低复杂度的LDPC并行分层译码算法研究  

Research on LDPC Parallel Hierarchical Decoding Algorithm with Low Complexity

作　　者：卢恒 魏华[1] 王国开 Lu Heng;Wei Hua;Wang Guokai(Chengdu University of Information Technology,Chengdu 610000,China)

机构地区：[1]成都信息工程大学,四川成都610000

出　　处：《邮电设计技术》2024年第10期54-59,共6页Designing Techniques of Posts and Telecommunications

摘　　要：为了提高准循环低密度奇偶校验(QC_LDPC)算法的译码性能,降低其译码算法的复杂性,加快译码收敛速度,提出了一种基于最小和译码算法的分层译码,以支持QC_LDPC快速译码,它由基校验矩阵通过循环移位得来,所以每一层之间的校验矩阵相互独立。相对于传统的BP算法,提出的分层算法复杂度更低,收敛速度更快。根据5G NR的标准,对该算法在各种情况下进行了仿真,当码长K=3960,码率为1/3时,使用BPSK进行调制、AWGN信道进行仿真,结果显示,此算法在SNR为1.3 dB左右时,得到了10-5的BER,该解码方法获得了0.1dB的性能增益、迭代次数的减少以及译码吞吐率的提升。In order to improve the decoding performance of quasi-cyclic low-density parity check(QC_LDPC),reduce the complexity of its decoding algorithm and accelerate the decoding convergence speed,a hierarchical decoding based on min-Sum decoding algorithm is proposed to support QC_LDPC fast decoding,which is obtained by cyclic shift of the basis check matrix,so the check matrix between each layer is independent of each other.Compared with the traditional BP algorithm,the proposed hierarchical algorithm has lower complexity and faster convergence.According to the standard of 5G NR,the algorithm is simulated in various scenarios.When the code length K is 3960 and the code rate is 1/3,BPSK modulation and AWGN channel are used to simulate,and the results show that this algorithm can achieve a BER of 10-5 at the SNR of about 1.3 dB.This decoding method achieve a performance gain of 0.1 dB,a reduction in iteration times,and an improvement in decoding throughput.

关 键 词：QC_LDPC码 分层译码 最小和译码算法 5G NR 低复杂度 

分 类 号：TN929.5[电子电信—通信与信息系统]