《计算机应用研究》|Application Research of Computers

一种基于CUDA的截断重叠维特比译码算法

Truncated overlap scheduling Viterbi decoding algorithm based on CUDA

免费全文下载 (已被下载 次)  
获取PDF全文
作者 李晨杰,王志旭
机构 南京邮电大学 通信与信息工程学院,南京 210003
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2019)06-047-1828-03
DOI 10.19734/j.issn.1001-3695.2017.12.0794
摘要 为解决信道译码在高吞吐量通信系统中的瓶颈问题,通过对CUDA并行计算的了解和对维特比译码并行实现的探索,为卷积码提出了一种基于CUDA的截断重叠维特比译码器。算法通过截断式的子网格图相互重叠的方式,并行执行独立的正向度量计算和回溯过程。实验结果表明,在保证了译码算法误码率性能的同时,获得了良好的吞吐量提升表现,相比现有的实现方式提升了1.3~3.5倍,降低了硬件开销,能够有效运用于实际高吞吐量通信系统中。
关键词 卷积码; 维特比译码; 并行计算; 计算机统一设备架构
基金项目
本文URL http://www.arocmag.com/article/01-2019-06-047.html
英文标题 Truncated overlap scheduling Viterbi decoding algorithm based on CUDA
作者英文名 Li Chenjie, Wang Zhixu
机构英文名 College of Telecommunications & Information Engineering,Nanjing University of Posts & Telecommunications,Nanjing 210003,China
英文摘要 In order to solve the bottleneck problem of channel decoding in high-throughput communication systems, this paper proposed a truncated overlap Viterbi decoder based on CUDA for convolutional codes to solve it by analyzing of parallel processing based on compute unified device architecture(CUDA) and exploring of the parallel implementation of Viterbi decoding. The algorithm performed both independent forward metrics computing and back-track procedure in parallel through the overlapping of truncated sub-grid. The experiment shows that the method keeps low BER, achieves a performance improvement of 1.3~3.5 times of the existing implementation and reduces hardware consumption. It can be effectively used in practical high-throughput communication systems.
英文关键词 convolutional codes; Viterbi decoder; parallel processing; compute unified device architecture(CUDA)
参考文献 查看稿件参考文献
 
收稿日期 2017/12/4
修回日期 2018/1/30
页码 1828-1830,1835
中图分类号 TP301.6
文献标志码 A