Researching On Bipartite Graph Construction Algorithm Based On ACE Spectrum In LDPC Codes

Low-Density Parity-Check (LDPC) codes have attracted a great amount of interest during the last decade due to their capacity-approaching capabilities and low complexity message-passing decoding algorithms. Constructing of bipartite graph of LDPC codes is very important. By optimizing bipartite graph, LDPC codes' error-floor can be lowered and its decoding performance can be improved.Based on the theoretical research of LDPC, several optimized algorithms for constructing Tanner graph of LDPC codes are addressed. The main results and contents are as follows.1. The principles of coding and decoding for low-density parity-check codes are briefly summarized. The main factors which can impact the performance of LDPC codes such as stopping sets, cycles and connectivity of Tanner graph are systematically analyzed, emphasis on the connectivity of Tanner graph.2. ACE algorithm is been analyzed in detail and an improved ACE algorithm is been introduced. Because short-loop composition is unavoidable in constructing LDPC codes, this algorithm introduce loop-remove to deal with it. The improved algorithm can further optimize the connectivity of bipartite graph and lower error-floor.3. ACE spectrum algorithm is an algorithm in order to measure connectivity of bipartite graph. By introduce EMD of cycles, an improved ACE spectrum algorithm is been introduced, which improve the ACE spectrum algorithm's measure accuracy.4. Based on the tree model of the bipartite graph of LDPC codes, an algorithm for counting short cycles for LDPC code is proposed. The algorithm can count short cycles for given length easily. Applying this algorithm, this paper counts short cycles for ACE algorithm.