ETD

• automotive radar
• fmcw
• forward-looking sar
• motion compensation
• moving target imaging
• sar

High-resolution sensing is crucial for autonomous driving to ensure reliable perception in dynamic environments. While Forward-Looking Synthetic Aperture Radar (FL-SAR) improves the limited angular resolution of compact automotive radars, the presence of non-cooperative Moving Targets (MTs) violates the stationary scene assumption of classic SAR imaging, resulting in phase perturbations that become geometric distortion, such as smearing, displacement, and defocusing in the reconstructed image.

This thesis work examines frequency modulated continuous wave (FMCW), based millimeter wave (MM-WAVE), FL-SAR with Multiple Single-Input Multiple-Output (SIMO) antenna (FL-SIMO-SAR) imaging for automotive scenario, in order to propose a processing for detection and refocusing of moving targets.
The proposed architecture employs a two-stage detection strategy fusing full and sub-aperture Range-Doppler maps, combined with single-snapshot ESPRIT for high-resolution Direction of Arrival (DoA) estimation. Target velocity vectors are estimated using a multi-aperture Least Squares approach, enabling Signal-Clutter separation. Finally, the image is reconstructed using a motion-compensated Backprojection algorithm, enhanced by adaptive spatial partitioning and local autofocus to correct phase errors for each target individually.

The method was validated through numerical simulations of dynamic scenarios modeling the kinematics of pedestrians, bicycles, and cars. The results demonstrate the framework capability to detect many targets with a compact array, successfully distinguishing even those with overlapping Range-Doppler responses. The processing chain achieves accurate parameter estimation and generates a focused image where moving targets are correctly repositioned.