Research Article
A Survey of Few-Shot Learning for Radio Frequency Fingerprint Identification
@INPROCEEDINGS{10.1007/978-3-030-90196-7_37, author={Hao Li and Yu Tang and Di Lin and Yuan Gao and Jiang Cao}, title={A Survey of Few-Shot Learning for Radio Frequency Fingerprint Identification}, proceedings={Artificial Intelligence for Communications and Networks. Third EAI International Conference, AICON 2021, Xining, China, October 23--24, 2021, Proceedings, Part I}, proceedings_a={AICON}, year={2021}, month={11}, keywords={Radio frequency fingerprint Few-shot learning Meta-learning}, doi={10.1007/978-3-030-90196-7_37} }- Hao Li
Yu Tang
Di Lin
Yuan Gao
Jiang Cao
Year: 2021
A Survey of Few-Shot Learning for Radio Frequency Fingerprint Identification
AICON
Springer
DOI: 10.1007/978-3-030-90196-7_37
Abstract
With the development of the Internet of Things technology, the radio frequency (RF) fingerprint identification technology of wireless communication equipment has also risen, providing new ideas for network security and RF perception systems. The existing RF fingerprint identification technology is mainly based on traditional machine learning or deep learning. In the face of small sample data or data imbalance, the classification effect is not satisfactory. Therefore, in this paper, we propose the use of Few-Shot Learning (FSL) to solve the problem of radio frequency fingerprint small sample recognition. We review the current RF fingerprint identification technology and FSL methods. What’s more, we analyze some available methods from two aspects. (i) From the perspective of data, the samples of RF signal training data set can be expanded manually or by using transformation function, can also be generated by generative model; (ii) From the perspective of algorithms, prior knowledge can be used to train the new model through fine-tuning, metric, and meta-learning. Finally, we look forward to the challenges and opportunities that the RF fingerprint identification technology may face from theory and application.
