A Two-Dimensional Range Ambiguity Suppression Method based on Blind Source Separation for Multichannel SAR Systems

Advanced multichannel spaceborne synthetic aperture radar (SAR) systems with multiple elevation beams allow to obtain high-resolution and wide-swath SAR images. However, the degradation of the SAR image quality due to range ambiguities is still an issue. In this paper, the range-ambiguity problem is analytically modeled and solved by employing a range-time and Doppler-frequency dependent mixing matrix which describes the linear superposition of multiple mutually range-ambiguous radar echoes. The overall framework can be regarded as a generalization of the cocktail party problem where each listener has to separate one speech signal out of a linear superposition of multiple voices. For the first time, the two-dimensional dependence of the antenna radiation pattern in the elevation and azimuth directions is considered by accounting for both the real-time beamsteering in elevation and the systematic variations of the mixing coefficients with Doppler frequency. A novel solution, based on higher-order blind source separation, is presented. The performance of the proposed method is numerically analyzed, with reference to an array-fed reflector antenna SAR system, by simulating a realistic acquisition scenario. To this aim, real SAR data and the actual antenna patterns of the Tandem-L mission proposal are considered.