SAR along-track formations: imaging and calibration

In recent years, the utilization of Synthetic Aperture Radar (SAR) satellites has become integral in various fields ranging from environmental monitoring to disaster management and defense intelligence. SAR satellites offer unparalleled capabilities in capturing high-resolution images of the Earth's surface, regardless of weather conditions or time of day. Among the numerous advancements in SAR technology, the concept of formation has emerged as a promising approach to enhance imaging capabilities, providing opportunities for improved spatial resolution, extended coverage, and enhanced data interpretation. The evolution of SAR satellites from traditional single-platform systems to multi-channel formations represents a paradigm shift in remote sensing methodologies. One of the most anticipated solutions is the Along-Track (AT) swarm, which aims to overcome the trade-off between resolution and coverage. The idea is that each receiver observes the same scene with a reduced Pulse Repetition Frequency (PRF), allowing a wide swath. The strong aliasing resulting from undersampling is resolved in post-processing by coherently combining data acquired by different receivers. The basic structure of the formation is one transmitter and multiple receivers, known as Single-Input-Multiple-Output (SIMO). Then, more complex configurations are being explored, such as the Multiple-Input-Multiple-Output (MIMO), where all satellites transmit simultaneously. The topic of AT SAR formations has been of great interest in the last few years, especially due to the maturity level of small satellites, such as CubeSats. However, there are still some open issues that must be solved before a successful launch of the new technology. So far, many works have considered ideal flying conditions in which the satellites follow the exact same orbit without any Cross-Track (XT) baseline. Moreover, the knowledge of acquisition parameters is commonly considered ideal. This thesis addresses reconstructing an unambiguous image from a set of undersampled SIMO acquisitions under realistic constraints. After analyzing state-of-the-art approaches, gaining from their strengths, and trying to overcome their pitfalls, I have developed an innovative method for reconstructing the desired high-resolution image. The technique performs the processing locally, introducing great flexibility and improving ambiguity suppression. By utilizing the Matching Pursuit (MP) algorithm, a sparse representation method, it is possible to investigate the data before the actual combination, allowing one to select the optimal strategy in a data-driven manner. Thus, it is possible to solve the issues related to wide antenna patterns and XT baselines and allow calibration of geometrical parameters before the actual data fusion. In developing this work, I was interested in using realistic datasets representing scenes with varying contrasts. The new methods were validated and tested with simulated data, using COSMO-SkyMed and Capella Space acquisition as references.