基于DSGD的分布式电磁目标识别

doi:10.12305/j.issn.1001-506X.2023.10.04

系统工程与电子技术 ›› 2023, Vol. 45 ›› Issue (10): 3024-3031.doi: 10.12305/j.issn.1001-506X.2023.10.04

• 电子技术 • 上一篇

基于DSGD的分布式电磁目标识别

王宏安¹^,², 黄达³, 张伟¹^,²^,*, 潘晔¹, 王祥丰³, 邵怀宗¹, 顾杰²

1. 电子科技大学信息与通信工程学院, 四川成都 611731
2. 电子信息控制重点实验室, 四川成都 610036
3. 华东师范大学计算机科学与技术学院, 上海 200062

收稿日期:2022-07-17 出版日期:2023-09-25 发布日期:2023-10-11
通讯作者: 张伟
作者简介:王宏安(1995—), 男, 硕士研究生, 主要研究方向为分布式机器学习、辐射源识别
黄达(1996—), 男, 硕士研究生, 主要研究方向为群体智能
张伟(1985—), 男, 高级工程师, 博士研究生, 主要研究方向为电子对抗
潘晔(1972—), 女, 副教授, 博士, 主要研究方向为信号处理
王祥丰(1987—), 男, 副教授, 博士, 主要研究方向为分布式优化、多智能体学习、可信机器学习
邵怀宗(1969—), 男, 教授, 博士, 主要研究方向为通信、雷达信号处理
顾杰(1974—), 男, 研究员, 博士, 主要研究方向为自适应信号处理
基金资助:
电子信息控制重点实验室开放基金;国家自然科学基金(U20B2070);国家自然科学基金(U19B2028)

Distributed electromagnetic target identification based on decentrallized stochastic gradient descent

Hongan WANG¹^,², Da HUANG³, Wei ZHANG¹^,²^,*, Ye PAN¹, Xiangfeng WANG³, Huaizong SHAO¹, Jie GU²

1. School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China
2. Science and Technology on Electronic Information Control Laboratory, Chengdu 610036, China
3. School of Computer Science and Technology, East China Normal University, Shanghai 200062, China

Received:2022-07-17 Online:2023-09-25 Published:2023-10-11
Contact: Wei ZHANG

摘要/Abstract

摘要：

分布式电磁目标识别利用分布式最优化和分布式计算等技术实现传统集中式电磁目标识别。分布式最优化方法结合分布式计算架构实现对最优化问题的分布式求解, 以分布式实现从问题信息、数据到最优目标识别模型的映射。利用去中心化随机梯度下降方法这一经典分布式最优化方法, 建立面向电磁目标识别的分布式计算架构和分布式电磁目标识别方法。实际电磁信号数据验证了所提算法的有效性。在分布式电磁目标识别算法与集中式识别算法性能均保持在90%以上时, 单节点训练时间下降50%以上, 显著提升了训练效率。

关键词: 电磁目标识别, 分布式, 去中心化, 随机梯度下降, 一致性约束

Abstract:

Distributed electromagnetic target identification aims to realize traditional centralized electromagnetic target identification by using technologies such as distributed optimization and distributed computing. The distributed optimization method combines the distributed computing architecture to realize the distributed solution to the optimization problem, and realizes the mapping from the problem information and data to the optimal target identification model in a distributed manner. This paper uses the decentralized stochastic gradient descent, which is a classical distributed optimization method, to establish a distributed computing architecture and a distributed electromagnetic target identification method for electromagnetic target identification. Based on the actual electromagnetic signal data, the effectiveness of the proposed algorithm is verified. When the performance of the distributed electromagnetic target identification algorithm and the centralized identification algorithm remains above 90%, the single node training time decreases by more than 50%, which significantly improves the training efficiency.

Key words: electromagnetic target identification, distributive mode, decentralized, stochastic gradient descent, consistency constraints

中图分类号:

TN971.1

王宏安, 黄达, 张伟, 潘晔, 王祥丰, 邵怀宗, 顾杰. 基于DSGD的分布式电磁目标识别[J]. 系统工程与电子技术, 2023, 45(10): 3024-3031.

Hongan WANG, Da HUANG, Wei ZHANG, Ye PAN, Xiangfeng WANG, Huaizong SHAO, Jie GU. Distributed electromagnetic target identification based on decentrallized stochastic gradient descent[J]. Systems Engineering and Electronics, 2023, 45(10): 3024-3031.

图/表 9

图1

图2

图3

图4

表1

表2

图5

图6

图7

参考文献 31

1	YUAN K , LING Q , YIN W T . On the convergence of decentralized gradient descent[J]. SIAM Journal on Optimization, 2016, 26 (3): 1835- 1854. doi: 10.1137/130943170
2	DE LUIGI C , JAUFFRET C . Estimation and classification of FM signals using time frequency transforms[J]. IEEE Trans.on Aerospace and Electronic Systems, 2005, 41 (2): 421- 437. doi: 10.1109/TAES.2005.1468738
3	WANG Y , GUI G , GACANIN H , et al. An efficient specific emitter identification method based on complex-valued neural networks and network compression[J]. IEEE Journal on Selected Areas in Communications, 2021, 39 (8): 2305- 2317. doi: 10.1109/JSAC.2021.3087243
4	ZHANG Q , GUO Y , SONG Z Y . Dynamic curve fitting and BP neural network with feature extraction for mobile specific emitter identification[J]. IEEE Access, 2021, 9, 33897- 33910. doi: 10.1109/ACCESS.2021.3060794
5	ZHA X , CHEN H , LI T Y , et al. Specific emitter identification based on complex fourier neural network[J]. IEEE Communications Letters, 2022, 23 (3): 592- 596.
6	WAN T , JI H , XIONG W N , et al. Deep learning-based specific emitter identification using integral bispectrum and the slice of ambiguity function[J]. Signal, Image and Video Processing, 2016, 2009- 2017.
7	QU L Z , LIU H , HUANG K J , et al. Specific emitter identification based on multi-domain feature fusion and integrated learning[J]. Symmetry, 2021, 13 (8): 1481. doi: 10.3390/sym13081481
8	HOU K K , LI N . Specific emitter identification based on CNN[J]. Journal of Physics Conference Series, 2021, 1971 (1): 012014. doi: 10.1088/1742-6596/1971/1/012014
9	张伟, 王沙飞, 林静然, 等. 基于孪生网络的电磁目标跨模式识别算法[J]. 电子学报, 2022, 50 (6): 1281- 1290.
	ZHANG W , WANG S F , LIN J R , et al. Cross-modal recognition algorithm of electromagnetic targets via siamese network[J]. Acta Electronica Sinica, 2022, 50 (6): 1281- 1290.
10	蒋季宏, 方宇轩, 张伟, 等. 小样本情景下基于特征融合的辐射源个体识别[J]. 电子信息对抗技术, 2022, 37 (3): 20- 25.
	JIANG J H , FANG Y X , ZHANG W , et al. Specific emitter identification based on feature fusion in few-shot scene[J]. Electronic Information Warfare Technology, 2022, 37 (3): 20- 25.
11	范广伟, 蔚保国, 晁磊, 等. 卫星导航电磁干扰识别分类器设计[J]. 系统工程与电子技术, 2014, 36 (2): 234- 238.
	FAN G W , YU B G , CHAO L , et al. Design of interference recognition and classification filter of satellite navigation electromagnetic environment[J]. Systems Engineering and Electronics, 2014, 36 (2): 234- 238.
12	TANG X J, CHEN W G, ZHU W G. Radar emitter recognition method based on AdaBoost and decision tree[C]//Proc. of the 2nd International Conference on Automation, Mechanical Control and Computational Engineering, 2017: 341-345.
13	HUANG Y K , JIN W D , GE P , et al. Radar emitter signal identification based on multi-scale information entropy[J]. Journal of Electronics & Information Technology, 2019, 41 (5): 1081- 1091.
14	王磊, 张志勇, 胥辉旗, 等. 基于VMD分解和多域联合分布的雷达辐射源识别[J]. 系统工程与电子技术, 2023, 45 (8): 2479- 2488.
	WANG L , ZHANG Z Y , XU H Q , et al. Radar emitter identification based on VMD and multi-domain distribution[J]. Systems Engineering and Electronics, 2023, 45 (8): 2479- 2488.
15	王枭, 何怡刚, 马恒瑞, 等. 面向电网辅助服务的虚拟储能电厂分布式优化控制方法[J]. 电力系统自动化, 2022, 46 (10): 181- 188.
	WANG X , HE Y G , MA H R , et al. Distributed optimization control method of virtual energy storage plants for power grid ancillary services[J]. Automation of Electric Power Systems, 2022, 46 (10): 181- 188.
16	HUO X, LIU M X. A novel cryptography-based privacy-preserving decentralized optimization paradigm[C]//Proc. of the IEEE International Conference on Pervasive Services, 2021: 607-612.
17	STIPANOVIC D M, TOMLIN C J. Decentralized optimization, with application to multiple aircraft[C]//Proc. of the IEEE Conference on Decision and Control, 2002, 1: 1147-1155.
18	ZHAO H S, CANNY J. Sparse allreduce: efficient scalable communication for power-law data[EB/OL]. [2022-06-29]. http//arxiv.org/abs/1312.3020.
19	AWAN A A, CHU C H, SUBRAMONI H, et al. Optimized broadcast for deep learning workloads on dense-GPU infiniband clusters: MPI or NCCL?[C]//Proc. of the 25th European MPI Users'Group Meeting, 2018.
20	PATARASUK P, YUAN X. Bandwidth efficient all-reduce operation on tree topologies[C]//Proc. of the IEEE International Parallel & Distributed Processing Symposium, 2007.
21	PATARASUK P , YUAN X . Bandwidth optimal all-reduce algorithms for clusters of workstations[J]. Journal of Parallel & Distributed Computing, 2009, 69 (2): 117- 124.
22	SERGEEY A, DELBALSO M. Horovod: fast and easy distri-buted deep learning in TensorFlow[EB/OL]. [2022-06-29]. https://arxiv.org/abs/1802.05799.
23	YANG Q , LIU Y , CHENG Y , et al. Federated learning[J]. Synthesis Lectures on Artificial Intelligence and Machine Learn-ing, 2019, 13 (3): 1- 207.
24	BOYD S , PARIKH N , CHU E , et al. Distributed optimization and statistical learning via the alternating direction method of multipliers[J]. Foundations & Trends in Machine Learning, 2010, 3 (1): 128.
25	CHANG T H , HONG M , WANG X . Multi-agent distributed optimization via inexact consensus ADMM[J]. IEEE Trans. on Signal Processing, 2014, 63 (2): 482- 497.
26	李瑞泽, 张双辉, 刘永祥. 基于卷积ADMM网络的高效结构化稀疏ISAR成像方法[J]. 系统工程与电子技术, 2023, 45 (1): 56- 70.
	LI R Z , ZHANG S H , LIU Y X . Computational efficient structural sparse ISAR imaging based on convolutional ADMM- net[J]. Systems Engineering and Electronics, 2023, 45 (1): 56- 70.
27	TUAN P V , KOO I . Distributed ADMM-based approach for total harvested power maximization in non-linear SWIPT system[J]. Wireless Networks, 2020, 26 (5): 1357- 1371.
28	XU T , WU W C . Accelerated ADMM-based fully distributed inverter-based volt/var control strategy for active distribution networks[J]. IEEE Trans. on Industrial Informatics, 2020, 16 (12): 7532- 7543.
29	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
30	TOLSTIKHIN I, HOULSBY N, KOLESNIKOV A, et al. MLP-mixer: an all-MLP architecture for vision[C]//Proc. of the NeurIPS, 2021.
31	VASWANI A , SHAZEER N , PARMAR N , et al. Attention is all you need[J]. Advances in Neural Information Processing Systems, 2017, 30, 2243- 2251.

训练方法	优化器	损失函数	学习率	批量大小
集中式	SGD	交叉熵	0.01	128
分布式	DSGD	交叉熵	0.01	128

训练方法	单节点样本数量	神经网络	平均训练时间/s	内存占用/G	训练集准确率/%	测试集准确率/%
集中式	126 000	ResNet101	647	4.15	99.97	99.94
集中式	126 000	MLP-mixer	371	4.51	99.80	98.47
分布式(3节点)	42 000	ResNet101	316	1.36	99.60	93.96
分布式(3节点)	42 000	MLP-mixer	157	1.52	92.58	91.93
分布式(4节点)	31 500	ResNet101	229	1.02	99.29	93.55
分布式(4节点)	31 500	MLP-mixer	106	1.14	93.26	92.79
分布式(5节点)	25 200	ResNet101	158	0.82	99.45	92.83
分布式(5节点)	25 200	MLP-mixer	67	0.91	92.40	91.81

[1]	刘文波, 姚翼荣, 张弓, 胡文. 超维计算概念、应用及研究进展[J]. 系统工程与电子技术, 2023, 45(7): 1938-1956.
[2]	郝宇航, 蒋威, 王增福, 兰华, 雍婷, 潘泉. 分布式MIMO体制天波超视距雷达仿真系统[J]. 系统工程与电子技术, 2023, 45(7): 1981-1989.
[3]	钱寒雨, 田晓曼, 肖冰, 曹璐, 王锐, 吴西伟. 分布式卫星集群博弈任务均衡分配方法[J]. 系统工程与电子技术, 2023, 45(6): 1797-1804.
[4]	孙田野, 孙伟, 吴建军. 改进Quatre算法的无人机编队快速集结方法[J]. 系统工程与电子技术, 2022, 44(9): 2840-2848.
[5]	胡泰洋, 张晋宇, 卢海梁, 李鹏飞, 李一楠, 吕容川. 基于数据融合的分布式综合孔径微波辐射高分辨率成像算法[J]. 系统工程与电子技术, 2022, 44(8): 2403-2409.
[6]	孙林, 毛忠阳, 康家方, 张磊. 基于最大化能效的海上中继通信频谱分配算法[J]. 系统工程与电子技术, 2022, 44(8): 2661-2667.
[7]	李鹏飞, 卢海梁, 韩涛, 党鹏举, 李一楠, 李浩, 吕容川. 基于漏波天线的分布式微波辐射计[J]. 系统工程与电子技术, 2022, 44(7): 2125-2133.
[8]	刘红亮, 陈超, 黄昆伟, 卢博, 岳凯. 针对分布式雷达网的多帧联合航迹起始方法[J]. 系统工程与电子技术, 2022, 44(7): 2143-2147.
[9]	曾守桢, 胡英杰. 基于高度残缺信息的分布式语言信任网络群决策方法[J]. 系统工程与电子技术, 2022, 44(6): 1907-1919.
[10]	高天旸, 胡笑旋, 夏维. 基于分布式协同进化的星座自主任务规划算法[J]. 系统工程与电子技术, 2022, 44(5): 1600-1608.
[11]	刘嘉琛, 董磊, 赵长啸, 陈泓兵. 基于形式化方法的DIMA动态重构仿真与验证[J]. 系统工程与电子技术, 2022, 44(4): 1282-1290.
[12]	刘高赛, 姜兴龙, 李华旺, 梁广. 基于位置感知的大规模LEO星座分布式路由算法[J]. 系统工程与电子技术, 2022, 44(11): 3529-3536.
[13]	王华华, 李延山, 张志豪, 梁家敏. 基于期望传播的分布式信号检测算法[J]. 系统工程与电子技术, 2022, 44(11): 3564-3570.
[14]	靳鹏, 李康. 基于改进合同网协议的分布式卫星资源调度[J]. 系统工程与电子技术, 2022, 44(10): 3164-3173.
[15]	罗哲, 权婉珍, 张朴睿, 杨小冈. 单边Lipschitz非线性多智能体系统一致性追踪控制[J]. 系统工程与电子技术, 2022, 44(1): 279-284.

基于DSGD的分布式电磁目标识别

Distributed electromagnetic target identification based on decentrallized stochastic gradient descent

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 31

相关文章 15

编辑推荐

Metrics

本文评价