Robust Localization of Close-Range Radar Reflections

In order to allow for a computationally efficient estimation of radar reflections, one commonly assumes the reflecting target to be in the far-field of the sensing array, such that the impinging wavefront is modeled as a linear phase-shift along the array. This works well in most radar scenarios, but causes significant performance degradation for close-range radar systems, wherein the curvature of the impinging wavefront may not be neglected. In this work, we examine how the used far-field assumption limits the resulting performance, illustrating how the (misspecified) Cramér-Rao lower bound (CRLB), taking the model mismatch into account, significantly devi-ates from the true CRLB for close-range targets. We further introduce a robust estimator that allows for the waveform cur-vature, showing that this computationally efficient estimator allows for superior performance for close-range reflectors.