Microwave Reflection Tomographic Array for Damage Detection of Civil Structures

Transcription

1 Microwave Reflection Tomographic for Damage Detection of Civil Structures Yoo Jin Kim, Luis Jofre, Member, IEEE, Franco De Flaviis, Member, IEEE, and Maria Q. Feng Abstract-- Microwave tomographic imaging technolog using a bi-focusing operator has been developed in order to detect the internal voids/objects inside concrete structures. The imaging sstem consists of several clindrical or planar arra antennas for transmitting and receiving signals, and a numerical focusing operator is applied to the eternal signals both in transmitting and in receiving fields. An imaging algorithm using numerical focusing operator was developed, which allows the recover of a 2-dimensional object from its scattered field. Numerical simulation demonstrated that a sub-surface image can be successfull reconstructed b using the proposed tomographic imaging technolog. For the eperimental verification, a prototpe planar antenna arra was fabricated and tested on a concrete specimen. Inde Terms imaging, reflection tomograph, damage detection I. INTRODUCTION Nondestructive assessment of concrete structures, currentl, heavil relies on visual inspections, which apparentl have some limitations. Majorit of highwa bridges are concrete bridges, and invisible damage such as voids and cracks inside concrete and debonding between rebars and concrete caused b corrosions and earthquakes is of significant concern. In the previous work []-[4], the authors developed an electromagnetic imaging technolog using surface-focused microwave. This was developed to assess the bonding condition of FRP-jacketed concrete structures. The eperimental results successfull detected the area and the location of the voids. Accurate information about the void s depth, however, cannot be obtained easil because the correlation of the depth and the measured signal has to be eperimentall determined. Apparentl, it is not eas for all possible damage cases. In this stud, a sub-surface-focused microwave imaging technolog using transmitting and receiving arras, as shown in Fig.., is developed, in order to assess damage within a concrete structural element far awa from its surface and to obtain the depth information about damage. This technolog can construct a microwave image, showing the information under the surface, thus it is possible to detect the depth of damage in concrete. Furthermore, this technolog uses transmitting and receiving arras with several antennas and the focusing point can be quickl adjusted b software without moving the hardware, which makes it possible to quickl measure a large area involving man points. The proposed sub-surface imaging technolog uses an arrangement consisting of several clindrical or planar arra antennas for transmitting and receiving signals, and a numerical focusing operator is applied to the eternal signals both in transmitting and in receiving fields. First, this paper describes a numerical focusing procedure, which allows the recover of a 2-dimensional tomographic image of an object from its scattered field. Second, some numerical simulations for the stud of reconstruction parameters and the verification of reconstruction algorithm are presented. Finall, eperimental verification through the tests on the concrete blocks demonstrated that the sub-surface focused imaging technolog, using the proposed antenna arra, can successfull detect the voids and defects inside the concrete structure. Receiving s I rm I tn Transmitting s I r2 I t2 I r I t rf Receiver Focusing Operator on rf Transmitter Focusing Operator on rf Focused Point Concrete Structure Fig.. Use of microwave reflection arras to focus waves on subsurface point. II. RECONSTRUCTION ALGORITHM A. Analtical Formulation The measurement geometr, shown in Fig. 2, uses N n N m elements, N n forming a clindrical transmitting arra and N m forming a clindrical receiving arra. A N n N m measurement matri can be obtained as follows: for ever selected transmitting element, the receiving arra is scanned

2 obtaining an N m -measurement column, then the procedure is repeated for the N n elements of the transmitting arra. Due to the basic 2D characteristic of the geometr under stud, each element consists of a long vertical antenna arra with a uniform current distribution. In practical terms, the length of the vertical antenna arra has to be greater than the transversal dimension of the focusing area. The concrete was assumed to be homogeneous and the effects from aggregates size were not considered in this theoretical model. 0.5m 0.4m Matching Cushion rf = ( f, f ) Transmitting s r0 = ( 0, 0 ) rtn = ( tn, tn ) rrm = ( rm, rm ) Reconstruction Cell (6.5mm ) 0.2m~0.4m Transmitting s Receiving s Defect (Object) Reconstruction 0.25m εr = m rf 0.03m 0.0m r0 Nn=32 T n(rtn) Rm(rrm) Nm=32 Receiving s Fig. 2. Measurement geometr for two clindrical transmitting and receiving arras of 32 elements. Following the electromagnetic compensation principle, the illumination of an object induces an equivalent electric current distribution, J eq ( 0, 0,z 0 ), and this distribution makes an electromagnetic image of the object in image reconstruction [5]-[6]. The reconstruction algorithm forms ever image point b means of the snthesis of two focused arras (transmitting and receiving arras), i.e. all the elements of both arras are weighted b focusing operator so as to be focused on a unique object point. This can be achieved b numerical treatment of the measurement matri. The focusing operator can be obtained b taking an inverse of the field induced b a current line. It is well known that the electric fields of the infinite electric line source are proportional to a Hankel function of the second kind whose argument is proportional to the distance from the source to the observation point. Therefore, the incident field at r i ( i, i ) when focusing on ever transmitting point r tn ( tn, tn ) on the reconstructing point r f ( f, f ) can be epressed as E (, ) = E (, ) I H ( k r r ) (3) s rm rm i 0 0 obj 0 e rm 0 where I obj is a constant for ever object containing its electromagnetic macroscopic characteristics. When focusing back the received field at r rm ( rm, rm ) on the point of interest r f ( f, f ), the electromagnetic image of E f ( f, f ) at r f ( f, f ) can be epressed as N m E (, ) = E (, ) I (, ) (4) f f f s rm rm Rm f f m= where I Rm ( f, f ), the focusing operator, is given b I (, ) = Rm f f H 0 ( k r r ) e rm f Finall, the entire process can be grouped as follows: E E E st, R st, R2 st, Rm I R E E E I st, 2R st, 2R2 st, 2Rm R 2 E (, ) = [ I I I ] f f f T T2 Tn R R C R R E E E I stnr, stnr, 2 stnrm, Rm (6) B. Reconstruction Parameters The formulation derived in the previous section has been applied to the case of two clindrical arras of 32 antennas each, at a frequenc of 0.0GHz (λ 0 =3.0cm, λ e =.3cm). Figs. 3 and respectivel show the focusing intensit of transmitting and receiving signals in concrete when focused at the point of (ρ =0.20m, φ =0 ), which is (=0.20 m, =0m) in rectangular coordinates. The focusing intensit reaches the maimum value (32) at the focused point. This value is consistent with the number of transmitting and receiving antennas. (5) N n (, ) = (, ) ( ) i i i Tn f f 0 e tn i E I H k r r () n= where I Tn ( f, f ), the focusing operator, is given b I (, ) = Tn f f H 0 ( k r r ) e tn f and k e is wavenumber in concrete (ε r =5.3). Scattered field measured at r rm ( rm, rm ) of a defect (object) placed at r 0 ( 0, 0 ) is Fig. 3. Focusing intensit in concrete (focused point: ρ =0.20m, φ =0 ). Transmitting intensit. Receiving intensit. 2

3 In order to stud the focusing capabilit of the sstem at different distances and view angles, nine point-like objects were placed at the boundar and the central landmark point of the reconstructing 2D cross section. The results, as shown in Fig. 4., show good uniformit in the focusing intensit levels, which is normalized from 0 to, at the 9 points. The impulsive shape of the focusing intensit at the focused point suggests a ver good behavior of the reconstruction algorithm. Finall, simulations using numerical measurements were conducted in order to verif the resolution capabilities of the sstem. As shown in Fig. 5., the results demonstrate that the sstem, due to the use of bi-focusing (focusing both in the transmitting and receiving arras), is able to achieve a resolution in the order of the wavelength, both in transversal and in the longitudinal directions, in the dielectric medium. This clearl improves the resolution of a conventional sstem using mono-focusing (focusing the receiving arra onl) [7] Y (m) Y (m) X (m) X (m) Y (m) X (m) Fig. 5. images of two point-like objects (Resolution of the sstem in normalized units from 0 to ). Transverse distance of 0.03m. Longitudinal distance of 0.03m. III. ANTENNA ARRAY DESIGN Fig. 4. Image of nine point-like objects (Impulse response of the sstem). - plane image (normalized unit from 0 to ). 3-D image A. Design of Slot s Based on the electrical and geometrical parameters of the concrete specimen to be inspected, an illuminating frequenc of 5.2GHz was chosen as it represents a reasonable tradeoff between the signal attenuation and the image resolution. The resolution can be 2.5cm at 5.2GHz, which is obtained from the wavelength (λ e ) in concrete. Also, the 5.2 GHz frequenc was chosen as an adequate :2 or :4 scale model for the geometries and wavelengths of the real columns and also to work with dimensional tolerances that facilitate the laborator. A microstrip slot antenna was chosen in order to be directl attached to the concrete surface or the matching cush- 3

4 ion. The microstrip slot antenna has the advantage of being able to produce either bi-directional or unidirectional radiation patterns with a large bandwidth. The strip and slot combination offers an additional degree of freedom in the design of the microstrip antenna [8]-[9]. At last, the geometr of the slot antenna was determined so as to obtain a wide bandwidth at the resonance frequenc of 5.2GHz. Fig. 6. illustrates the design based on the considerations described above. 64 (8 8) slots for each arra and a dimension of 20cm 20cm were selected as the represent a reasonable tradeoff between the resolution and the reconstructed area covered b the antenna arra. 20 cm.25 cm Receiving 00 Ω 7 Ω Transmitting 00 Ω 2.5 cm λ/4 Reflector (Steel) λ/2 Slot 7.5 cm λ/4 Microstrip Feed 7 Ω 00 Ω 00 Ω Attach to Object to be Investigated 50 Ω Fig. 6. Planar rectangular microwave antenna arra. Conceptual design of planar rectangular slot antenna arra. Eight elements linear arra with corporate feed configuration. As shown in Fig. 6., the planar rectangular microwave antenna is composed of transmitting and receiving arras, each consisting of 8 8 slot antennas. Each of these two planar 8 8 arras consists of a parallel fed 8-element uniform vertical arra producing a tomographic focused slice perpendicular to the vertical ais of the structure, and an electronicall switched 8-element horizontal arra able to focus at a particular point inside the previous tomographic focused slice. The transmitting antenna arra focuses the illuminating fields on a particular point inside the volume of investigation and the receiving arra focuses the receiving beam on the same point. The antenna arra is a sandwich structure with two metallic grounded substrates separated b a light foam laer. The grounded substrate close to the target concrete structure contains the radiating halfwavelength (λ/2) slots in the eterior ground plane and the 00Ω microstrip feeding line on the interior side. The second grounded plane, quarter-wavelength (λ/4) apart from the slot plane, acts as a reflector in order to produce unidirectional radiation towards the volume of investigation. B. Measurement Results The fabricated antenna arra was placed on a concrete specimen with the slots facing the surface of the concrete, so that the wave radiates through the concrete. The reflection (S ii ) and transmission parameters (S ij ) were tested in order to investigate the matching condition and mutual coupling. Fig. 7. shows a good radiation performance at the illuminating frequenc, and that interference between the co-lateral elements is as low as -20 db. Magnitude (db) Magnitude (db) Magnitude (db) S ii of Transmitting s Frequenc (GHz) S jj of Receiving s Frequenc (GHz) 0 S ij Measurement -0 From 8 To Frequenc (GHz) (c) Fig. 7. Measurement results of 8 6 antenna arra. S ii measurement of transmitting arra. S jj measurement of receiving arra. (c) Eample of S ij measurement (S 89 ). 4

5 IV. NUMERICAL SIMULATION Numerical simulations were carried out to test the proposed algorithm for the image reconstruction of a concrete specimen with internal air voids and steel bars. The reconstruction parameters including effective focusing area were investigated. Totall 7 cases were modeled with 5.2GHz as an illuminating frequenc and the image of each case was reconstructed. Some of these numerical simulations were compared with the results with 0.0GHz as the illuminating frequenc. This comparison showed that the resolution can be improved b using the higher illuminating frequenc. For the simulation, planar rectangular antenna arras were used with 8 transmitting/receiving antennas for the case of 5.2GHz and 6 transmitting/receiving antennas for the case of 0.0GHz. For simplicit, onl a 2D rectangular geometr was considered for the investigation and onl Y polarization of the antenna was used. Etension to 3D can be easil obtained b using the spherical Green s function as a focusing operator instead of the Hankel function. A. Effective Focusing (EFA) Reconstruction parameters such as focusing intensit and impulse response of the sstem for the planar antenna arra were investigated again, because the geometr of the fabricated antenna arra and the frequenc range are different from the clindrical sstem used previousl. Based on the results of a parametric stud using point-like objects, the effective focusing area (EFA), in which the focusing operation is effective, was determined. In other words, the focusing intensit level and impulse response shape have a good uniformit in the effective focusing area. Also, the effective penetration depth is determined b the effective focusing area. The shape and size of the effective focusing area (penetration depth) strongl depend not onl on the electrical size of the measuring line, but also on the arrangement of the transmitting and receiving elements, which means that the EFA of a planar antenna arra differs from the one for circular antenna arra. Through a series of simulation using point-like objects, in which the focusing intensit level and shape of impulse response at different locations were compared, the effective focusing area was determined as the shape of a trapezoid, as represented in Fig m 0. m m 25 m 0.25 m B. Simulation Results Using5.2GHz Based on the reconstruction parameters, especiall the effective focusing area, numerical simulations were carried out. For the direct problem in which the antenna radiates a wave and receives the scattered field, a 3D structure simulator, CST Microwave Studio TM, was used for measuring the transmission parameters (S ij ) in each arra. Totall, seven geometries were modeled with a 6-slot antenna. In the simulation, the background material was modeled as concrete so as to radiate the wave through the concrete. Open boundar conditions were applied at the top and the bottom of the geometr in order to generate absorbing conditions. Using the S ij measurements from each arra, the measurement matri of Eqn. (6) was assembled and the electromagnetic image of E f ( f, f ) at r f ( f, f ) was obtained as presented previousl. All the objects used in the numerical simulation were located inside of the effective focusing area. The reconstructed images for each case are reported in Fig. 9 along with the model description and the reconstructed area. 5 m 0.2m 0. m Air Sphere (r = 0.0m) 5 mm Effective Focusing - Fig. 8. Effective focusing area of planar rectangular antenna arra 5

6 0. m m 5 m m m 5 mm 25 m m 0.25 m m m m m (c) (d) (e) m 0.0 m 0. m m m 0.0 m 0.0 m m (f) (g) Fig. 9. Simulation results with descriptions. Case. Case 2. (c) Case 3. (d) Case 4. (e) Case 5. (f) Case 6. (g) Case 7. In case, the air sphere was successfull reconstructed with eact location and size as represented in Fig. 9. The reconstructed area was limited to 5cm in -direction because the open boundar in CST Microwave Studio TM was set at the -ais, from which the distance to the antenna arra is 5cm. The reconstructed area in the -direction for the other cases was also determined b the location of an open boundar (0cm in the -direction). Cases 2, 3 and 4 are similar with the ones in the parameter investigations in Section II, in which the sstem was able to achieve resolution of the order of a wavelength in the dielectric medium. Cases 2 and 3 show the transverse and longitudinal resolution of the sstem, respectivel. The wavelength in concrete at 5.2GHz is about 2.5cm, which provides the resolution of the sstem of 2.5cm. As represented in Fig. 9, the locations of two steel bars separated with transverse distance of 2.5cm (λ e ) were detected apparentl, although the resolution was not sufficient enough to reconstruct the eact shapes of the bars. Fig. 9(c) shows the reconstructed image of two steel bars placed at a 6

7 longitudinal distance of 2.5cm (λ e ). In this case, the two bars could be successfull identified, but the eact shapes could not be reconstructed due to the lack of resolution. From the results of cases 2 and 3, it was demonstrated that the resolution of the sstem with planar antenna arra is of the order of a wavelength in concrete (2.5cm), which can identif the approimate shapes and locations. In case 4, si steel bars were placed at the boundaries of the effective focusing area for the reconstructing of a 2D cross section. The result, as shown in Fig. 9(d), demonstrated that the objects near the antenna arras are more clearl identified rather than the ones at the farther boundaries of effective focusing area, which is marked b black line, from the antenna arra. In cases 5 and 6, a square and a rectangular air block were placed inside the concrete material, respectivel. As shown in Fig. 9(e) and Fig. 9(f), the reconstructed image shows the eact location and the approimate size of the air blocks in both cases, although there was some noise. These two cases were investigated again at the higher frequenc in order to demonstrate the improvement of resolution. In case 7, a steel square bar and an air block were placed inside the concrete material. The difference in the dielectric constant between steel and concrete is much bigger than the one between air and concrete. For this reason the air void is more difficult to detect than the steel bar. As shown in Fig. 9(g), although the steel is located far from the antenna arra, the main features are correctl located, including size and position, while for the air void onl the location is detected. Using the results of the numerical simulation, the efficienc of the image reconstruction algorithm was tested through different cases, including steel, air voids, or combinations of the above. Within the effective focusing area, the location and size of objects were successfull detected. C. Simulation Results Using0.0GHz In the antenna design, the illuminating frequenc was determined as 5.2GHz as a good compromise between a set of different parameters, such as size, resolution, attenuation, and simplicit of prototpe. For that reason, the resolution of the sstem, as shown in the previous simulation results, was limited to 2.5cm and the eact shape of the object inside concrete could not be reconstructed. Better resolution, however, can be obtained b increasing the illuminating frequenc. In order to demonstrate the improvement of resolution using higher illuminating frequenc, cases 5 and case 6 were investigated again with 0.0GHz as an illuminating frequenc. For the numerical simulation, the planar rectangular antenna arras with 6 transmitting and 6 receiving antennas were used. The results in Fig. 0 show that the resolution of the reconstructed image can be improved b increasing the illuminating frequenc and that the eact shape and location of the voids can be reconstructed when using 0.0GHz (c) Fig. 0. Resolution improvement using higher frequenc. Case 5 using 5.2GHz. Case 5 using 0.0GHz. (c) Case 6 using 5.2GHz. (d) Case 6 using 0.0GHz. (d) V. EXPERIMENTAL IMPLEMENTATION A. Eperimental Setup and Calibration The effectiveness of the proposed sub-surface imaging technolog using the developed microwave antenna arra was investigated through a series of eperiments on a concrete panel and a concrete block. In the eperiment, a concrete panel with a hole at the center and a rectangular concrete block were used. The concrete panel was first used for calibration (to be described). Then, a sphere hole was placed on one of the faces of the concrete panel to simulate damage. Two tpes of artificial voids were generated inside a concrete block with a dimension of 30cm 30cm 30cm. The first involved a square Strofoam block of 2cm on the side and the other involved a rectangular Strofoam bar of 7

8 5cm 2cm 2cm. The Strofoam block was inserted into the concrete during the pouring of concrete with the distance of 3cm from the face of concrete to the face of Strofoam in both cases. The samples using steel bars in the air were also prepared. All the eperimental cases were described in Table I. The eperimental setup consisting of a network analzer, a switch bo, coaial cables and antenna arras is represented in Fig.. The network analzer was used to evaluate the transmitted portion of signal through the medium (S 2 ). The switch bo is a RF network capable of controlling multiple antennas in the arra and selecting them individuall to perform S 2 measurements. No TABLE I. Descriptions of eperimental cases Description Steel Sphere At Center of Concrete Panel Rectangular Steel Bar In the Air Rectangular Steel Bar In the Air Square Void (2cm 2cm) In Concrete Rectangular Void (2cm 5cm) In Concrete ( ) Eact Locations of Objects (, ) (cm) 4 cm 20 cm (0,0) 6 cm 20 cm (3,0) 6 cm 20 cm (3,-2) 8 cm 20 cm (4,0) 8 cm 20 cm (4,0) Transmitting Receiving d,9 d,6 Concrete Panel d8,0 Steel Plate Fig. 2. Description of calibration scheme Each transmission measurement, without all the effects from coaial cables and switch bo, which is from the transmitting arra to the receiving one, is proportional to the zero-order Hankel function of the second kind, H 0 (k e,conc d ij ), whose argument is proportional to the distance from the transmitting arra to the receiving one, d ij [0]. All the measurement effects from the coaial cables and the switch bo (calibration factor), therefore, can be calculated b dividing each transmission measurement of calibration (S ij (cal_mea) ) b the zero-order Hankel function of the second kind, and the calibrated transmission measurement (S ij (cal) ) can be obtained b dividing each transmission measurement (S ij (mea) ) b a calibration factor, as follows for the case of transmitting arra i and receiving arra j. where S ( cal ) ij d8,6 ( mea ) Sij = (7) Cal _ Factor Cal _ ( cal _ mea ) Sij factor = (8) ( ) H k d 0 e, conc ij The amplitudes of the calibrated measurements in concrete without defects are proportional to the distance between the transmitting and the receiving arras, and smmetric with respect to the diagonal in the measurement matri. The calibration factors of each transmission measurement at 5.2GHz were used for calculating the calibrated measurement matri in the eperiments. Fig.. Eperimental setup In order to remove the effect of wave reflections and loss in the coaial cables and/or in the switch bo from the measurement matri, calibration with respect to each transmission measurement (S ij (mea) ) is needed. For the calibration purpose, a concrete panel without defect and steel plate were used as represented in Fig. 2. B. Eperimental Results A continuous 5.2GHz sinusoidal EM wave was generated from the signal analzer and sent to the test specimen. Switch bo controls the location of transmitting and receiving arras from S,9 to S 8,6. Transmission measurements of each transmitting and receiving arra pair at 5.2GHz were assembled into a measurement matri. Each measured signal was divided b the calibration factor at 5.2GHz as described in previous section, and was assembled into the calibrated measurement matri to be multiplied b numerical focusing operators. The reconstructed images of a center cut in each case are plotted in Fig. 3 b the amplitude of electric current distributions. All the images contain steel objects or air voids. 8

9 In case, the location of the steel sphere is detected as well as the size, although there are edge effects from the small size of the concrete specimen, which add some noises to the result. Case 2 and case 3 are dealing with radiation into the air rather than concrete. As a result, the steel bars are detected b the location: reconstructed image in case 3 is shifted to the negative -direction with respect to the image in case 2, as epected. The shape of the steel bar could not be eactl reconstructed because of the lack of wave radiation into the air. Case 4 and case 5 are more realistic cases, which demonstrate air voids inside concrete block. Two-dimensional descriptions are illustrated in Fig. 3(d) and Fig. 3(e). The results from both cases show that the images of square and rectangular Strofoam s are successfull reconstructed in terms of size and location, as represented in Fig. 3(d) and Fig. 3(e). The resolution, which is 2.5cm at 5.2GHz, was not enough to reconstruct the eact size, which can be improved b using the higher frequenc, as eplained in the previous numerical simulation. m Steel Plate Steel Bar m 0.0 m Air (c) m m Concrete Block Concrete Panel Steel Sphere (r = 0.0m) - - m m m Steel Plate Air - - m Concrete Block (d) - m m m - m Steel Bar - m (e) Fig. 3. Eperimental results. Case. Case 2. (c) Case 3. (d) Case 4. (e) Case 5. 9

10 VI. CONCLUSIONS A sub-surface focused microwave imaging technolog was developed in this stud for detecting damages or objects inside concrete structures. The following conclusions can be drawn from simulation analsis and eperiments: () An imaging reconstruction algorithm using a bifocusing operator was formulated for the sub-surface imaging technolog and verified through a series of numerical simulations. The results showed that the algorithm gives a uniformit of focusing intensit level and the resolution in the order of the wavelength in a dielectric medium (concrete in this case) can be achieved b focusing both in transmitting and receiving arras. A slot antenna arra working in front of a dielectric medium was designed and fabricated. Through testing, it is found that the slot antenna arra is appropriate to use for damage detection of concrete structures in terms of its radiation performance and bandwidth. (3) The effective focusing area of a planar antenna arra is found to be a trapezoid, in which the focusing operator is effectivel working. (4) From the results of the numerical simulation, steel or air objects within the effective focusing area can be successfull detected in terms of the location and the size. It was also verified that the resolution can be improved b increasing the illuminating frequenc. (5) From the results of eperiments using a concrete panel and concrete block, it is demonstrated that the steel objects and air void can be successfull detected in terms of the location in depth and the approimate size. [] M. Q. Feng, F. De Flaviis, Y. J. Kim, and R. Diaz, Application of Electromagnetic Waves in Damage Detection of Concrete Structures, Proceedings of the International Smposium on Smart Structures and Materials, SPIE, pp. 8-26, Newport Beach, CA., March [2] M. Q. Feng, F. De Flaviis, and Y. J. Kim, Use of Microwaves for Damage Detection of Fiber Reinforced Polmer-Wrapped Concrete Structures, Journal of Engineering Mechanics, ASCE, Vol. 28, No. 2, pp , [3] Y. J. Kim, F. De Flaviis, L. Jofre, and M. Q. Feng, Microwave- Based NDE of FRP-Jacketed Concrete Structures, Proceeding of the 46 th International SAMPE Smposium, Vol. 46, Long Beach, CA, Ma [4] Y. J. Kim, F. De Flaviis, L. Jofre, and M. Q. Feng, Microwave Clindrical Reflection Imaging For Structural Damage Detection, Proceedings of IEEE AP-S/URSI Smposium, pp , Boston, MA, June 200. [5] I. M. Gironés, L. Jofre, M. Ferrando, M. De Los Rees, and J. Ch. Bolome, Microwave Imaging with Crossed Linear s, IEE Proceedings, Vol. 34, Pt. H, No. 3, pp , June 987. [6] A. Broquetas, J. Romeu, J. M. Rius, A. Elias-Fuste, A. Cardama, and L. Jofre, Clindrical Geometr: A Further Step in Active Microwave Tomograph, IEEE Transactions on Microwave Theor and Techniques, Vol. 39, pp , Ma 99. [7] S. X. Pan and A. C. Kak, A Computational Stud of Reconstruction Algorithms for Diffraction Tomograph: Interpolation vs. Filtered- Backpropagation, IEEE Trans. Acoust. Speech Signal Processing, Vol. ASSP-3, pp , October 983. [8] R. Garg, P. Bhartia, I. Bahl, and A. Ittipiboon, Microstrip Design Handbook, Artech House, Boston, 200. [9] C. Chen, W. E. McKinzie, III, and N. G. Aleopoulos, Stripeline- Fed Arbitraril Shaped Printed-Aperture s, IEEE Trans. s and Propagation, Vol. 45, No. 7, pp , Jul, 997 [0] C. A. Balanis, Advanced Engineering Electromagnetics, Wile, New York, 989. ACKNOWLEDGEMENT This work was supported b the National Science Foundation under Grants CMS and CMS REFERENCES 0

11 List of Captions TABLE I. Descriptions of eperimental cases Fig.. Use of microwave reflection arras to focus waves on sub-surface point. Fig. 2. Measurement geometr for two clindrical transmitting and receiving arras of 32 elements. Fig. 3. Focusing intensit (focused point: ρ =0.20m, φ =0 ). Transmitting intensit. Receiving intensit. Fig. 4. Image of nine point-like objects (Impulse response of the sstem). - plane image (normalized unit from 0 to ). 3-D image Fig. 5. images of two point-like objects (Resolution of the sstem in normalized units from 0 to ). Transverse distance of 0.03m. Longitudinal distance of 0.03m. Fig. 6. Planar rectangular microwave antenna arra. Conceptual design of planar rectangular slot antenna arra. Eight elements linear arra with corporate feed configuration. Fig. 7. Measurement results of 8 6 antenna arra. S ii measurement of transmitting arra. S jj measurement of receiving arra. (c) Eample of S ij measurement (S 89 ). Fig. 8. Effective focusing area of planar rectangular antenna arra Fig. 9. Simulation results with descriptions. Case. Case 2. (c) Case 3. (d) Case 4. (e) Case 5. (f) Case 6. (g) Case 7. Fig. 0. Resolution improvement using higher frequenc. Case 5 using 5.2GHz. Case 5 using 0.0GHz. (c) Case 6 using 5.2GHz. (d) Case 6 using 0.0GHz. Fig.. Eperimental setup Fig. 2. Description of calibration scheme Fig. 3. Eperimental results. Case. Case 2. (c) Case 3. (d) Case 4. (e) Case 5.