Signal Reconstruction Algorithm for Azimuth Multichannel SAR System Based on a Multiobjective Optimization Model

This article establishes a multiobjective optimization model to suppress the azimuth ambiguity power and noise simultaneously in signal reconstruction for a multichannel synthetic aperture radar (SAR) system. This multiobjective optimization model extends the theory of multichannel signal processing for reconstructing the SAR signal from the aliased signals. Linear scalarization and a quadratically constrained method for the multiobjective optimization model are applied to obtain l1norm optimization, l2norm optimization, and quadratically constrained optimization, respectively, in signal reconstruction. Azimuth ghosts can intuitively reflect the effects of azimuth ambiguity on SAR images. The l1norm optimization solution leads to a minimum upper bound of azimuth ghosts. A lowest azimuth ambiguity-to-signal ratio (AASR) can be derived by l2norm optimization. By relaxing the constraint of total ambiguity power suppression, one can obtain a minimum noise level in the case of quadratically constrained optimization. The reconstruction performances of the multiobjective optimization model in terms of AASR, signal-to-noise ratio (SNR), and signal-to-ambiguity-plus-noise ratio (SANR) are investigated with respect to the pulse repetition frequency (PRF) and compared with other methods for a multichannel SAR system.