Creating the optimal bipartite graph and decoding of a Low Density Parity Check (LDPC) codeword on it is deemed an NP-complete problem. Much work in the last decade has gone into proper construction of LDPC codes with maximum girth and optimized stopping sets; to ensure the BER approaches channel capacity. When an edge alteration to the graph is proposed, it completely changes the performance of the graph and usually leads to the re-analysis of the entire graph’s properties. In this work we propose a method of lowering the error floor experienced in an LDPC code by intelligently inserting a set of known bits in the message frame. This deactivates paths in the graph located around trapping sets without modification of edges in the graph, and provides high log-likelihood information to the induced sub-graph for improved performance.
History
Publication title
2015 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)
Pagination
2994-2998
ISBN
978-1-4673-6997-8
Department/School
School of Engineering
Publisher
IEEE-Inst Electrical Electronics Engineers Inc
Place of publication
New Jersey, USA
Event title
IEEE International Conference on Acoustics, Speech, and Signal Processing