用于LDPC码快速译码的改进加权比特翻转算法

An Improved Weighted Bit-Flipping Algorithm for High-Speed LDPC Decoding

为了提高低密度奇偶校验码(LDPC)译码器的译码速度,提出了一种基于部分并行比特选择机制的快速多比特翻转算法.根据接收向量中错误具有随机分布的特点,将所有比特划分成若干子块,从每个子块挑选出1个候选翻转比特,再从这些候选比特中挑选出合理数目的比特进行翻转,完成译码迭代.此外,通过引入树形搜索和数据池技术降低该算法核心模块的计算复杂度,以进一步增加算法硬件实现时的译码速度.分析结果表明,相较于多比特翻转算法,利用所提出的算法和相关硬件实现技术,译码器的吞吐量能得到明显的提高.仿真结果验证了快速多比特翻转算法的有效性.

In order to improve the decoding speed of low density parity check code （LDPC） codes, a fast weighted bit flipping algorithm with partially parallel bit-chosen is proposed. In this scheme, all bits are firstly divided into several blocks according to the stochastic error distribution in the received vector. And then one bit is chosen from each block to form the candidate set of flipped bits. Finally, a proper number of bits are chosen from the candidate set and flipped to implement the decoding. Furthermore, the tree sorting and data pool methods are adopted to decrease the computational complexity of the key module of the proposed algorithm. Compared with the muhi-bit flippin （MBF） algorithm by simulations, the proposed scheme and the method of hardware implement can achieve significant improvement in the decoding speed.