Datenbestand vom 19. Juli 2019
Tel: 089 / 66060798
Mo - Fr, 9 - 12 Uhr
Fax: 089 / 66060799
aktualisiert am 19. Juli 2019
978-3-8439-0353-0, Reihe Elektrotechnik
Versatile Compression of Multidimensional Spectral Data for Space Instruments
245 Seiten, Dissertation Technische Universität Braunschweig (2011), Softcover, A5
Scientific spacecrafts are equipped with a multitude of different instruments, providing data rates of up to several Gbit/s. This exceeds the maximum spacecraft telemetry rate by several orders of magnitude for most missions. Today, scalable on-chip architectures for FPGAs facilitate a highly integrated design of a common unit for a broad range of payload instruments. For such Data Processing Unit (DPU) a versatile data compression module is needed. Since standardized compression algorithms are available for images, it is of special interest whether these algorithms may also prove valuable for widely used spectrometer type of instruments.
Objective and subjective quality assessment criteria for a versatile data compression are compiled from typical reference instruments with different spectral sources, using artificial, simulated and real sensor data. Standardized algorithms are assessed and compared based on selected criteria. In contrast to other performance assessments, the computational complexity and the memory capacity needed are quantitatively analysed in detail since available energy and hardware resources are very limited on board spacecrafts. A significant reduction in complexity might be more attractive than a modest increase in compression quality. Ratio-complexity curves are plotted as complement to the commonly used ratio-distortion curves. Finally, both are combined to a distortion over complexity diagram.
From detailed verification matrices that summarize the different criteria at a comparable scale, the most versatile data compression algorithm is finally selected and implemented exemplarily in software also with necessary enhancements for one-dimensional spectral data. The wavelet-transform based image compression standard of the Consultative Committee for Space Data Systems (CCSDS-122) provides the best trade-off between performance and complexity.
Further enhancements and extensions for the CCSDS standard are assessed in detail, especially for processing of spectral data with limited resources available on board spacecrafts. Overlapped tiling is presented to transform unlimited line size data with limited memory resources in a sliding-window fashion. An embedded peak detection coding scheme is designed to be completely integrated in the coding structure of CCSDS algorithm and adapts to unknown spectral peaks automatically, delivering near-lossless compression.