针对准循环LDPC(QC—LDPC)码基于洪水消息传递机制译码算法的不足,该文提出了一种快速的分组串行译码算法。该算法通过将LDPC码的校验节点(或变量节点)按一定规则划分成若干个子集,在每一轮迭代过程中,依次对各个子集中的校验节点(或变量节点)并行地进行消息更新,提高了译码速度。同时根据分组规则,提出了一种有效的分组方法,并通过分析发现基于循环置换阵的准循环LDPC码非常适合采用这种分组译码算法进行译码。通过对不同消息传递机制下准循环LDPC码译码算法性能的仿真比较,验证了在复杂度不增加的情况下,该译码算法在继承了串行译码算法性能优异和迭代收敛快等优点的同时,极大地提高了准循环LDPC码的译码速度。分析表明,分组串行译码算法译码速度至少为串行译码算法的p倍(p为准循环LDPC码校验矩阵中循环置换阵的行数或列数)。
A new fast decoding method of Quasi-Cyclic LDPC(QC-LDPC)codes, named semi-serial decoding algorithm, is presented through contraposing the shortcoming of flood decoding method.For the algorithm, the message was updated in a serial sequence of m subsets of check nodes(or variable nodes), while the check nodes(or variable nodes) in the same subset are updated simultaneously. In this way, the decoding speed would be faster than the common serial decoding algorithm several times. An efficient grouping method suitable for the quasi-cyclic LDPC codes based on the circulant architecture is proposed.Simulation results show that,under the same complexiry, the proposed algorithm fastens the decoding speed of quasi-cyclic LDPC codes, meanwhile, it inherits all the merits of the serial massage-passing decoding algorithm. The conclusion is that the speed of the semi-serial decoding method is at least p times than that of the serial one(p is the number of rows of the cyclic permutation matrix in the parity check matrix of quasi-cyclic LDPC codes).