Abstract
We propose a joint source-channel decoding approach for multidimensional correlated source signals. A Markov random field (MRF) source model is used which exemplarily considers the residual spatial correlations in an image signal after source encoding. Furthermore, the MRF parameters are selected via an analysis based on extrinsic information transfer charts. Due to the link between MRFs and the Gibbs distribution, the resulting soft-input soft-output (SISO) source decoder can be implemented with very low complexity. We prove that the inclusion of a high-rate block code after the quantization stage allows the MRF-based decoder to yield the maximum average extrinsic information. When channel codes are used for additional error protection the MRF-based SISO source decoder can be used as the outer constituent decoder in an iterative source-channel decoding scheme. Considering an example of a simple image transmission system we show that iterative decoding can be successfully employed for recovering the image data, especially when the channel is heavily corrupted.
| Original language | English |
|---|---|
| Journal | IEEE Transactions on Signal Processing |
| Volume | 54 |
| Issue number | 10 |
| Pages (from-to) | 3688-3701 |
| Number of pages | 14 |
| ISSN | 1053-587X |
| DOIs | |
| Publication status | Published - 01.10.2006 |
Funding
Manuscript received November 23, 2004; revised October 26, 2005. The associate editor coordinating the review of this manuscript and approving it for publication was Dr. David J. Miller. This work was in part supported by the German Research Foundation (DFG) under grant KL1080/3-1. This paper was presented in part at the ITG Conference on Source and Channel Coding (SCC), Erlangen, Germany, January 2004, and at the IEEE Conference on Acoustics, Speech, and Signal Processing (ICASSP), Montreal, QC, Canada, May 2004.
