系统工程与电子技术 ›› 2026, Vol. 48 ›› Issue (11): 4066-4079.doi: 10.12305/j.issn.1001-506X.2026.11.36

• 通信与网络 • 上一篇    下一篇

基于并行计算的深海多波束水声信号快速帧同步方法

邬恩1,2,3(), 王巍1,2,3, 彭海源1,2,3, 田亚男1,3, 黄海宁1,2,3   

  1. 1. 中国科学院声学研究所,北京 100190
    2. 中国科学院大学,北京 100049
    3. 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
  • 收稿日期:2025-09-29 修回日期:2025-11-13 出版日期:2026-10-25 发布日期:2026-01-24
  • 通讯作者: 黄海宁 E-mail:wuen22@mails.ucas.ac.cn
  • 作者简介:邬 恩(2000—),男,博士研究生,主要研究方向为信号与信息处理、多载波水声通信
    王 巍(1983—),男,研究员,博士,主要研究方向为信号与信息处理、水声通信
    彭海源(1998—),男,博士研究生,主要研究方向为信号与信息处理、稳健水声通信、扩频水声通信
    田亚男(1995—),女,博士后,主要研究方向为水声通信与组网、信号处理及水下环境噪声
  • 基金资助:
    国家自然科学基金(62401559)资助课题

Fast frame synchronization method for deep-sea multi-beam underwater acoustic signals based on parallel computing

En Wu1,2,3(), Wei Wang1,2,3, Haiyuan Peng1,2,3, Yanan Tian1,3, Haining Huang1,2,3   

  1. 1. Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China
    2. University of Chinese Academy of Sciences,Beijing 100049,China
    3. Key Laboratory of Science and Technology on Advanced Underwater Acoustic Signal Processing,Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China
  • Received:2025-09-29 Revised:2025-11-13 Online:2026-10-25 Published:2026-01-24
  • Contact: Haining Huang E-mail:wuen22@mails.ucas.ac.cn

摘要:

针对深海水声通信系统接收机在多波束信号接收过程中对处理实时性的严格要求,提出一种基于中央处理单元与图形处理单元(graphic processing unit, GPU)异构计算平台的多波束帧同步快速方法。该方法提出了多波束频域匹配滤波并行架构与峰值索引同步搜索的GPU优化算法,对帧同步过程中的关键计算步骤进行并行化设计与优化。详细描述了该方法的实现流程,并在仿真平台上对所提方法的加速性能与同步准确性进行了验证。结果表明,在满足毫秒级同步误差的情况下,所提方法可实现最高6.85倍的加速比。进一步将该方法部署于实际水声通信系统平台,开展外场试验,实验结果表明,该方法在满足系统实时性要求的同时,能够在深海复杂信道条件下准确识别信号起始位置与最优接收方向,具有良好的准确性与可靠性。

关键词: 水声通信, 波束形成, 帧同步, 图形处理单元

Abstract:

To address the real-time processing requirements of receivers in deep-sea underwater acoustic communication systems during multi-beam signal reception, a fast multi-beam frame synchronization method is proposed based on a heterogeneous platform that combines a central processing unit and a graphics processing unit (GPU). This method proposes a GPU-optimized algorithm featuring a multi-beam frequency-domain matched filtering parallel architecture and synchronous peak index search, which parallelizes and optimizes key computational steps in frame synchronization. This paper elaborates on the implementation workflow of the approach and validates its acceleration performance and synchronization accuracy on a simulation platform. Under the condition of satisfging millisecond-level synchronization error, a maximum speed-up ratio of 6.85 is achieved. The method is further deployed on a practical underwater acoustic communication system platform and is tested in field experiments. Experimental results show that the method meets the system’s real-time processing requirements while achieving high accuracy and reliability in detecting the signal start position and optimal receiving direction under complex deep-sea channel conditions.

Key words: underwater acoustic communication, beamforming, frame synchronization, graphics processing unit(GPU)

中图分类号: