A practical low-density coding scheme for binary dirty paper channel

Abstract

Recently, dirty-paper coding has been attracting a lot of attentions due to its connection with various problems such as data hiding, precoding for ISI channels and MIMO Gaussian broadcasting. One of the main challenges is to design codes with practical encoding and decoding algorithms. In this paper we propose a simple coding scheme for the binary dirty-paper channel -with the interference known non-causally to the encoder- by compounding low-density graphical structures. In our construction, the codewords of a channel code are both the bin indexes and parity-syndromes for a quantizer code. This novelty allows us to use reinforced belief propagation algorithm as a quantization algorithm for a LDPC code. Our construction has a natural nested and coset structure and achieves the capacity of dirty paper channel using random linear component codes and maximum- likelihood encoding/decoding algorithms. By selecting the constituent codes from the ensemble of LDPC codes, this scheme allows for efficient iterative (suboptimal) algorithms at both encoder and decoder.