Abstract
This thesis addresses the problem of stand-off ground-penetrating radar (GPR) antenna systems, which are principally used in demining applications. The detection of buried target is difficult due to fundamental limitations. Electromagnetic signals are highly attenuated in soil and the attenuation increases with frequency. Depth resolution on the target can only be achieved with broadband systems. As a consequence, the requirements on the antenna are severe. Antennas having a large bandwidth and a clean radiated response are needed to achieve depth resolution of the target. Moreover, stand-off systems such as the one used in demining applications require sufficient gain characteristics and mobility. The physics behind the GPR problem is complicated due to the large number of parameters affecting the radiation and reception of electromagnetic signals. Natural soils are often inhomogeneous and the close environment creates a considerable amount of clutter. Modelling techniques play an important role in understanding the electromagnetic wave propagation in natural ground. They are very useful in the design of antenna systems. The above issues are addressed in two different investigation routes. One route of investigation looks at the design of directive broadband antennas. The other route of investigation concerns the way of enhancing the modelling capabilities of GPR antenna problems. A novel stand-off antenna is proposed. The structure is resistively loaded and resembles a skeleton TEM horn. The design of the TEM horn antenna is undertaken by using an equivalent transmission line model. The structure is optimised to meet the design goals. Prototypes of the derived antenna are built and used within a real GPR system. GPR images of mines are obtained using the prototypes. Additionally, the agreement between measurements and predictions is good. Concerning the second route of investigation of this project, a hybrid method based on plane wave spectra interactions is introduced. The method enables field prediction of a GPR situation. It is based on splitting the complete problem into two sub-geometries. One geometry represents the antenna and the other represents the ground with buried targets. Each sub-geometry is analysed independently. Field prediction is undertaken by merging the two sub-geometries via a plane wave spectra formulation. The proposed hybrid method is validated with a dipole and a TEM horn antenna problem. Good agreement is found between the Method of Moment results and the hybrid method results.