Author: Zhenyu Mao

Publisher: Stanford University

Published: 2011

Total Pages: 159

ISBN-13:

Source coding or lossy data compression has been extensively studied ever since Shan- non formally described the problem more than half of a century ago. Yet source coding schemes approaching the theoretical Shannon limit even on simple stationary sources such as IID Gaussian or IID uniform sources remain elusive. In this dissertation, both source coding and the related rate-constrained simulation problem are studied. Four necessary conditions for asymptotically optimal sliding-block or stationary codes for source coding and rate-constrained simulation are derived. The first two necessary conditions are proved for stationary ergodic sources while the remaining two necessary conditions are proved for IID sources. A new code design algorithm which attempts to satisfy all four necessary conditions is presented. The code structure has intuitive similarities to classic random coding arguments as well as to "fake process" methods and alphabet-constrained methods. Experimental results show that the new coding design algorithm provides comparable or superior performance with respect to previously published methods on common IID and autoregressive examples, often by significant margins. In many cases, such as IID Gaussian, IID uniform and discrete IID binary sources, the performance approaches the theoretical Shannon limit.