Onboard Quantization for Interferometric and Multichannel Synthetic Aperture Radar (SAR) Systems

Kurzfassung

This work focuses on onboard quantization for synthetic aperture radar (SAR) systems. For present and next-generation spaceborne SAR missions, an increasing volume of onboard data is going to be required, due to the employment of large bandwidths, multiple channels, and the imaging of large swath widths at fine spatial resolutions. This implies strong requirements in terms of onboard memory and downlink capacity, which are limited resources and often represent a bottleneck in the design of SAR missions. In this context, SAR raw data quantization represents an aspect of utmost importance for the design of spaceborne SAR missions, since the number of bits employed to digitize the recorded radar signal directly affects the performance of the SAR products and defines the total volume of data to be managed by the system. In this dissertation, the impact of quantization on SAR and interferometric (InSAR) performance has been investigated using experimental data acquired by the TanDEM-X mission, and a novel performance optimized block-adaptive quantization (PO-BAQ) has been introduced, which allows for a joint optimization of the data rate and of the interferometric performance. Furthermore, a new azimuth-switched quantization (ASQ) has been developed, which allows for the efficient implementation of non-integer quantization rates, hence granting higher flexibility of performance and resource allocation. In the second part of the thesis, the attention is then focused on the development of new methods for data volume reduction for future SAR systems, with particular focus on multichannel SAR and staggered SAR. The significant oversampling and correlation properties of the azimuth SAR raw signal has been exploited by applying a proper encoding and quantization to the SAR raw data, hence allowing for a considerable reduction of the data volume to be transmitted to the ground at the cost of a modest increase of onboard computational effort.