Data compression for space applications

The project on which this thesis is based to develop practical data compression techniques for the Soviet space mission Mars 94 - ELISMA project. Data compression techniques will be necessary because the amount of space-acquired data generated by the ELISMA project will be about five times the telemetry capacity available for this project.

Several new data compression algorithms have been proposed and are presented in this thesis: (1) Fast-adaptive Huffman coding. This algorithm splits the sample space into two trees for encoding, yielding fast-adaptive properties. These properties are significant when data are segmented so as to limit the error propagation. Also, the algorithm is able to offer a faster computation and requires smaller amount of memory than the conventional adaptive Huffman coding. (2) Adaptive cosine transform coding. The adaptive cosine transform and the bit allocation based on the marginal return technique are combined in this algorithm. The method of coefficients normalization has been found. Also, the error-correcting coding technique is included to make the transmitted data robust enough against the channel noise. (3) Adaptive Hadamard transform coding. This algorithm is identical to the adaptive cosine transform coding except in the transform core. Although this algorithm does not perform as well as the adaptive cosine transform coding in terms of the mean square error, it offers a faster transform algorithm. (4) Uniform vector quantization coding. This algorithm uniformly partitions vectors according to their activity and offers the simple procedure of codebook generation. This makes it suitable for real-time tasks. (5) Adaptive uniform vector quantization coding. This algorithm is based on the uniform vector quantization coding. It employs entropy coding techniques to obtain further compression efficiency, and adapts its codebook sizes to the data activity of the different subimages, resulting in fewer reconstructed errors. (6) Modified block truncation coding. This algorithm includes optimal quantization, differential and entropy coding, and error control.

Simulation results are given, the performance is analyzed, and the suitability of the new algorithms is discussed for the Mars 94 - ELISMA project.