Signal Processing Methods for Long-Range UAV-SAR Focusing with Partially Unknown Trajectory

This work provides the signal processing workflow to focus Unmanned Aerial Vehicles (UAV) Synthetic Aperture Radar (SAR) images with partially unknown or corrupted trajectories. The processing chain is divided into two modular blocks. The former involves a novel and low-complexity autofocusing technique. This method applies geometric corrections directly to the nominal trajectory, getting rid of the assumption of a constant phase correction, that does not hold with highly variable squint and off-nadir angles, as for the case of UAV-borne SAR. The latter modular block concerns a Fast Factorised Back Projection (FFBP) based focusing scheme. In particular, for a given scenario the processor computes a priori the computational burden in order to define the most appropriate reference system and the degree of hierarchical merging to focus the scene at the minimum computational cost. In light of this, the proposed focusing algorithm is able to deal with a complex trajectory typical of UAVs and to focus a large image at high resolution. Two scenarios are considered. The former is a UAV-borne SAR experiment with a small aperture, a long range, and a wide area covered by a large antenna aperture. The latter is a proper UAV-borne SAR stripmap scenario. Here, the proposed focusing scheme performs better than the traditional FFBP. Finally, the results are supported by a numerical simulation to prove the effectiveness of the whole processing scheme.