基于GPU的OMCSS水声通信M元解扩算法并行实现  

GPU-based parallel implementation of M-ary despreading algorithm for OMCSS underwater acoustic communication

作　　者：彭海源 王巍[1] 李德瑞 刘彦君 李宇[1] 迟骋 田亚男 PENG Haiyuan;WANG Wei;LI Derui;LIU Yanjun;LI Yu;CHI Cheng;TIAN Yanan(Key Laboratory of Science and Technology on Advanced Underwater Acoustic Signal Processing,Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China;U niversity of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院声学研究所先进水下信息技术重点实验室,北京100190 [2]中国科学院大学,北京100049

出　　处：《系统工程与电子技术》2025年第3期978-986,共9页Systems Engineering and Electronics

基　　金：国家自然科学基金(E311130101)资助课题。

摘　　要：针对正交多载波扩频(orthogonal multi-carrier spread spectrum,OMCSS)水声通信系统接收信号快速处理需求,提出一种基于图形处理模块(graphic processing unit,GPU)的M元解扩算法的并行实现方法。首先,分析M元解扩算法在GPU平台上实现的可行性,针对算法内部基础运算单元进行并行优化处理。然后,为了进一步提升GPU并行运行速度,对算法进行基于并发内核执行的M元并行解扩计算架构设计。在中央处理器(central processing unit,CPU)+GPU异构平台上对算法性能进行测试。测试结果表明,设计的M元并行解扩算法相比M元串行解扩算法在运行速度上有最大90.47%的提升,最大加速比为10.5。In response to the requirements for fast processing of received signals in the orthogonal multi-carrier spread spectrum(OMCSS)underwater acoustic communication system,a parallel implementation of the M-ary despreading algorithm based on the graphic processing unit(GPU)is proposed.First,the feasibility of implementing the M-ary despreading algorithm on the GPU platform is analyzed,and parallel optimization processing is performed on the internal basic computing units of the algorithm.Then,in order to improve the parallel processing speed,M-ary parallel despreading computing architecture based on concurrent kernel execution is designed.The algorithm performance is tested on a central processing unit(CPU)+GPU heterogeneous platform.The test results show that the M-ary parallel despreading algorithm designed in this paper has a maximum improvement of 90.47%in running speed compared to the M-ary serial despreading algorithm,and the maximum speedup ratio is 10.5.

关 键 词：正交多载波扩频 水声通信 M元解扩 图形处理模块 并行实现 

分 类 号：TN929.3[电子电信—通信与信息系统]