Specific Emitter Identification Based on Two Residual Networks

In order to solve the scarcity of labeled samples in actual combat scenarios makes it difficult to converge the deep learning model in a complex electromagnetic environment. This paper combines the residual learning model of residual network and complex-valued residual network, we proposes a feature fusion that makes full use of a method that has a small amount labeled data. First, we need input reprocessed data into two residual learning networks. And then, we regard the data features extracted by the two residual networks as the real and imaginary parts of the classifier to train the complex network classifier. Finally, Using signal data of the radiation source to be identified to verify the effectiveness of the algorithm. Experiments have proved that compared with the baseline method, this method could improve the recognition effect of communication radiation sources under the condition of little labeled samples.