AI-Based Performance-Optimized Quantization for Future SAR Systems

Next generation SAR systems will bring a huge improvement in terms of SAR performance and coverage through the use of large bandwidths and digital beamforming techniques in combination with multiple acquisition channels. This will allow for overcoming the limitations imposed by conventional SAR imaging for the acquisition of wide swaths and, at the same time, of finer resolutions. The significant improvements that can be achieved in terms of performance are associated with the generation of large volumes of data, which, in turn, set harder requirements for the onboard memory and downlink capacity of the system.In this scenario, an efficient quantization of SAR raw data is of crucial importance, as it defines the amount of onboard memory and it directly affects the quality of the generated SAR products. In the Performance-Optimized BAQ (PO-BAQ), the basic concept of the original BAQ is further extended according to the approach proposed in [1], which represents a first attempt for an optimization of the resource allocation depending on the performance requirement defined for the final higher-level SAR/InSAR product. As quantization errors are significantly influenced by the local distribution of the SAR intensity, such an optimization is achieved by exploiting the a priori knowledge of the SAR backscatter statistics of the acquired scene.In this contribution we investigate the feasibility of deriving a Performance-Optimized bitrate map through a machine learning-based architecture, in view of a future possible onboard realization.