RF Fingerprint Recognition Based On Spectrum Waterfall Diagram
Di Liu, Mengjuan Wang, Hao Wang · 2021 18th International Computer Conference on Wavelet Active Media Technology and Information Processing (ICCWAMTIP) · 2021
RF fingerprint technology has received great attention and research due to its security and operability. Early research on radio frequency fingerprints mainly revolved around transient signals, and transient characteristics are currently the most mature research category. With the development of wireless communication technology, the existing radio frequency fingerprint identification technology has gradually been unable to meet practical needs. In recent years, with the rapid development of image recognition technology, some scholars have proposed to apply image recognition technology to RF fingerprint identification, thereby converting radio signal recognition problems into image recognition target detection problems, and making full use of advanced image recognition technology to improve equipment recognition accuracy. This paper proposes to convert the original I/Q signal into a spectrum waterfall diagram and input it into the image recognition model for training and recognition. The experimental results show that this method can effectively improve the accuracy of RF fingerprint recognition.