基于SDAE_SVDD的通信辐射源个体开集识别方法  

Individual Open Set Recognition of Communication Emitter Based on SDAE_SVDD

作　　者：刘志文 陈旗[1] 满欣[1] LIU Zhiwen;CHEN Qi;MAN Xin(School of Electronic Engineering,Naval University of Engineering,Wuhan 430034,China)

机构地区：[1]海军工程大学电子工程学院,武汉430034

出　　处：《电子信息对抗技术》2023年第4期26-31,共6页Electronic Information Warfare Technology

摘　　要：针对现有通信辐射源个体识别研究在遇到开集问题时识别性能不高的问题,提出了一种基于堆栈去噪自编码器和支持向量描述(Support Vector Data Description,SVDD)的开集识别方法。该方法通过堆栈去噪自编码器实现降噪和特征压缩提取,将特征输入SVDD进行通信辐射源个体开集识别实验。结果表明,在不同开放度下,该方法可以将未知通信辐射源个体和已知通信辐射源个体以高准确率区分出来,进而将开集识别转为闭集识别。同时,对已知通信辐射源个体识别有很好的识别准确率和抗噪声能力。Aiming at the problem that the existing research on specific identification of communication emitter has low identification performance when encountering open set problems,an open set recognition method based on stack denoising autoencoder and support vector data description(SVDD)is proposed.Noise reduction and feature compression extraction are realized through stack denoising autoencoder.Features are put into SVDD for individual open set identification experiments of communication emitter.The results show that under different degrees of openness,the method can distinguish unknown communication emitter individuals and known communication emitter individuals with high accuracy,and then turn the open set identification into closed set identification.Meanwhile,the known communication emitter individuals can be identified with good identification accuracy and anti-noise ability.

关 键 词：辐射源个体识别 开集识别 深度学习 自编码器 SVDD 

分 类 号：TN911.7[电子电信—通信与信息系统]