A Greedy Approach for Sparse Angular Aperture Radar

We present a novel algorithm for radar imaging of point scatterers using a sparse number of spatially separated sensors. Such sparse sensing scenarios are prototypical of many applications wherein a limited number of sensors are distributed over a geographical area; or where environmental and/or systemic constraints enforce a sparse sampling of angular aperture. Our underlying assumption is that the image is sparse with respect to the Gabor basis set. We then introduce the concept of an orbit-viz. the locus of all projections made by a spatial basis-and formulate the radar imaging problem as that of sparsifying the number of orbits that comprise the radon measurements of the source. We demonstrate how our algorithm outperforms FFT-based and Compressive-sensing based reconstruction algorithms for point-scatterer images, describe relevant theoretical performance bounds of our algorithm, and point to future research arising from this work