| Generating Code Representations Suitable for Belief Propagation Decoding |
| Citation: |
Yedidia, J.S.,; Chen, J.; Fossorier, M., "Generating Code Representations Suitable for Belief Propagation Decoding", Proceedings of the 40th Annual Allerton Conference on Communications, Control and Computing, October 2002 |
| Date: | September 2002 |
| MERL Contact: | Jonathan Yedidia |
We describe a method for transforming a generalized parity check (GPC) matrix representation of a block linear binary code into another GPC matrix representation of the same code. The output GPC matrix has some attractive features from the point of view of iterative decoding algorithms like belief propagation. In particular, the number of ones in each row is reduced, and there are no cycles of length four in the equivalent graphical representation of the code. We illustrate the method for toy examples including the Golay code, and also for a Euclidean Geometry (n=255,k=127) code. The decoding performance of the belief propagation algorithm using our new GPC matrices improves significantly when they are used on the binary erasure channel, but the results are mixed for the additive white Gaussian noise channel--for the Euclidean Geometry code, the performance still improves, but for the Golay code it deteriorates. |
|
