Anti-reconnaissance method for wireless communication signals based on AdvGAN
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
万方数据
深度学习技术凭借其强大的特征提取能力,在信号认知方面取得广泛应用,这对有保密需求的无线通信系统的保密性安全带来极大威胁.针对上述问题,提出一种基于对抗生成网络(generating adversarial examples with adversarial net-works,AdvGAN)的无线通信信号反侦察方法.首先实现两种不同的调制信号识别模型;再使用3种对抗样本生成方法构造伪装信号;最后叠加在原始信号上并在调制信号识别模型上进行测试.实验结果表明,所提方法能够使侦收方的智能调制识别模型的识别准确率大幅下降,在信噪比10 dB条件下,使侦收方未知模型识别准确率下降约66%,从而有效反制侦收方的智能识别模型.
Deep learning technology has been widely used in signal recognition by virtue of its powerful feature extraction capability,which brings great threat to the confidentiality security of wireless communication systems with confidentiality needs.To address the above problems,this paper proposes a wireless communication signals anti-reconnaissance method based on generating adversarial examples with adversarial networks(AdvGAN).Firstly,two different modulated signal recognition models are realized.Then three antagonistic sample generation methods are used to construct camouflage signals.Finally,they are superimposed on the original signals and tested on the modulated signal recognition model.The experimental results show that the method proposed in this paper can make the recognition accuracy of the intelligent modulation recognition model of the reconnaissance side drop dramatically,and make the recognition accuracy of the unknown model of the reconnaissance side drop by about 66%under the condition of SNR is 10 dB,so as to effectively counteract the intelligent recognition model of the reconnaissance side.
modulated signal recognitiondeep learningcommunication anti-reconnaissancesignal camouflagegenera-tive adversarial networks
NETL
NSTL
万方数据
