Improving the performance of LDGM codes over indoor channels by exploiting the channel statistics

We propose a scheme for improving the decoding of linear codes with low-density generator matrix (LDGM codes) over indoor-like channels. The key idea is to model the channel of interest as a hidden Markov model, which is estimated and exploited in the decoding process. No a priori knowledge about the hidden Markov model is assumed at the decoder site. The proposed system clearly outperforms systems in which the channel statistics are not considered in the decoding process.