中文摘要：

对于LDPC码的译码算法——和积算法，提出了一种新的基于差分的译码算法，其主要思想是：在LDPC码的二部图上所传递的消息是概率的差分值，而对于校验节点和消息节点的更新都是在特定的加法域中进行．针对校验节点的更新，还可以选择若干个绝对值最小的差分值进行运算，以进一步降低复杂度．与传统的基于对数似然比的译码方法相比，新算法的计算复杂度降低很大，而译码性能和收敛速度没有明显损失．

英文摘要：

As for the sum-product decoding algorithm of low density parity check （LDPC） codes, a difference-based decoding method is proposed. The main idea is shown as follows： the messages passed on the bipartite graph of LDPC are based on the probability differences, and the updates of both check nodes and message nodes are carried out on some special addition domain. As for the update of check nodes, only some lowest differences can be selected for reducing computational complexity. In contrast with the traditional log-likelyhood-ratio-based （LLR-based） decoding method, the proposed method has much lower complexity, while it has no obvious loss of performance and convergence.