首页|基于CPU+GPU混合架构的雷达信号处理方法

基于CPU+GPU混合架构的雷达信号处理方法

扫码查看
针对雷达系统复杂化趋势,提出了基于CPU+GPU混合架构的信号处理方法。依据运算复杂度进行任务的细粒度划分,将复杂处理任务解耦为逻辑处理和运算处理两部分,并映射到CPU+GPU核心进行处理。根据雷达信号处理的拆解包、脉冲压缩、相参积累、目标检测、目标检测后处理及解模糊过程,将对应处理映射到CPU+GPU核心以提高加速比。实验结果表明,相比多核DSP信号处理方法,该方法具有更好的性能。
A Radar Signal Processing Method Based on CPU+GPU Hybrid Architecture
Aiming at the complexity trends of the radar system,a signal processing method based on CPU+GPU hybrid architecture is proposed.According to the computational complexity,the fine granularity division of task is performed,and the complicated processing task is decoupled into two parts of logic processing and operation processing and mapped into CPU+GPU cores to be processed.According to the unpacking,pulse compressing,coherent-integration,target detection,post-processing of target detection and ambiguity resolution of radar signal process,the corresponding processes are mapped into CPU+GPU cores to improve the acceleration ratio.The experiment results show that the proposed method has better performance than the multi-core DSP signal processing method.

CPU+GPUHybrid architecturesignal processingmulti-core DSP

常艳、何涛、朱占宇

展开 >

中国航空工业集团公司雷华电子技术研究所,江苏 无锡 214063

苏州大学电子信息学院,江苏 苏州 215031

CPU+GPU 混合架构 信号处理 多核DSP

国家安全学术联合基金重点资助项目

U2130202

2024

火力与指挥控制
火力与指挥控制研究会,火力与指挥控制专业情报网

火力与指挥控制

CSTPCD北大核心
影响因子:0.312
ISSN:1002-0640
年,卷(期):2024.49(7)
  • 4