Optimum signal processing for multichannel SAR: With application to high-resolution wide-swath imaging

A new method for processing multichannel synthetic aperture radar (SAR) data to achieve desirable image characteristics is presented. The method is optimal because it is derived by minimizing a mean-square-error cost function and generalizes current methods for high-resolution wide-swath SAR signal processing. The proposed method is easily implementable, can support a wide range in the pulse repetition frequency (PRF), including cases with highly nonuniform spatial sampling, and is robust against PRFs where current projection techniques fail, cases where the PRF is ideally suited to clutter suppression. Point spread functions for the proposed algorithms are presented, and the theory and simulations are further corroborated by results using multichannel SAR data measured by RADARSAT-2. We demonstrate that, if RADARSAT-2 were able to illuminate a 250-km swath (300 km ground range), then, conceptually, the new method would be able to process the highly nonuniformly sampled data to provide an extremely wide mode at approximately 5-m azimuth resolution.