Accurate optimal doppler centroid estimation for SAR data

The paper addresses the problem of finding an optimal estimation of the Doppler centroid for Synthetic Aperture Radar (SAR) data. The original idea is to exploit a bandwidth much wider than the PRF, say 3-5 times, by selecting non-aliased point targets. Non-aliased band of natural and isolated targets with close-to-ideal features is evidenced by spotlight processing of strip map acquisitions and accurate Doppler centroid is estimated by means of a joint Maximum Likelihood (ML) estimator. Lower bound of the estimate is determined and results on both simulated and real X-band SAR data are shown.