利用同步信息修正的LDPC和积译码

LDPC Sum-Product Decoding Based on Synchronized Information

摘要: 低密度奇偶校验码(LDPC)是一类具有稀疏校验矩阵的线性分组码，和积算法是理论可行的LDPC软解码算法，但其实用受制于计算复杂度。最小和算法及其它已知改进算法采用近似来有效简化每次迭代的过程，以部分性能为代价，在一定程度上弥补了前述缺陷。本文提出了基于同步信息修正的和积算法，在保持原算法每次迭代复杂度以保证性能的前提下，利用同步信息对算法进行修正，有效减少了算法迭代次数，因此降低了计算复杂度，提高了实用性。

Abstract: Low density parity check code (LDPC) is a subclass of linear block code with sparse check matrix. Sum-product algorithm is efficient for decoding LDPC theoretically, but not practically due to its complexity. Improved algorithms including minimum sum algorithm reduce computational com-plexity by approximation but performance is degraded. In this paper, we introduce corrected sum-product algorithm based on synchronized information, which improves the practicability by reducing the iterations without performance degradation.