系统工程与电子技术 ›› 2026, Vol. 48 ›› Issue (6): 2105-2112.doi: 10.12305/j.issn.1001-506X.2026.06.31

• 通信与网络 • 上一篇    下一篇

基于GPU的DSSS水声通信多普勒快速估计方法

彭海源1,2(), 李宇1, 朱雨男1, 普湛清1, 王巍1,*   

  1. 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
  • 收稿日期:2025-02-26 修回日期:2025-03-27 出版日期:2026-06-25 发布日期:2025-05-23
  • 通讯作者: 王巍 E-mail:phy20200309@163.com
  • 作者简介:彭海源(1998—),男,博士研究生,主要研究方向为信号与信息处理、扩频水声通信
    李 宇(1977—),男,研究员,博士,主要研究方向为水声信号处理、阵列信号处理、水声通信与网络技术
    朱雨男(1996—),男,博士后,主要研究方向为水声通信与网络技术
    普湛清(1989—),男,研究员,博士,主要研究方向为水声通信与网络技术
  • 基金资助:
    中国科学院青年创新促进会(E3291302)资助课题

GPU-based fast Doppler estimation method for DSSS underwater acoustic communication

Haiyuan PENG1,2(), Yu LI1, Yu’nan ZHU1, Zhanqing PU1, Wei WANG1,*   

  1. 1. Key Laboratory of Science and Technology on Advanced Underwater Acoustic Signal Processing,Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China
    2. University of Chinese Academy of Sciences,Beijing 100049,China
  • Received:2025-02-26 Revised:2025-03-27 Online:2026-06-25 Published:2025-05-23
  • Contact: Wei WANG E-mail:phy20200309@163.com

摘要:

直接序列扩频(direct sequence spread spectrum,DSSS)水声通信中基于交叉模糊函数(cross ambiguity function,CAF)的多普勒估计方法具有计算密集、求解耗时长的缺点。为了提高多普勒估计效率,提出一种基于图形处理器(graphic processing unit,GPU)的多普勒快速估计方法。首先,为了降低CAF多普勒估计方法的计算复杂度,提出一种改进两步式多普勒估计算法。其次,从设备内存预加载、运算单元并行化、GPU算法部署3个方面进行算法实现。最后,使用实际海试数据对方法进行测试并与CPU估计方法进行对比。测试结果表明,提出的GPU多普勒快速估计方法能够实现DSSS水声通信信号多普勒因子的准确估计。与CPU估计方法相比,最大加速比为263.65,极大地提高了多普勒估计算法的执行效率。

关键词: 直接序列扩频, 水声通信, 图形处理器, 多普勒估计

Abstract:

The Doppler estimation method based on cross ambiguity function (CAF) in direct sequence spread spectrum (DSSS) underwater acoustic communication has the disadvantages of being computationally intensive and time-consuming to solve. In order to improve the efficiency of Doppler estimation, a fast Doppler estimation method based on graphics processing unit (GPU) is proposed. Firstly, in order to reduce the computational complexity of the CAF Doppler estimation method, an improved two-step Doppler estimation algorithm is proposed. Secondly, the algorithm is realized from three aspects: device memory preloading, parallelization of operation units, and deployment on the GPU platform. Finally, the method is tested using actual sea trial data and compared with the CPU estimation method. The test results show that the GPU fast Doppler estimation method proposed in this paper can achieve accurate estimation of the Doppler factor of DSSS underwater acoustic communication signals. Compared with the CPU estimation method, the maximum speedup ratio is 263.65, which greatly improves the execution efficiency of the Doppler estimation algorithm.

Key words: direct sequence spread spectrum (DSSS), underwater acoustic communication, graphic processing unit (GPU), Doppler estimation

中图分类号: