Fixed Points and Stability of Density Evolution

Density evolution is a dynamic system in a space of probability distributions repre- senting the progress of iterative decoders in the infinite block length limit. In this paper we establish some basic results concering this process. In particular we show that the decoding threshold is equivalent to to appearance of non-trivial fixed point solutions to the density evolution equations. In the case of LDPC codes we prove the sufficiency of the previously published stability condition for stability of the δ∞ fixed point and slightly strengthen the necessity result.

[1]  Niclas Wiberg,et al.  Codes and Decoding on General Graphs , 1996 .

[2]  Rüdiger L. Urbanke,et al.  The capacity of low-density parity-check codes under message-passing decoding , 2001, IEEE Trans. Inf. Theory.

[3]  Laurent Decreusefond,et al.  On the error-correcting capabilities of cycle codes of graphs , 1994, Proceedings of 1994 IEEE International Symposium on Information Theory.

[4]  Rüdiger L. Urbanke,et al.  Exact thresholds and optimal codes for the binary-symmetric channel and Gallager's decoding algorithm A , 2000, IEEE Transactions on Information Theory.

[5]  Rüdiger L. Urbanke,et al.  Design of capacity-approaching irregular low-density parity-check codes , 2001, IEEE Trans. Inf. Theory.

[6]  Robert G. Gallager,et al.  Low-density parity-check codes , 1962, IRE Trans. Inf. Theory.

[7]  David J. C. MacKay,et al.  Good Codes Based on Very Sparse Matrices , 1995, IMACC.

[8]  A. Glavieux,et al.  Near Shannon limit error-correcting coding and decoding: Turbo-codes. 1 , 1993, Proceedings of ICC '93 - IEEE International Conference on Communications.

[9]  T. Richardson,et al.  Multi-Edge Type LDPC Codes , 2004 .