Joint Source-Channel Coding for Deep-Space Image Transmission

The research described in this publication was carried out in part at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. The work by the University of Southern California and JPL was funded through the NASA/JPL Director’s Research and Development Fund (DRDF) and the Strategic University Research Partnerships (SURP) program. © 2011. All rights reserved. abstract. — A new coding scheme for image transmission over a noisy channel is proposed. Similar to the standard image encoders, the scheme includes a transform coding stage and scalar quantization. Differently, the proposed scheme considers the image as a set of parallel subsources where subsources are equal-size partitions of transform domain subbands. The scheme is based on joint source-channel coding in a twofold sense: 1) The rate allocations to different subsources are optimized considering both their individual operational rate-distortion curves and channel parameters; 2) instead of separate entropy coding and channel coding stages, the proposed scheme directly encodes quantization indices into channel codewords using a single linear encoder. Both of these properties are essential to increase robustness against channel noise and to decrease delay due to retransmissions. Comparisons are made with respect to concatenation of the state-of-the-art image coding and channel coding schemes used by the Jet Propulsion Laboratory using Mars Exploration Rover mission images.