基于深度聚类的通信辐射源个体识别方法

针对非合作通信条件下缺少标签数据的通信辐射源个体识别问题,提出了一种基于深度聚类的通信辐射源个体识别方法。利用自编码器网络强大的特征提取和数据重构能力对原始I/Q数据进行表征学习,提取个体识别的指纹特征,同时将表征学习过程和特征聚类过程进行联合优化,使表征学习和特征聚类契合度更高,更好地完成无标签条件下的通信辐射源个体识别。通过对5种ZigBee设备采集的信号进行实验,结果表明在信噪比高于0 dB时,可以达到85%以上的识别准确率,证明了本文方法的有效性和稳定性。

A Communication Emitter Identification Method Based on Deep Clustering

Aimed at the problem that individual identification of communication radiation sources has a certain lack of label data under conditions of non-cooperative communication, a method of individual identification of communication emitter is proposed based on deep clustering. The powerful feature extraction and data reconstruction capabilities of the auto-encoder network are utilized for carrying out the representation learning of the original I/Q data, extracting the fingerprint features of individual recognition, and jointly optimizing the representation learning process and the feature clustering process, so as to achieve a higher fit between the representation learning and the feature clustering, and complete still greater individual identification of the communication emitter without labels. The experimental results show that the recognition accuracy is more than 85% when the SNR is above 0 dB. And the proposed method is valid and stable.