Compression of correlated binary sources using turbo codes

We propose the use of punctured turbo codes for compression of correlated binary sources. Compression is achieved because of puncturing. The resulting performance is close to the theoretical limit provided by the Slepian-Wolf (1973) theorem. No information about the correlation between sources is required in the encoding process. The proposed source decoder utilizes iterative schemes, and performs well even when the correlation between the sources is not known in the decoder, since it can be estimated jointly with the iterative decoding process.

I. Introduction

Let and be two correlated binary information sequences taking values . The correlation between the sources is introduced in the following way. •

Generate the symmetric i.i.d. sequence .

•

Define the sequence as , where -dicates modulus 2 addition and is a random variable, independent of , which takes value 1 with probability value 0 with probability .

Notice that this model is equivalent to a model in which a sequence S is observed by two independent sensors, producing the sequences and , with and , where and are independent random variables taking value 1 with probabilities and , and value 0 with probabilities and .