anda bandwidth of (7.5 kHz BW) providing range gates of 8.75 ns, 17.5ns, and 35 ns respectively.

Range gating is represented by the time-domain output of the videoampli er in Fig. 4 (neglecting amplitude terms) by [1]

(1)where is time in seconds, and

is theround-trip time from theradar to thecenter of the range

gate.0018-926X/$31.00 © 2012 IEEE



5496 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 60, NO. 11, NOVEMBER 2012

Fig. 1. Measured S21 magnitude response of the crystal lter used for the spa-tial frequency range gate.

Fig. 2. Time domain simulated scattered eld and spatial frequency lter re-sponse.

This spatial frequency range-gate is used to greatlyattenuate the spa-tial frequency response that corresponds to the slab, thereby reducingthe dynamic range requirement of the digitizer. This approach is ad-vantageous when imaging through lossy slabs because measured lossesthrough solid concrete, not including path loss, typically exceed themaximum dynamic range available from most digitizers of reasonablecost. For example, a 10 cm thick solid concrete slab was measured to provide 45 dB and a 20 cm thick solid concrete slab was measured to provide 90 dB two-way path loss at 3 GHz, stack-ups of various mate-rials further increase attenuation [17].

The spatial frequency range gate lterused in these experiments wasmeasured (Fig. 1). Using the model previously developed [1], a range pro le of a 10 cm thick solid concrete wall with a 7.6 cm radius perfectelectric conductor cylinder at a range of 6 m and 9.1 m respectivelywas simulated and the frequency response of the IF lter in time do-main was shifted thereby placing the wall in the stop-band (Fig. 2).The product of the lter and time domain response show that the rel-ative magnitude of the cylinder is greatly increased compared to thewall thus requiring less dynamic range for a through-wall measure-ment (Fig. 3). In addition to this, multi-path between the back-side of the wall and cylinder is also increased relative to the wall amplitude.The IF lter plays the major role in reducing the wall re ection, in thiscase, reducing the wall re ection by approximately 55 dB. Similar re-sults are realized when the wall thickness is increased to 20 cm.

This range-gating technique facilitates the use of long duration LFMwaveforms, thereby providing increasing average transmit power to be

Fig. 3. Simulated time domain response after spatial frequency crystal lter response is applied.

Fig. 4. Block diagram.

radiated at the target scene while using a low peak-power (and there-fore lower cost) transmitter. In addition to this, the long duration LFMwaveform eases data acquisition requirements so that inexpensive dig-itizers can be used. In this case, a 200 KSPS 16 bit digitizer was usedfor all experiments. The objective of this design was to provide high performance at a low cost using readily available components because




Fig. 5. Array layout (units in inches) showing antenna element locations (large circles), bi-static baselines, and effective mono-static phase centers (small circles).

applying low-cost design principles up-front to a proof of concept pro-totype will more easily facilitate itswidespread adaptation if it is shownto be useful.

This FMCW radar uses separate transmit and receive antenna el-ements to minimize transmitter-to-re ceiver coupling. Using separatetransmit and receive elements generally provides better isolation than

using a circulator with a single element because the circulator’s perfor-mance depends on the re ected power of the element it is fed to. For example, if the of the element were dB this would offer only20 dB of isolation. The measured S11 of elements used in this radar varies over the wide bandwidt h, but in general each element provides a better than dB S11 between 2–4 GHz, dipping as low as dBS11 between 3–4 GHz. The coupling between elements was measuredat the radar center frequency of 3 GHz to be dB from ANT19 toANT9 and from ANT19 to ANT11. It is for this reason that the best possible isolation is achieved by using separate transmit and receiveelements.

A conventional 2.24 m long array providing (5.08 cm) elementspacing for this radar at 3 GHz would require a total of 88 antenna ele-ments consisting of 44 separate pairs of transmit and receive elements.This large number of elements signi cantly increases the cost andcom- plexity of the system. Fortunately the practical application does notrequire instantaneous beamforming. By using switched antenna arraytechniques, fewer transmit and receive elementsare required to provide

spacing across the array, at the expense of time required to timedivision multi plex (TDM) the radar transmit and receive ports to theappropriate antenna elements. For the array described in this commu-nication, only8 receiveelements and 13 transmit elementsare required.

The transmit and receive ports of the FMCW radar are fed to twofan-out switch matrices (Fig. 4) that rout transmit signals to transmitelements (ANT1-13) and received signals from receive elements(ANT14-21) . The radar transmitter provides a peak output power of 10 mW, but the transmit fan-out switch matrix (SW1-4) has approxi-

mately 10 dB of insertion loss, resulting in a 1 mW peak power at thetransmit antenna elements. The receive fan-out switch matrix (SW5-7)also has insertion loss, therefore in order to preserve the noise gure,each receive element has an LNA mounted directly on it (LNA1-8).The resulting cascaded noise gure is estimated to be approximately3.3 dB.

The transmit and receive elements are physically separated into twosub-ar rays made up of ANT1-13 and ANT14-21. Each element is alinear tapered slot antenna etched on FR-4 substrate. The physical lo-cation of each element is represented by a large circle in Fig. 5, wherethe receive elements are on the top row and the transmit elements areon the bottom row.

Atany giventimethe transmitterand receiverports are routed to onlyone antenna pair. This pair represents a bi-static radar baseline. Only44 pairs (or baselines) are used to synthesize a aperture, whereeach of the pairs is represented by straight lines drawn between el-

Fig. 6. Radar system.

ements (Fig. 5). Each bi-static pair of transmit and receive elementsfunctions like a mono-static element approximately located half-wayalong the line drawn between [20], [21]. The middle row of small cir-cles in Fig. 5 indicates where the effective mono-static elements arelocated. The beamforming algorithm is a type of SAR imaging algo-rithm known as the range migration algorithm (RMA) [19]. This algo-rithm assumes a uniformly sampled aperture. These assumptionsare an approximation to the actual effective mono-static element posi-tion which, according to analysis [2], holds true for targets at practicalstand-off ranges of 4.5 m or greater. Due to the relatively small size of this SAR, the target scene of interest is within the main beam of eachelement.

The switched-antenna-array radar system is shown in Fig. 6 and thefront of thearray is shown in Fig. 7. A Labview graphical user interface(GUI) controls the switch matrices, pulses the transmitter, digitizes thede-chirped video signal, and computes then displays the SAR image, providing a pulse rate frequency (PRF) of approximately 22 Hz and animage rate of approximately 0.5 Hz. Although the near real-time con-

cept is demonstrated here, real-time video frame-rate imaging is pos-sible by using a more ef cient data acquisition pipeline, imaging al-gorithm, and GUI implementation, thereby providing a maximum pos-sible theoretical PRF of 400 Hz and image rate of 9 Hz.

Each antenna pair is calibrated to a 1.52 m tall 1.9 cm diameter copper pole exactly 3.35 m down range and centered to the middle of the array in free-space. This pole is treated as if it were a point target.Calibration coef cients are applied to each frame of data before theSAR image is processed [2].

Complete details of implementation are presented [22].

III. R ESULTS

Imaging through dielectric slabs is achieved for a variety of movingand stationary targets. In all scenarios, background subtraction wasused to reduce in-scene clutter (such as support beams and other in-frastructure) because all measurements were conducted within the au-




Fig. 7. Antenna array.

thor’s garage. It was shown in Fig. 3 that background subtraction isnot necessary to image through a concrete slab but in practice it helpsto reduce clutter. Any practical implementation of this system woulduse a high frame rate on the order of 10 Hz or greater. With a higher frame rate, frame-to-frame coherent background subtraction would re-veal the location of all moving targetswhile rejecting stationary clutter.Thereare many scenarios in which this would be valuable, for example,looking for human targets inside of a building. The results in the fol-lowing experiments will show that the location of the targets behindthe slab were clearly discernible.

To quantify the performance of this radar system, the same targetscene is both simulated and measured in the laboratory. This targetscene consists of a cylinder with a radius of 15.2 cm located downrange at 907 cm, cross range at 25 cm, and behind a slab located at6.1 m down range made of solid concrete with a thickness of 10 cm.A 2D simulation was performed using a combination of wave matrix

theory [23] and the scattering solution to a cylinder [24], details shown[1]. The simulated image of this target scene is shown in Fig. 8(a).For comparison, the measured image of this target scene is shown inFig. 8(b). The measured cylinder is about 20 dB above the noise oor,where the noise oor is clearly shown in the image. The point spreadfunctions of both simulated and measured imagery are similar, showingthe validity of the model except for the close-in cross-range sidelobeswhich are slightly elevated.

A. Sidelobes

Down-range and cross-range cuts are plotted (Fig. 9) of the simu-lated and measured images (Fig. 8). The measured and simulated im-ages compare well in down-range, except that the rst sidelobe closest

to the radar appears to be elevated in the measurement. Measured andsimulated cross-range main lobes are in close agreement, however the rst sidelobes are approximately 3 dB above the simulated result. Thedifferences are likely due to transmit leakage into adjacent elementsthrough one of several paths; the relatively low measured isolation of the antenna switches at 4 GHz (35 dB), feed line coupling because allfeed lines are bundled tightly into two harnesses (transmit and receive),and mutual coupling of antenna elements. All aforementioned affectswould shift the assumed bi-static phase centers thereby increasing thecross-range sidelobes.

B. Resolution

It was observed that theoretical best-possible resolution in free spaceis nearly identical to the measured range resolution when imagingthrough a lossy slab with the system placed at a stand-off range of ( m).

Fig. 8. Simulated (a) and measured (b) cylinder (radius cm)through a lossy slab (solid concrete with thickness cm).

The expected free space range resolution for an un-weighted pulsecompressed waveform depends on the chirp bandwidth calculatedusing [19]

(2)

where is the speed of light in free space. The expected down-rangeresolution based on chirp bandwidth is 7 cm. The down-range resolu-tion derived from the measurements is 9.4 cm. This result shows thatwhen imaging through a lossy slab of concrete at a stand-off range theswitched-antenna-array radar is performing close to the smallest theo-retical range resolution possible and that the slab has negligible effecton range resolution when the radar is located at a stand-off range.

Radar images are formed using a SAR imaging algorithm. Becauseof this the cross-range resolution depends on the length of the arrayand the location of the point target being measured relative to the frontof the array in both down-range and cross-range. Cross-range resolu-tion is calculated for the un-weighted case by [19]

(3)

where is the range to the point target, is the angle from thecenter of the aperture to the point target, and is the change in targetaspect angle from 0 to across the aperture. The expected cross-rangeresolution for the cylinder located at 907 cm down-range and 25 cmcross-range is 20.4 cm. The measured cross range is 12.6 cm, which




Fig. 9. Measured and simulated down-range (a) and cross-range (b) sidelobes.

Fig. 10. Large-target imagery example: A 30.5 cm diameter cylinder is imagedthrough a 10 cm thick lossy dielectric.

appears to be better than free-spacebut this is likely dueto element coupling (as mentioned above) or array calibration causing subtle cancella-tion close to the mainlobe. Regardlessof this, themeasured cross-rangesidelobes show that the switched-antenna-array radar is performingclose to the smallest theoretical cross-range resolution possible whenimaging through a lossy slab at a stand-off range.

Fig. 11. Small-target-imagery example: three 12 oz soda cans are imagedthrough a 10 cm thick lossy dielectric slab.

C. Large and Small Target Imagery

A larger RCS cylinder with a diameter of 30.5 cm was imagedthrough the slab (Fig. 10). The full extent of the radar image is clearlyshowing the target position. But more importantly the lossy slablocated 610 cm down-range in front of the cylinder is not shown,demonstrating the effectiveness of both the range-gated FMCW radar architecture and coherent background subtraction for eliminating theunwanted returns.

Three 12 oz aluminum soda cans were imaged through the slab(Fig. 11). Although the RCSs of these targets are signi cantly smaller than both the 15.2 cm and 30.5 cm diameter cylinders, the location of each is clearly shown, demonstrating this radar’s sensitivity.

IV. CONCLUSION

A rail SAR with a range-gated FMCW radar architecture was pre-viously shown to be effective at imaging through a lossy slab of solidconcrete at stand-off ranges exceeding 6 m. Unfortunately the dataacquisition time was approximately 20 minutes, which is far too slowfor practical applications. With this range-gated FMCW architecture,a phased array radar system was implemented using switched antennatechniques, providing near-equivalent performance to the rail SAR while reducing data acquisition time to 2 s providing a 0.5 Hz framerate. In addition, a 2D through-slab model was applied to the problemgeometry showing that modeled and measured through-slab results arein close agreement. Furthermore, the location of the slab is effectivelyrange gated and coherently subtracted out of all imagery, and low RCStargets such as soda cans can be imaged through a lossy slab made of concrete at stand-off ranges. In short, the concept of imaging throughlossy slabs using this switched array system was demonstrated, butthis system could be improved. The limiting factor causing the 0.5 Hzframe rate is the data acquisition pipeline which was not optimizedfor high performance, instead it was developed to rapidly provethese concepts. Future work should lead to development of a systemwith increased transmit power, reduced noise gure, should includeadditional frequency trade-space analysis, and demonstrate real-timevideo frame-rate imaging of 10 Hz or greater.

R EFERENCES

[1] G. L. Charvat, L. C. Kempel, E. J. Rothwell, C. Coleman, and E. L.Mokole, “A through-dielectric radar imaging system,” IEEE Trans. An-tennas Propag. , vol. 58, no. 8, pp. 2594–2603, Aug. 2010.




[2] G. L. Charvat, L. C. Kempel, E. J. Rothwell, C. Coleman, and E. L.Mokole, “An ultrawideband (UWB) switched-antenna-array radar imaging system,” presented at the Int. Symp. on Phased Array Systemsand Technology, Waltham, MA, Oct. 12–15, .

[3] P. J. F. Swart, J. Schier, A. J. van Gemund, W. F. van der Zwan, J.P. Karelse, G. L. Reijns, P. van Genderen, L. P. Ligthart, and H. T.Steenstra, “The Colorado multistatic FMCW radar system,” in Proc. IEEE Eur. Microwave Conf. , Oct. 1998, vol. 2, pp. 449–454.

[4] V. Katkovnik, M. S. Lee, and Y. H. Kim, “High-resolution signal pro-cessing for a switch antenna array FMCW radar with a single channelreceiver,” in Proc. IEEE Sensor Array and Multichannel Signal Pro-cessing , Aug. 4–6, 2002, pp. 543–547.

[5] M. S. Lee, V. Katkovnik, and Y. H. Kim, “System modeling and signal processing for a switch antenna array,” IEEE Trans. Signal Processing ,vol. 52, no. 6, pp. 1513–1523, Jun. 2004.

[6] M. A. Barnes, S. Nag, and T. Payment, “Covert situational awarenesswithhandheldultra-wideband short pulse radar,” in Proc. Radar Sensor Technology VI, SPIE , 2001, vol. 4374.

[7] S. Nag, M. A. Barnes, T. Payment, and G. W. Holladay, “An ultra-wideband through-wall radar for detecting the motion of people in realtime,”in Proc. Radar Sensor Technology and Data Visualization, SPIE ,2002, vol. 4744.

[8] A. R. Hunt, “Image formation through walls using a distributed radar sensor array,” in Proc. 32nd IEEE Applied Imagery Pattern Recogni-tion Workshop , Oct. 15–17, 2003, pp. 232–237.

[9] M. Mahfouz, A. Fathy, Y. Yang, E. E. Ali, and A. Badawi,“See-through-wall imaging using ultra wideband pulse systems,”in IEEE Proc. 34th Applied Imagery and Pattern Recognition Work- shop , Oct. 19–21, 2005.

[10] R. Benjamin, I. J. Craddock, E. McCutcheon, and R. Nilavalan,“Through-wall imaging using real-aperture radar,” in Proc. Sensors,and Command, Control, Communications, and Intelligence (C3I)Technologies for Homeland Security and Homeland Defense IV , May20, 2005, vol. 5778.

[11] Y. Yang, C. Zhang, S. Lin, and A. E. Fathy, “Development of an ultrawideband Vivaldi antenna array,” in Proc. IEEE Antennas and Propa- gation Society Int. Symp. , Jul. 2–8, 2005, vol. 1A, pp. 606–609.

[12] A. Berri and R. Daisy, “High-resolution through-wall imaging,” in Proc. Sensors, and Command, Control, Communications, and Intelli- gence (C3I) Technologies for Homeland Security and Homeland De- fense V , 2006, vol. 6201, p. 6201J.

[13] H. Burchett, “Advances in through wall radar for search, rescue andsecurity applications,” in Proc. Institution of Engineering and Tech-nology Conf. on Crime and Security , Jun. 13–14, 2006, pp. 511–525.

[14] W. Zhiguo, L. Xi, and F. Yuanchun, “Moving target position withthrough-wall radar,” in Proc. Int. Conf. on Radar , Oct. 2006.

[15] Y. Yang and A. Fathy, “Design and implementation of a low-cost real-time ultra-wide band see-through-wall imaging radar system,” in Proc. IEEE/MTT-S Int. Microwave Symp. , Jun. 3–8, 2007, pp. 1476–1470.

[16] K. E. Browne, R. J. Burkholder, and J. L. Volakis, “Through-wall op- portunistic sensing system utilizing a low-cost at-panel array,” IEEE Trans. Antennas Propag. , vol. 59, no. 3, pp. 859–868, Mar. 2011.

[17] P. R. Hirschler-Marchand, “Penetration Losses in Construction Mate-rials and Buildings,” MIT Lincoln Laboratory Project Rep. TrACC-1Rev. 1, Jul. 19, 2006.

[18] M. I. Skolnik , Radar Handbook , 3rd ed. New York, NY: McGraw-Hill, 2008.

[19] W. G. Carrara, R. S. Goodman, and R. M. Majewski , Spotlight Syn-thetic Aperture Radar Signal Processing Algorithms . Boston, MA:Artech House, 1995.

[20] N. J. Willis , Bista tic Radar . Raleigh, NC: Scitech Publishing, 1995.[21] S. Gabig, K. Wilson, P. Collins, J. Terzuoli, A. J. G. Nesti, and J. For-

tuny, “Validation of near- eld monostatic to bistatic equivalence the-orem,” in Proc. IEEE Int. Geosci. Remote Sensing Symp. , 2000, vol. 3, pp. 1012–1014.

[22] G. L. Charvat, “A Low-Power Radar Imaging System,” Ph.D. Dis-sertation, Dept. Elect. Computer Engrg., Michigan State Univ., EastLansing, MI, 2007.

[23] R. E. Collin , Field Theory of Guided Waves , 2nd ed. Piscataway, NJ:IEEE Press, 1991.

[24] C. A. Balanis , Advanced Engineering Electromagnetics . New York, NY: Wiley, 1989.

a through-dielectric ultrawideband (uwb)

Documents