Comparison of antenna dispersion and digital signal processing effects in ultrawideband Ground Penetrating Radar systems

摘要

In Ground Penetrating Radar it is of great interest to have a pulse width as narrow as possible to achieve best resolution capability. However, ultrawideband antennas may cause distortion to the radar signal due to dispersion, which is dependent on the frequency and bandwidth, as well as the direction of radiation. As dispersion causes an increase of the peak's width, the resolution capability is degraded. When frequency-based radar systems are utilized, such as stepped-frequency radar, a transformation of the recorded frequency-domain data to time domain is often required. There are different means of transformation at hand, the most popular being inverse Fourier Transform. Generally, such a transformation involves certain degrees of freedom in the processing of the data,which affect the appearance or even distort the resulting signal in time domain. In the same way, distortion may also be introduced to signals obtained by pulse radar, which is operating in time domain, when digital signal processing filters are applied to the time-domain data, modifying their frequency content. In contrast to dispersion these effects can be controlled — but not avoided. As an example, the typical sinc-shaped distortion of the time-domain signal after inverse Fourier Transform may be avoided by windowing of the frequency-domain data, a well-known basic technique in the digital signal processing domain. As dispersion, also windowing causes a broadening of the peak's width. However, this leads to the important question, to which degree both, dispersion and windowing, affect the time domain signal andwhich of them hasmore severe impact. In this paperwe investigate both effects. Wecompare dispersion of different ultrawideband Vivaldi, Bowtie and Loaded Bowtie antennas with the ideal (theoretical) non-dispersive one. And we compare these results with the distortion effects introduced to the radar signal by windowing frequency-domain data prior to inverse Fourier Transform.