基于LDPC编码器的多码率并行算法的设计研究  

Design Study of Multi-Code Rate Parallel Algorithm Based on LDPC Encoder

作　　者：孙紫檀 周益 易小谦 赵臣 李生贵 SUN Zitan;ZHOU Yi;YI Xiaoqian;ZHAO Chen;LI Shenggui(Comac Flight Test Center,Shanghai 201323,China)

机构地区：[1]中国商用飞机有限责任公司民用飞机试飞中心,上海201323

出　　处：《自动化与仪器仪表》2024年第5期33-37,共5页Automation & Instrumentation

摘　　要：为了提高通信信息传输过程中的译码速度,研究以低密度校验码为对象,选取了置信传播译码算法为研究对象,并对其进行并行性能优化。该优化过程分为两步,第一步是采用对数似然比法对算法的译码过程进行运算简化,第二步是通过通用处理器提升算法的并行速率。对优化算法的性能进行验证后,结果表明置信传播译码算法在信噪比为4.0时,BP算法的耗时为372 ms。优化后的置信传播译码算法耗时降低到了287 ms。故研究提出的优化算法可用于通信领域的并行译码工作中。Low-density parity check codes were investigated in order to increase the decoding speed during communication information transfer.The confidence propagation decoding algorithm was chosen as the research subject,and its parallel performance was optimized.There are two steps in the optimization process.The first stage is to use the logarithmic likelihood ratio approach to simplify the algorithm’s decoding process.The second step is to use a universal processor to increase the algorithm’s parallel speed.The confidence propagation decoding process takes 372 ms when the signal-to-noise ratio is 4.0,according to the results of the optimization algorithm's performance verification.The time consumption drops to 287 ms with the enhanced confidence propagation decoding technique.Therefore,the optimization algorithm proposed in the study can be used in parallel decoding work in the field of communication.

关 键 词：LDPC 多码率 并行运算 BP算法 GPU 

分 类 号：TN911[电子电信—通信与信息系统] TP39[电子电信—信息与通信工程]