| 摘要: |
| 首先提出基于初始对数似然比（Log-Likelihood Ratio，LR）与路径度量值（Path Metric,PM）的PM-LLR-SCL译码算法，在接收端初始LLR和PM值之间建立映射关系，并通过重排PM值完成翻转功能。其次，提出基于极化码生成矩阵的行重特性和PM值的PM-RW-SCL译码算法，不仅考虑了Polar码的最小码距和极化子信道可靠度，同时将路径分裂每一层的PM值引入到译码策略中，从而提高了译码性能。仿真结果显示，与串行抵消列表比特翻转（Successive Cancellation List Bit-flip,SCLF）相比，提出的PM-LLR-SCL算法最大可获得约0.23 dB的性能增益，而基于路径数量的复杂度降低了约62%；与基于行权重的串行抵消列表翻转译码算法相比，PM-RW-SCL算法最大可获得约1.5 dB的性能增益，而复杂度降低了约39%。 |
| 关键词: Polar码 串行抵消列表比特翻转译码 路径度量值 行重特性 |
| DOI：10.20079/j.issn.1001-893x.220822005 |
|
| 基金项目:国家自然科学基金资助项目（62261003,61961004）；广西重点研发计划项目（2021AB05081） |
|
| SCL Decoding Algorithm for Polar Code Combined with Path Metric and Row Weight Property |
| ZHOU Quan,CHEN Haiqiang,ZENG Qiaoli,LIAO Lanjuan,SUN Youming,LI Xiangcheng |
| (1.School of Computer,Electronics and Information,Guangxi University,Nanning 530004,China;2.Guangxi Key Laboratory of Multimedia Communications and Network Technology,Nanning 530004,China) |
| Abstract: |
| A Successive Cancellation List(SCL) decoding algorithm for Polar code based on initial Log-Likelihood Ratio(LLR) and Path Metric(PM),called PM-LLR-SCL,is presented.For the presented algorithm,the mapping relationship between LLR and PM is constructed.Then the flipping operation is replaced by the PM re-arrangement.Second,an SCL decoding algorithm based on the row-weight property of the generator matric and the PM,called PM-RW-SCL is further presented.The presented PM-RW-SCL algorithm not only considers the minimum-distance of the Polar code and the reliability of the sub-channel,but also introduces the PM into the decoding strategy so as to improve the decoding performance.Simulation results show that,compared with the SCL Bit-flip(SCLF),the presented PM-LLR-SCL algorithm can achieve about 0.23 dB performance gain at the most,while the complexity can be reduce by about 62%.Similarly,compared with the SCLF based on row weight,the presented PM-RW-SCL algorithm can achieve about 1.5 dB gain at the most,while the complexity can be reduce by about 39%. |
| Key words: Polar code SCLF decoding path metric row-weight property |
