Few-Shot Domain Adaption-Based Specific Emitter Identification Under Varying Modulation
Lijun Yin, Xue Fu, Shengnan Shi, Pengfei Liu, Yun Lin, Yu Wang, Hikmet Sari
Abstract
Specific emitter identification (SEI) is an effective Internet of things (IoT) data flow protection technique of identifying individual emitters via unique characteristics of different emitters. However, deep learning-based methods are difficult to deal with the identification of emitters of variable modulation signals, especially when there are only few samples in the target domain. In order to solve the problem of small-sample variable modulation classification, this paper designs a SEI method based on transfer learning (TL), which freezes some layers on the basis of the CVNN network trained in the source domain, generalizes the transferable general features learned into the new modulation signal recognition task, and uses the Maximum Mean Discrepancy (MMD) metric as the regularization of supervised learning to reduce the distribution difference between the source domain and the target domain in the latent space. Experimental results show that the proposed method has a great improvement in recognition performance compared with the method of separate training and direct transfer of the target domain. Even with very few samples, there is a certain performance guarantee.