A GPU Implementation of a MAP Decoder for Synchronization Error Correcting Codes

In this paper we present a parallel implementation of a MAP decoder for synchronization error correcting codes. For a modest implementation effort, we demonstrate a considerable decoding speedup, up to two orders of magnitude even on consumer GPUs. This enables the analysis of much larger codes and worse channel conditions than previously possible, and makes applications of such codes feasible for software implementations.

[1]  Vahid Tarokh,et al.  A survey of error-correcting codes for channels with symbol synchronization errors , 2010, IEEE Communications Surveys & Tutorials.

[2]  David J. C. MacKay,et al.  Reliable communication over channels with insertions, deletions, and substitutions , 2001, IEEE Trans. Inf. Theory.

[3]  Joseph R. Cavallaro,et al.  Implementation of a High Throughput 3GPP Turbo Decoder on GPU , 2011, J. Signal Process. Syst..

[4]  Jun Li,et al.  The fast evaluation of hidden Markov models on GPU , 2009, 2009 IEEE International Conference on Intelligent Computing and Intelligent Systems.

[5]  Marc André Armand,et al.  Interleaved LDPC codes, reduced-complexity inner decoder and an iterative decoder for the Davey-MacKay construction , 2011, 2011 IEEE International Symposium on Information Theory Proceedings.

[6]  Johann A. Briffa,et al.  An Improved Decoding Algorithm for the Davey-MacKay Construction , 2010, 2010 IEEE International Conference on Communications.

[7]  Johann A. Briffa,et al.  Codebook and marker sequence design for synchronization-correcting codes , 2011, 2011 IEEE International Symposium on Information Theory Proceedings.

[8]  Marilyn Wolf,et al.  Design space exploration of the turbo decoding algorithm on GPUs , 2010, CASES '10.

[9]  J.A. Briffa,et al.  Improvement of the Davey-MacKay construction , 2008, 2008 International Symposium on Information Theory and Its Applications.