A Specific Emitter Identification Method Based on Dual Neural Networks

In response to the challenge of low identification accuracy attributed to the incomplete extraction of radio frequency fingerprint (RFF) features in existing specific emitter identification (SEI) methods, this paper proposes a novel SEI method based on dual neural networks. First, representative modulation signals are selected and directly employed as inputs to a neural network. then, comprehensive RFF features are extracted from the radio frequency (RF) signals utilizing deep learning techniques. Finally, a SoftMax classifier is deployed to make classification decisions based on the extracted RFF features, thereby achieving accurate emitter identification. Experimental results demonstrate the efficacy of the proposed method, demonstrating high identification accuracy across three classes of emitters under two modulation types of RF signals.