一种低存储量需求的高效LDPC译码方法

提出了一种高效的低密度奇偶校验码（LDPC）译码方法,相对于传统译码方法,仅需增加少量存储量便能获得接近2 倍的吞吐率增益.本文将修正的最小和算法与分组双向消息传递（STMP）译码算法相结合,提出了基于最小和的分组双向消息传递算法（MS-STMP）,并给出了相应的迭代交叠方案.随后讨论了交叠过程中2 种运算单元对存储器的访问.最后以中国地面数字电视传输（DTTB）标准中使用的一组LDPC 码为例,计算了MS-STMP 算法相对于传统双向消息传递（TPMP）算法的吞吐率增益和额外增加的存储量.计算结果表明,MS-STMP 算法平均增加44%的存储量,而将吞吐率平均提高到1.85 倍,相对于交叠（OMP）算法有明显的优势.

An efficient decoding of LDPC codes with low memory requirement

An efficient decoding with low memory requirement is presented in this paper. With someextra memory bits,the proposed method could achieve nearly 2 times throughput compared with theconventional decoding. This paper combines min-sum decoding with Sliced Two-phase Message Passing（STMP） decoding, and proposes an overlapping scheme. The accessing of the operation units to the memoryunits is also discussed in order to analyze the access violation. A group of Low Density Parity Check（LDPC）codes are selected from Chinese Digital Television Terrestrial Multimedia Broadcasting（DTTB） standard,and the throughput gain as well as the ratio of extra memory bits for the proposed decoding over theconventional Two Phase Message Passing（TPMP） are calculated. The calculated results indicate thatMS-STMP decoding needs only extra 44% memory bits to achieve 1.85 times throughput in even,which ismore efficient than the Overlapped Message Passing（OMP） decoding.