计算机工程

• •    

基于对数似然比和极化信道可靠度的SCF译码算法

  

  • 发布日期:2021-01-13

SCF decoding algorithm based on log likelihood ratio and channel error probability

  • Published:2021-01-13

摘要: 传统的串行抵消比特翻转(Successive Cancellation Bit-Flipping, SCF)译码算法仅仅用对数似然比 (Log-Likelihood-Ratio, LLR)的绝对值去衡量信息比特译码结果的可靠情况,造成误块率(Block Error Rate , BLER)过高和翻转的尝试次数较多。为此提出一种基于 LLR 和信道极化可靠度的 SCF 译码算法 (PLR-SCF)。分析 SC 译码算法发生错误译码的原因并通过仿真观察对数似然比、信道极化可靠度和信 息位所在的位置与 SC 译码算法发生首个判决错误(The First Decision Error, TFDE)之间的关系。利用这 些因素设计一个能准确的衡量信息位发生译码错误程度的度量公式。仿真结果表明,提出的译码算法的 BLER 相对于传统的 SCF 译码算法得到了降低,特别是在高信噪比(Signal-to-Noise Ratio,SNR)下获得 最大信噪比增益大约为 0.12dB,并且翻转尝试次数与 SCF 相比减少了 13.6%。

Abstract: The traditional successive cancellation bit-flipping (SCF) decoding algorithm only uses the absolute value of log like ratio (LLR) to measure the reliability of information bit decoding, resulting in high block error rate (BLER) and more attempts to flip. This paper proposes a SCF decoding algorithm based on LLR and channel polarization reliability(PLR-SCF).Analyzing the causes of error decoding in SC decoding algorithm and the relationship between log likelihood ratio, channel polarization reliability and the location of information bits with the first decision error (TFDE) of SC decoding algorithm. Using these factors, a measurement formula is designed to accurately measure the decoding error of information bits. Simulation shows that BLER of the proposed decoding algorithm is lower than that of the traditional SCF decoding algorithm. Especially in high signal-to-noise ratio(SNR),the maximum SNR gain is 0.12dB and the number of flip attempts are reduced by 13.6% compared with SCF.