There is great desire to employ radio waves to sense objects visually obstructed behind walls or other non-metallic objects. Through-wall imaging (TWI) has a variety of potential applications including avalanche/earthquake rescue, mine detection, and reconnaissance, among others. Synthetic aperture radar (SAR) imaging approaches are typically implemented for TWI due to their inherent high cross-range resolution. However, the generation of SAR images requires movement of the array aperture along a prescribed path. This requirement introduces challenges, as each discrete SAR interrogation location along the path must be precisely known to within the phase resolution of the system in order to maintain coherency (i.e. create a focused image). This is extremely difficult for SAR platforms based upon ground or airborne vehicles, especially in hostile environments. Further, backscatter collection via SAR is inevitably slow (especially when compared to collection speeds of real apertures), and therefore cannot be utilized within applications which are time sensitive.
This dissertation introduces a portable UWB “snapshot” radar sensing system as an alternative to SAR. This system avoids SAR deficiencies by not requiring aperture movement along a path when collecting backscatter from multiple locations. Instead, a single finite planar antenna aperture employing time division multiple-input multiple-output (MIMO) processing is utilized. Ultimately, this allows high resolution image creation quickly and accurately. The subject array measures 55 x 46 cm in aperture, is light-weight (and therefore portable), and exhibits a large 0.9 to 2.3 GHz 3 dB bandwidth. Most importantly, it maintains an extremely low-profile while mounted over a ground plane. The UWB attribute is augmented via the introduction of virtual phase centers interwoven within actual array phase centers. The virtual phase centers provide an artificial increase in array element density and are realized through a unique MIMO array excitation process.
This dissertation focuses on the system design and implementation, followed by quick and efficient formulation of high resolution (specifically along the cross-range dimension) images. The general radar imaging problem, along with a variety of scattering models, is also discussed. Various linear imaging methods are considered along with an efficacious concept, known as coherence factor (CF) weighting, which dramatically enhances the target to clutter ratio (TCR) within an image. High resolution images are then forged via image fusion. That is, an image with enhanced clarity is generated by properly overlaying individual snapshot images from various angularly diversified scene aspects. In addition, several resource (i.e. processing time and computational memory) intense imaging methods are considered which enhance the inherently poor cross-range resolution associated with an image forged from a single snapshot interrogation location. These methods promote sparsity or region homogeneity via a one-norm minimization process.
Finally, a variety of images based upon simulated (via a high frequency scattering code) and experimental (via the proposed snapshot imaging radar) backscatter are presented to demonstrate the efficacy of these approaches. More precisely, several free-space and through-wall (cinderblock) scenes are utilized which encompass complex scattering environments containing metal spheres, trihedrals, plates, cylinders, and a variety of extraneous scatterers.