联合LeNet和知识图谱的链路建立行为识别方法  

Joint LeNet and Knowledge Graph Approach to Link Building Behavior Recognition

作　　者：刘燕澳 孙佳琛 丁国如 徐以涛 宋叶辉 汤鹏 Liu Yan’ao;Sun Jiachen;Ding Guoru;Xu Yitao;Song Yehui;Tang Peng(College of Communications Engineering,Army Engineering University,Nanjing 210007,China)

机构地区：[1]陆军工程大学通信工程学院,南京210007

出　　处：《航空兵器》2024年第6期112-119,共8页Aero Weaponry

基　　金：国家自然科学基金(62171462,62301604,62231027)。

摘　　要：在电磁对抗环境下,通信行为识别是电磁空间信号挖掘利用的重要一环。针对非合作无线网络中链路建立行为识别检测困难以及传统方法特征表征能力弱的问题,本文提出了联合LeNet和知识图谱的链路建立行为识别方法,该方法在数据预处理的基础上,建立了特征数据库,并使用知识图谱(Knowledge Graph,KG)进行特征的可视化。此外,该方法能够对低信噪比条件下的IQ信号也有很好的识别效果,在实测数据的基础上,信噪比为-20 dB的情况下,平均准确率达到90%以上,能够对链路建立行为进行有效识别。In the electromagnetic countermeasure environment,communication behavior recognition is an important part of the mining and utilization of electromagnetic space signals.In order to solve the problem of the difficulty of link establishment behavior recognition and detection in non-cooperative wireless networks and the weak feature representation ability of traditional methods,this paper proposes a link establishment behavior recognition method combining LeNet and knowledge graph,which establishes a feature database based on data preprocessing and uses knowledge graph to visualize the features.In addition,the method can also have a good recognition effect on IQ signals under the condition of low signal-to-noise ratio,and the average accuracy rate of the signal-to-noise ratio is more than 90%when the signal-to-noise ratio is-20 dB on the basis of the measured data,which can effectively identify the link establishment behavior.

关 键 词：链路建立行为识别 非合作无线网络 LeNet 知识图谱 深度学习 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]