The performance of a new method for decoding binary error correcting codes is presented, and compared with established hard and soft decision decoding methods. The new method uses a modified form of the maxsum algorithm, which is applied to a split (partially disconnected) modification of the Tanner graph of the code. Most useful codes have Tanner graphs that contain cycles, so the aim of the split is to convert the graph into a tree graph. Various split graph configurations have been investigated, the best of which have decoding performances close to maximum likelihood.
CITATION STYLE
Paire, J. T., Coulton, P., & Farrell, P. G. (2001). Graph configurations and decoding performance. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 2260, pp. 158–165). Springer Verlag. https://doi.org/10.1007/3-540-45325-3_15
Mendeley helps you to discover research relevant for your work.