基于Welch功率谱和卷积神经网络的通信辐射源个体识别

针对低信噪比条件下通信辐射源个体识别率低的问题,提出了一种基于Welch功率谱和卷积神经网络的通信辐射源个体识别方法.构建了由20个基于ZigBee协议的物联网设备组成的测试平台,将ZigBee信号前同步码部分的Welch功率谱数据作为辐射源指纹特征送入卷积神经网络进行分类.该方法在低信噪比条件下很好地保留了辐射源的指纹特征,结合卷积神经网络强大的微特征提取能力,对辐射源进行了有效分类.实验结果证明,在瑞利信道及低信噪比条件下,所提方法的识别效果明显优于其他方法.

Communication transmitter individual identification based on Welch power spectrum and convolution neural network

To solve the problem of low individual recognition rate of communication transmitter under low signal to noise ratio(SNR),a method for individual recognition of communication transmitter based on Welch power spectrum and convolutional neural network(CNN) is proposed.A test platform consisting of 20 ZigBee devices of the Internet of Things is built,and Welch power spectrum data of the preamble code of the ZigBee signal is sent into the CNN as the fingerprint features of the transmitter for classification.The method preserves the feature integrity of the fingerprint of the radiation source under low SNR,and makes use of the powerful ability of micro feature extraction of the CNN to effectively classify the radiation source.Experimental results show that the proposed method is better than other methods in Rayleigh channel and low SNR condition.