基于最小熵的地基MIMO-SAR幅相校正GPU加速与嵌入式实现

doi:10.12305/j.issn.1001-506X.2026.08.10

系统工程与电子技术 ›› 2026, Vol. 48 ›› Issue (8): 2615-2625.doi: 10.12305/j.issn.1001-506X.2026.08.10

• 传感器与信号处理 • 上一篇

基于最小熵的地基MIMO-SAR幅相校正GPU加速与嵌入式实现

郤伟杰, 赖涛, 梁高天, 王青松, 黄海风

中山大学电子与通信工程学院，广东深圳 518107

收稿日期:2025-04-01 修回日期:2025-05-12 接受日期:2025-06-11 出版日期:2026-03-12 发布日期:2026-03-12
通讯作者: 赖涛
作者简介:郤伟杰（2000—），男，硕士研究生，主要研究方向为MIMO雷达天线与阵列设计、MIMO通道幅相误差校正
梁高天（1998—），男，博士研究生，主要研究方向为地基SAR成像与干涉、三维阵列信号处理、空间信息自动化感知
王青松（1983—），男，副教授，博士，主要研究方向为超遥感图像精化处理、智能视觉导航、协同探测感知与信息融合
黄海风（1976—），男，教授，博士，主要研究方向为空间电子和智能感知领域的基础理论和关键技术攻关研究
基金资助:
深圳市科技计划(SGDX20230116092503007)；广东省先进智能感知技术重点实验室基金(2023B1212060024)资助课题

GPU acceleration and embedded implementation for amplitude-phase correction of ground-based MIMO-SAR based on minimum entropy

Weijie XI, Tao LAI, Gaotian LIANG, Qingsong WANG, Haifeng HUANG

School of Electronics and Communication Engineering，Sun Yat-sen University，Shenzhen 518107，China

Received:2025-04-01 Revised:2025-05-12 Accepted:2025-06-11 Online:2026-03-12 Published:2026-03-12
Contact: Tao LAI

摘要/Abstract

摘要：

针对多输入多输出(multiple input multiple output，MIMO)合成孔径雷达(synthetic aperture radar，SAR)中的通道幅相误差问题，最小化图像熵幅相校正算法是一种有效的解决方法，但算法运算量大、估计时间长，难以用于实时校正。对此，提出一种最小熵幅相校正的图形处理单元(graphics processing unit，GPU)实现方法，在数据传输和运算方面使用混合精度计算，保证算法精度的同时大幅提升了运算效率；通过异步并行流处理技术，充分利用GPU资源对算法进行加速，并最终将其部署在嵌入式GPU平台上。通过对实测数据的对比分析，桌面级GPU以及嵌入式GPU上的测试结果证明了所提方法的准确性和高效性，该算法可以对幅相误差进行有效估计，加速比最高可达到105%。

关键词: 合成孔径雷达, 多输入多输出雷达, 通道幅相校正, 最小熵算法, 图形处理单元加速

Abstract:

Addressing the channel amplitude and phase error issue in multiple input multiple output (MIMO) synthetic aperture radar (SAR), the minimum image entropy amplitude and phase correction algorithm serves as an effective solution. However, due to the large computational load and long estimation time, the algorithm is difficult to use for real-time correction. To this end, a graphics processing unit (GPU) implementation method for minimum entropy amplitude and phase correction is proposed. This method employs mixed-precision computation in data transmission and computation, ensuring algorithm accuracy while significantly enhancing computational efficiency. By leveraging asynchronous parallel stream processing technology, GPU resources are fully utilized to accelerate the algorithm, which is ultimately deployed on an embedded GPU platform. Through comparative analysis of measured data, test results on desktop GPU and embedded GPU demonstrate the accuracy and efficiency of the proposed method. The algorithm can effectively estimates amplitude and phase errors, achieving an acceleration ratio of up to 105%.

Key words: synthetic aperture radar (SAR), multiple input multiple output (MIMO) radar, channel amplitude-phase correction, minimum entropy algorithm, graphics processing unit (GPU) acceleration

中图分类号:

TN 911.7

郤伟杰, 赖涛, 梁高天, 王青松, 黄海风. 基于最小熵的地基MIMO-SAR幅相校正GPU加速与嵌入式实现[J]. 系统工程与电子技术, 2026, 48(8): 2615-2625.

Weijie XI, Tao LAI, Gaotian LIANG, Qingsong WANG, Haifeng HUANG. GPU acceleration and embedded implementation for amplitude-phase correction of ground-based MIMO-SAR based on minimum entropy[J]. Systems Engineering and Electronics, 2026, 48(8): 2615-2625.

图/表 13

图1

图2

表1

图3

图4

图5

图6

图7

表2

图8

表3

表4

图9

参考文献 30

1	LI J, STOICA P. MIMO radar with colocated antennas[J]. IEEE Signal Processing Magazine, 2007, 24 (5): 106- 114. doi: 10.1109/MSP.2007.904812
2	MOREIRA A, PRATS-IRAOLA P, YOUNIS M, et al. A tutorial on synthetic aperture radar[J]. IEEE Geoscience and Remote Sensing Magazine, 2013, 1 (1): 6- 43. doi: 10.1109/MGRS.2013.2248301
3	王敏, 马晓川, 鄢社锋, 等. 阵列幅度/相位误差的有源校正新方法[J]. 信号处理, 2015, 31 (11): 1389- 1395. doi: 10.3969/j.issn.1003-0530.2015.11.001
	WANG M, MA X C, YAN S F, et al. A new active calibration method for array amplitude/phase errors[J]. Journal of Signal Processing, 2015, 31 (11): 1389- 1395. doi: 10.3969/j.issn.1003-0530.2015.11.001
4	王纪平, 刘衍琦, 邓钱钰, 等. 基于PAST的毫米波雷达MIMO阵列幅相误差校正算法[J]. 信号处理, 2024, 40 (9): 1633- 1647. doi: 10.12466/xhcl.2024.09.006
	WANG J P, LIU Y Q, DENG Q Y, et al. A PAST-based amplitude and phase error calibration algorithm for millimeter-wave radar MIMO arrays[J]. Journal of Signal Processing, 2024, 40 (9): 1633- 1647. doi: 10.12466/xhcl.2024.09.006
5	王鼎, 邬钧霆, 叶国华, 等. 一种均匀线阵互耦和幅相误差校正算法[J]. 信号处理, 2010, 26 (1): 38- 45.
	WANG D, WU J T, YE G H, et al. A calibration algorithm for mutual coupling and amplitude-phase errors in uniform linear arrays[J]. Journal of Signal Processing, 2010, 26 (1): 38- 45.
6	LI Y, ER M H. Theoretical analyses of gain and phase error calibration with optimal implementation for linear equispaced array[J]. IEEE Trans. on Signal Processing, 2006, 54 (2): 712- 723. doi: 10.1109/TSP.2005.861892
7	LIU A F, LIAO G S, ZENG C, et al. An eigenstructure method for estimating DOA and sensor gain-phase errors[J]. IEEE Trans. on Signal Processing, 2011, 59 (12): 5944- 5956. doi: 10.1109/TSP.2011.2165064
8	WANG J R, LAI T, WANG Q S, et al. Entropy-based parameterized amplitude and phase correction for MIMO ground-based SAR[J]. IEEE Trans. on Geoscience and Remote Sensing, 2024, 62, 5215014. doi: 10.1109/tgrs.2024.3402582
9	KELKAR A, DICK C. NVIDIA aerial GPU hosted AI-on-5G[C]//Proc. of the IEEE 4th 5G World Forum, 2021: 64−69.
10	TIAN H N, HUA W, GAO Y, et al. Research on real-time imaging method of airborne SAR based on embedded GPU[C]//Proc. of the 3rd China International SAR Symposium, 2022.
11	YANG T, XU Q B, MENG F T, et al. The distributed imaging processing method of space-borne SAR based on embedded GPU[C]//Proc. of the CIE International Conference on Radar, 2021: 255−258.
12	SOUROURI M, GILLBERG T, BADEN S B, et al. Effective multi-GPU communication using multiple CUDA streams and threads[C]//Proc. of the IEEE 20th International Conference on Parallel and Distributed Systems, 2014: 981−986.
13	LIU K, ZHANG Y H, CAO Y, et al. MIMO-SAR image antialiasing for cascaded mmWave radar sensor[C]//Proc. of the IEEE Radar Conference, 2023.
14	WANG W Q, PENG Q, CAI J. Diversified MIMO SAR waveform analysis and generation[C]//Proc. of the 2nd Asian-Pacific Conference on Synthetic Aperture Radar, 2009: 270−273.
15	LIN Y, GUO M, CHEN Q, et al. TDM FMCW MIMO radar for imaging and change detection[C]//Proc. of the 8th Asia-Pacific Conference on Antennas and Propagation, 2019.
16	XUE B, ZHANG G, LEUNG H, et al. Multi-sub-chirp signal synthesis for millimeter-wave radar based on dechirp processing[C]//Proc. of the IEEE Radar Conference, 2024.
17	ZHU D. A novel approach to residual video phase removal in spotlight SAR image formation[C]//Proc. of the IET International Conference on Radar Systems, 2007.
18	MOHAMMAD T S O, HAMED E T. The new spectral conjugate gradient method for minimization[C]//Proc. of the International Conference on Data Science and Intelligent Computing, 2022: 273−277.
19	PAZDERIN A, YUFEREV S. Power flow calculation by combination of Newton-Raphson method and Newton’s method in optimization[C]//Proc. of the 35th Annual Conference of IEEE Industrial Electronics, 2009: 1693−1696.
20	CHOQUETTE J, GANDHI W, GIROUX O, et al. NVIDIA A100 tensor core GPU: performance and innovation[J]. IEEE Micro, 2021, 41 (2): 29- 35. doi: 10.1109/MM.2021.3061394
21	CUI H, DAHNOUN N. Real-time stereo vision implementation on NVIDIA Jetson TX2[C]//Proc. of the 8th Mediterranean Conference on Embedded Computing, 2019.
22	HOU Q D, QIU C, MU K H, et al. A cloud gaming system based on NVIDIA GRID GPU[C]//Proc. of the 13th International Symposium on Distributed Computing and Applications to Business, Engineering and Science, 2014: 73−77.
23	HE B, HUYNH H P, MONG R G S. GPGPU for real-time data analytics[C]//Proc. of the IEEE 18th International Conference on Parallel and Distributed Systems, 2012: 945−946.
24	WANG C X, SONG P H, ZHAO H Y, et al. Randomized testing framework for dissecting NVIDIA GPGPU thread block-to-SM cheduling mechanisms[C]//Proc. of the IEEE 29th International Conference on Parallel and Distributed Systems, 2023: 1843−1852.
25	PARK S, CHO K, BAHN H. Analysis of thread block scheduling algorithms for general purpose GPU systems[C]//Proc. of the IEEE Asia-Pacific Conference on Computer Science and Data Engineering, 2021.
26	CHIEN S, PENG I, MARKIDIS S. Performance evaluation of advanced features in CUDA unified memory[C]//Proc. of the IEEE/ACM Workshop on Memory Centric High Performance Computing, 2019: 50−57.
27	GANJI D C, ASHFAQ S, SABOORI E, et al. DeepGEMM: accelerated ultra low-precision inference on CPU architectures using lookup tables[C]//Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2023: 4656−4664.
28	HAIDAR A, TOMOV S, DONGARRA J, et al. Harnessing GPU tensor cores for fast FP16 arithmetic to speed up mixed-precision iterative refinement solvers[C]//Proc. of the International Conference for High Performance Computing, 2018: 603−613.
29	COOK S. CUDA programming: a developer’s guide to parallel computing with GPUs[M]. San Francisco: Morgan Kaufmann Publishers, 2012.
30	SCALCON F P, TAHAL R, AHRABI M, et al. AI-powered video monitoring: assessing the NVIDIA Jetson orin devices for edge computing applications[C]//Proc. of the IEEE Transportation Electrification Conference and Expo, 2024.

算法核心步骤	乘法次数	加法次数
FFT/IFFT	$MN{\log _2}\;M$	$MN{\log _2}\;M$
误差补偿	$8MN$	$4MN$
熵计算	$3MN$	$2MN$
相位梯度	$5MN\left( {{{\log }_2}\;MN + 2} \right)$	$5MN\left( {{{\log }_2}\;MN + 1} \right)$
相位步长	$15MN\left( {{{\log }_2}\;MN + 2} \right)$	$5MN\left( {{{\log }_2}\;MN + 1} \right)$
幅度梯度	$10MN\left( {{{\log }_2}\;MN + 1} \right)$	$3MN\left( {3{{\log }_2}\;MN + 1} \right)$
幅度步长	$15MN\left( {{{\log }_2}\;MN + 2} \right)$	$15MN\left( {{{\log }_2}\;MN + 1} \right)$
搜索方向	$3M$	$3M$

处理平台	型号	核心配置
CPU	Intel（R） Core（TM） i9-12900H	2.5 G主频，32 G内存
GPU	GeForce RTX 4090	16384 个 CUDA 核心，24 G显存
GPU	Jetson Orin NX	1024 个 CUDA 核心，8 G显存

GPU平台和算法精度	幅度误差	相位误差
RTX 4090（双精度）	2.87×10⁻¹¹	3.30×10⁻¹¹
RTX 4090（混合精度）	1.06×10⁻³	7.52×10⁻³
Jetson NX（双精度）	2.99×10⁻¹¹	3.56×10⁻¹¹
Jetson NX（混合精度）	1.07×10⁻³	7.38×10⁻³

平台	方案	耗时/s	加速比
CPU	—	216.69	—
RTX 4090	常规	5.579	38.747
RTX 4090	本文算法	2.044	105.757
Jetson NX	常规	164.214	1.316
Jetson NX	本文算法	69.871	3.094

[1]	孙慧霞, 周生华, 杜小荷, 南静怡, 郭小路, 林明聪. 脉压宽主瓣子阵MIMO雷达收发联合设计[J]. 系统工程与电子技术, 2026, 48(8): 2602-2614.
[2]	刘畅, 王凌宇, 夏浪, 李岳峰, 林欣, 刘艳阳, 黄鹏辉. 基于小波散射激励与SimAM的轻量化抗噪卷积神经网络的目标识别[J]. 系统工程与电子技术, 2026, 48(7): 2184-2194.
[3]	金梓涵, 江率, 刘彦斌, 曹越, 喻忠军. 基于脉间捷变频的低空聚束SAR快速成像算法[J]. 系统工程与电子技术, 2026, 48(7): 2229-2240.
[4]	付卫红, 彭文洪, 刘乃安. 面向高效检测的YOLOv11轻量化SAR图像船舶检测模型[J]. 系统工程与电子技术, 2026, 48(7): 2241-2250.
[5]	赵巍山, 尤思洋, 黄丽佳, 周光尧. 基于散射拓扑增强的双流SAR飞机检测网络[J]. 系统工程与电子技术, 2026, 48(6): 1848-1858.
[6]	袁永康, 王君旭, 华煜明, 张磊. 地基雷达对月多普勒补偿成像方法[J]. 系统工程与电子技术, 2026, 48(6): 1859-1868.
[7]	杨泽莹, 陈卓, 徐志明, 艾小锋, 吴其华, 王玲. 空间目标ISAR图像部件正向标记方法[J]. 系统工程与电子技术, 2026, 48(4): 1165-1173.
[8]	黄天佑, 林惠孚, 杨超, 赖涛, 王青松, 黄海风. SAR移频欺骗干扰距离主瓣展宽问题解决方法[J]. 系统工程与电子技术, 2026, 48(4): 1174-1185.
[9]	胡继军, 韩伟, 张国玉, 李洋, 杨洁, 田甜. 基于行为识别网络的SAR工作模式识别方法[J]. 系统工程与电子技术, 2026, 48(4): 1218-1229.
[10]	翟佳, 画麒嘉, 王梓权, 张子恺, 王睿琦, 刘金玲, 胡美琪, 武晨辉. 基于多级别融合的可见光和SAR图像联合识别[J]. 系统工程与电子技术, 2026, 48(3): 817-825.
[11]	李莹莹, 吴昊, 陈凤, 林郁卓越, 王雪莹, 冯鑫. 一种基于天线指向的大方位扫描滑聚模式星载SAR几何定位技术[J]. 系统工程与电子技术, 2026, 48(3): 846-858.
[12]	杨雪莹, 李高鹏, 张云. 语义引导的星载多角度SAR三维重建方法[J]. 系统工程与电子技术, 2026, 48(2): 430-446.
[13]	刘梦莉, 舒高峰, 李宁. 基于多干扰机协同调制的多通道SAR-GMTI欺骗干扰方法[J]. 系统工程与电子技术, 2026, 48(2): 490-502.
[14]	胡继军, 韩伟, 金龙, 周希娃, 张国玉, 周钦月. 雷达导引头的多站组网单比特干扰对抗方法[J]. 系统工程与电子技术, 2026, 48(2): 503-514.
[15]	陈海青, 汪刘应, 刘顾, 王龙, 葛超群, 陈孟州. 合成孔径雷达图像去噪算法研究进展[J]. 系统工程与电子技术, 2026, 48(1): 94-105.

基于最小熵的地基MIMO-SAR幅相校正GPU加速与嵌入式实现

GPU acceleration and embedded implementation for amplitude-phase correction of ground-based MIMO-SAR based on minimum entropy

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 30

相关文章 15

编辑推荐

Metrics

本文评价