论文部分内容阅读
A new class of binary low-density parity-check (LDPC) codes is proposed based on B 2 (mod m) sequences. The parity-check matrix of such a code has a column weight of three and a row weight of an arbitrary integer,and a quasi-cyclic structure.The parity-check matrix also has a girth at least 8,and corresponds to a code with minimal distance at least 12.When m is prime,an 8-cycles reduction method is presented to completely avoid the two types of 8-cycles within the total four types existed in the Tanner graph.Simulation results show that,for a prime integer m,the new LDPC code outperforms the random (quasi-) regular counterpart generated by the PEG algorithm.Finally,a heuristic algorithm based on a strategy called neighboring extension search is presented to search for the B 2 (mod m) sequences whose lengths approach or meet the upper bound.
A new class of binary low-density parity-check (LDPC) codes is proposed based on B 2 (mod m) sequences. The parity-check matrix of such a code has a column weight of three and a row weight of an arbitrary integer , and a quasi-cyclic structure. The parity-check matrix also has a girth at least 8, and corresponds to a code with minimal distance at least 12.When m is prime, an 8-cycles reduction method is presented to completely avoid the two types of 8-cycles within the total four types existed in the Tanner graph. Simulation results show that, for a prime integer m, the new LDPC code outperforms the random (quasi-) regular counterpart generated by the PEG algorithm. Finally, a heuristic algorithm based on a mechanism called Neighbor extension search is presented to search for the B 2 (mod m) Sequences whose lengths approach or meet the upper bound.