A Low Complexity Feature Extraction for the RF Fingerprinting Process

Feature extraction, as an important part of the RF fingerprinting process, aims to extract the subtle features that can reflect the difference of different radio frequency(RF) devices. How to reduce the sample dimension, reduce the test training time while ensuring the classification recognition rate is an important aspect of its research. In this paper, the wavelet coefficients are extracted from the RF transmit signal. By combining the ReliefF algorithm with the PCA algorithm, the signal dimension is reduced to 29%. Experimental results show that the proposed method can effectively reduce the data dimension and time consumption while ensuring the SVM classification accuracy, which it is specially suitable for applying in lightweight identification of the terminal environments.