Vis enkel innførsel

dc.contributor.authorEide, Egil S.nb_NO
dc.date.accessioned2014-12-19T11:17:25Z
dc.date.available2014-12-19T11:17:25Z
dc.date.created2000-09-27nb_NO
dc.date.issued2000nb_NO
dc.identifier122538nb_NO
dc.identifier.urihttp://hdl.handle.net/11250/228350
dc.description.abstractThis dissertation focuses on subsurface imaging of objects closely below the ground surface using ground penetrating radar (GPR). The main purpose is to detect and image man-made objects such as buried landmines or subsurface utility systems like cables and pipes. The thesis treats the various system aspects that have an influence on the performance of a GPR system for mapping of objects at shallow depths. High-resolution mapping at shallow depths requires radar equipment that has a high bandwidth and a high dynamic range for sensing weak scatterers closely below a strong reflecting surface. The focus in this thesis is to suggest and evaluate novel techniques for GPR such as new ultra-wideband radar hardware, wideband antennas and signal processing subsystems. Imaging through a lossy, dispersive and inhomogeneous soil represents a challenge due to the complicated wave propagation conditions in the soil. Through simple modeling, we describe the fundamental behavior of electromagnetic waves in the soil. The results from these simulations serve as a basis for designing an ultrawideband GPR system that can provide 3-D reflection data of targets at shallow to moderate depths. After reviewing the most common waveforms and techniques for GPR, this thesis reports the design and test of the RadioStar ultra-wideband radar. The radar is capable of operating in the frequency range 10 MHz — 3.4 GHz using digitally generated waveforms from an Arbitrary Waveform Generator. In this way we can test and evaluate a large number of waveforms and sequences of waveforms by simply changing software. For our GPR experiments, we have used the steppedfrequency waveform. Through the processing gain from the pulse compression of the frequency domain data, we obtain a system performance of 176 dB. The developed radar system has the capability of scanning up to 64 antenna pairs arranged in an array by utilizing two 64-way RF-switches. The performance of a GPR system depends strongly on the antenna which is one of the most critical system components. Large bandwidth and a well-defined impulse response are two of the most important characteristics of a GPR antenna. In addition, the isolation between the transmitter and receiver antenna should be as high as possible for optimum system performance and dynamic range. For shallow imaging, we have developed a novel ultra-wideband bow-tie transmit/receive antenna pair that can be viewed as a quasi monostatic antenna with diplexing properties. This is accomplished by mounting the antennas perpendicularly on a V-shaped ground plane and using microwave absorbers for loading. The antenna is shown to possess an excellent impulse response with low late-time ringing, as well as a well-controlled radiation pattern over a wide frequency range. The antenna pair is used as a building element in a switched antenna array consisting of 31 pairs for cross-track scanning. A fractal structure consisting of an interleaved pattern of large and small antennas covering different parts of the frequency spectrum is built and tested This structure gives an array that both fulfills the Nyquist spatial sampling criterion and covers a wide frequency range. Seismic migration techniques can easily be adapted to GPR imaging. The close relation between the properties of acoustic and electromagnetic waves allows us to employ well-documented methods from the seismic industry for the purpose of reconstructing 3-D subsurface images. In this thesis, we have implemented a 3-D version of the Stolt migration technique for processing the data from the antenna array. This method has its strength in terms of computing efficiency, but it assumes a constant wave velocity in the medium. This is not always the case in the soil, and for the typical GPR geometry, where the antennas are elevated above the ground, more elaborate methods will give better focusing of the images. These methods are suggested for future applications of the system. The radar system, including the antenna array and the image reconstruction processing, has been verified through 3-D imaging experiments of buried objects in a laboratory sandbox. The results of these experiments demonstrate that the array antenna has a potential for rapid acquisition of 3-D data in applications where the survey speed is crucial. Furthermore, the laboratory and field tests have documented several system artifacts that should be removed in the next system upgrade. Detection of metallic targets has been demonstrated both in the laboratory sandbox and through outdoor field tests. The detection of surrogate plastic mines proved to be much more difficult under outdoor conditions, but improved hardware and image reconstruction methods may still increase the detection capability. This thesis contains the following novel contributions in the field of Ground Penetrating Radar: • The development of an ultra-wideband radar system that uses digitally generated waveforms together with a wideband upconversion RF-architecture. • The development of a compact ultra-wideband transmit/receive antenna pair that consists of bow-tie monopole antennas arranged on a common ground plane. • The development of a switched ultra-wideband antenna array for rapid 3-D data acquisition. The concept of thinned spatial sampling allows the use of antenna elements of various sizes along the aperture. In this way, both the low-frequency coverage and the high-frequency spatial sampling requirements can be fulfilled at the same time. • The development of digital pre-processing algorithms together with implementation of a 3-D Stolt migration scheme for processing of data from the antenna array.nb_NO
dc.languageengnb_NO
dc.publisherFakultet for informasjonsteknologi, matematikk og elektroteknikknb_NO
dc.titleRadar imaging of small objects closely below the earth surfacenb_NO
dc.typeDoctoral thesisnb_NO
dc.source.pagenumber187nb_NO
dc.contributor.departmentNorges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi, matematikk og elektroteknikknb_NO
dc.description.degreedr.ing.nb_NO
dc.description.degreedr.ing.en_GB


Tilhørende fil(er)

Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel