A Consideration of Signal Processing for Spotlight Synthetic Aperture Radar (Ambiguity Functions)

Abstract

An analysis of spotlight mode synthetic aperture radar (SAR) imaging was carried out using a model based solely on the recognizable similarity it shares with tomographic reconstruction. Doppler effects were relegated to a pre-processing step. Results of this investigation included a determination of the importance of: (1) rectangular versus polar formatting of data for imaging, (2) limited time-bandwidth of chirp signals if stretch processing is employed, (3) residual Doppler effects, and (4) spherical wavefront curvature. Furthermore, the applicability of the convolution back-projection algorithm for imaging spotlight SAR data was established.

A second part of the thesis was concerned with the design of two-dimensional tapers, or windows, for Fourier transformation of data collected over arbitrarily shaped apertures. If the intent is to minimize sidelobe energy, the generalized prolate spheroidal eigen-equation of Slepian can be solved in a simple iterative manner using a digital computer to obtain the taper. Computer solutions for several aperture shapes of interest were presented.

A novel analysis of interpolation artifacts was given to explain the appearance of defocused, spurious targets which appear from digital processing of data which has been interpolated from a partial polar raster onto a rectangular raster. The polar raster arises, in the context of the SAR application, from the manner of data collection, whereas digital processing is facilitated using data sampled on a rectangular raster.

Finally, the bi-static SAR problem was examined in its relation to other forms of near-field imaging and direct parallels were drawn. With less restrictive geometries, the optimal processor was again seen to be the spatially varying matched filter.