Distributed conditional arithmetic coding based on adaptive source-symbol purging

Abstract

Distributed Source Coding (DSC) is a coding architecture that utilizes the correlation among multiple sources to compress them. It can be achieved by using entropy coding instead of channel codes. We propose a new DSC scheme based on arithmetic coding named “Distributed Conditional Arithmetic Coding based on Adaptive Source-symbol Purging”. Since the encoder only encodes a part of symbols in the source sequence, more compression can be obtained. For the source sequence with memory, the stronger the correlation strength within the source, the more compression gain obtained by this scheme. Compared with traditional arithmetic coding-based DSC schemes, this scheme is very effective when coding source sequences with strong internal correlation and short block length.