基于差分等势星球图的通信辐射源个体识别方法

Individual Identification Method of Communication Radiation Sources Based on Differential Contour Stellar Images

为了准确地识别和认证物联对象,阻止用户身份假冒和设备克隆等问题的发生,提出一种基于差分等势星球图的通信辐射源个体识别方法;通过差分等势高星球图的射频指纹特征提取,在不对接收机的载波频率偏差和相位偏差进行估计和补偿的情况下,获取较稳定的通信辐射源个体(发射机)射频指纹;对同厂家、同型号、同批次的20个无线保真网卡设备进行识别测试。结果表明:差分等势星球图通过点密度特征可以恢复一定低信噪比时的星座图丢失的统计特征,更全面地描述信号的细微特征;在相同的深度卷积神经网络模型架构下,相较于传统的基于星座图的统计图域方法,所提出的方法可以在很大程度上提高识别准确率。

To accurately identify and authenticate objects of internet of things,and prevent occurrence of problems such as user identity impersonation and device cloning,a method for individual identification of communication radiation sources based on differential contour stellar images was proposed.Through radio frequency fingerprint feature extraction of differential contour stellar images,stable radio frequency fingerprint of the communication radiation source individual(i.e.transmitter)was obtained without estimating and compensating carrier frequency deviation and phase deviation of the receiver.Identification tests were carried on 20 wireless fidelity network card devices with the same manufacturer,the same type,and the same batch.The results show that differential contour stellar images can restore statistical characteristics lost in planisphere under a certain low signal-to-noise ratio by using point density feature,and can describe subtle characteristics of signals more comprehensively.Under the same deep convolutional neural network model architecture,compared with traditional planisphere-based statistical graph domain methods,the proposed method can greatly improve identification accuracy.