A RF Fingerprint Recognition Method Based on Deeply Convolutional Neural Network

Radio frequency (RF) fingerprinting is a process of identifying a device or annunciator. Because of the specificity of the wireless transmitting circuit and the imperfections of the components that make up the transmitting circuit, it is possible to extract the RF fingerprint used to identify the wireless transmitter from the wireless signal. At present, the traditional RF fingerprint identification method is performed based on a preset determination formula, and there are problems such as high prior information requirement and limited application range. To solve these problems, this paper proposes a RF fingerprint recognition method based on convolutional neural network (CNN). The research focuses on three aspects: RF fingerprint extraction, convolutional neural network design and wireless transmitter identification and verification. The result proves that this method can complete the wireless transmitter identification well.