Optimal Parameter Choice for Wyner-Ziv Coding of Laplacian Sources with Decodere Side-Information

source coding with side-information, Wyner-Ziv, Laplacian distribution, Gaussian distribution, channel codes, LDPC codes, Turbo codes, arithmetic codes A large number of practical coding scenarios deal with sources, for instance transform coefficients that can be well modeled as Laplacians. In regular practical coding of such sources, samples are often quantized by a family of uniform quantizers possibly with a deadzone, and then entropy coded. For the Wyner-Ziv coding problem when correlated sideinformation is available at the decoder, the side-information can be modeled as obtained by additive Gaussian or Laplacian noise on the source. This paper deals with optimal choice of parameters for practical coding of such sources in presence of side-information, using the same quantizer structure as in the regular codec, assuming that the variances of the source and additive noise are known. We first consider memoryless coding which may be the only option in some coding scenarios, and then follow up by considering coding using powerful channel codes with soft decoding that approach the Slepian Wolfe bound. We show that in the latter case, at practical block lengths and code complexities, not pure channel coding but a hybrid combination of source coding and channel coding provides optimal rate-distortion performance. A good understanding of the optimal parameter choice mechanism is essential for building practical codecs that can be used in a variety of scenarios.