Predictive Quantization for Onboard Data Reduction in Future SAR Systems

Abstract:
Next-generation satellite SAR systems will provide unprecedented high-resolution data products thanks to multi-static configurations, wider swath widths, and multiple polarizations. As a drawback, a considerable amount of onboard data is going to be generated, leading to onboard storage and downlink capacity challenges. In this context, the need for an onboard compression technique featuring low memory requirements and low complexity is of great interest, due to the strict hardware and cost constraints of satellite platforms. In this paper, an investigation of predictive coding applied to SAR systems is carried out. Specifically, two types of novel predictive encoders are presented: Dynamic Predictive BAQ (DP-BAQ) and Non-Causal Predictive BAQ (NC-PBAQ). A theoretical investigation is initially introduced and the results are finally reported by taking into account the trade-off between data reduction performance and computational requirements. DP-BAQ gain in terms of data reduction is directly dependent on both the performance requirements and the system constraints (e.g. the antenna size), and has already been proven to reach 20-25% of data reduction for the case of DLR’s Tandem-L mission proposal. The non-causality introduced by NC-PBAQ shows promising results with respect to the performance of DP-BAQ, at least from a theoretical point of view. Moreover, the computational effort required for DP-BAQ encoding is less demanding, therefore could be implemented in state-of-the-art spaceborne SAR systems.