摘要
近些年,随着高频地波雷达技术的发展,大量的遥感数据给实时信号处理带来了巨大的挑战。高频地波雷达信号处理机的主流设计方案通常采用DSP+FPGA技术或基于高性能CPU技术来实现信号的实时处理,但DSP+FPGA方案中系统的软硬件耦合较为紧密,开发周期长且不易维护,而高性能CPU技术对计算机性能要求较高,大大增加了硬件成本。在保证信号实时处理的前提下,为了缩短开发周期,提高系统的可维护性和降低开发成本,本文基于CPU+GPU架构设计了的信号处理方案,实现了高频地波雷达实时信号处理的基本流程,获取了目标的距离信息和速度信息。实验结果表明,本文提出的CPU+GPU架构下的信号处理方案与同期中央处理器(CPU)运算平台相比,具有10倍以上的加速比,验证了方案的可行性。
With the development of High Frequency Surface Wave Radar（HFSWR） technologies in recent years, the huge amount of remote sensing data brings challenges for real-time imaging processing. The ma- instream design scheme of High frequency ground wave signal processor is usually DSP＋FPGA technolo- gy or based on high performance CPU technology to achieve real-time signal processing, but the hardware and software of system using DSP＋FPGA is more compact ,it is difficult to maintain because of long de- velopment cycle. High performance CPU technology requires high performance on the hardware,it greatly increasing the cost of hardware. In order to shorten the development cycle, improve the maintainability of the system and reduce the development cost under the premise of guaranteeing signal real-time processing, this paper designs a signal processing scheme based on CPU＋GPU architecture, realizes the basic flow of real-time signal processing of HF Ground wave Radar, including data pre-processing, the distance infor- mation and speed information of the aim are acquired. Experiment results shows that the proposed signal processing scheme under CPU＋GPU architecture has a speedup of 10 times compared with the CPU plat- form of the same period, which verifies the feasibility of the scheme.
出处
《中国海洋大学学报（自然科学版）》
CAS
CSCD
北大核心
2017年第2期56-61,共6页
Periodical of Ocean University of China
基金
海洋公益性行业科研专项项目(201505002)
国家自然科学基金重点项目(61132005)资助~~
