提出了一种改进的分层修正最小和的LDPC译码算法,该算法充分考虑到了译码器硬件结构的特性,使用了部分信息节点提前中止迭代的方法,降低了译码器处理数据的位宽。同时,在这种算法的基础上,设计出了结构简单的译码器,该译码器在资源使用非常少的情况下可以获得较高的译码吞吐量,同时保持译码器译码性能和相应的浮点算法很接近。另外通过合理地设计LDPC码校验矩阵(H矩阵)和译码器数据处理单元,使得译码器可以支持多种码长码率LDPC码译码。这样结构特点的译码器,在低功耗以及需要多种码长码率的编码进行数据传输的领域有着非常高的应用价值。
The paper proposes an improved layered modified minimal sum algorithm (ILMMS algorithm) that can reduce the data width of the processing data in the decoder by means of stopping message updating of some bit nodes during iteration. Based on the ILMMSA, a simple, semi-parallel decoder structure is proposed. Such kind of decoder structure can decode LDPC code at a very high speed with rather low logic resource usage while keeping the decoding performance as good as float decoding algorithm. Also by properly designing H matrix and data processing unit, the decoder can decode multi code length and multi code rate LDPC code. Such kind of features is very useful in areas that need low power and multi code length and code rate data communication.