Multi-scale feature extraction and feature selection network for radiation source identification

Convolutional neural nelworks currently applied to radiation source identification process the time-series IQ(in-phase and quadralure-phase) signals in two ways; one way transforms them inlo images, and the other way extracts shallow features of the IQ time-series data. The former way leads to a large computalional efforl of the algorithm, while the latter way leads to a low accuracy of the recognition rate. To address the above problems, a multi-scale feature extraction and feature selection network was proposed. After inputling the IQ signal, the shallow and multi-scale features of the IQ signal were extracted by the multi-scale feature extraction network. Then the data dimension of multi-scale features was reduced by the feature selection network. Feature enhancement was achieved by the adaptive linear rectilicalion unit, and a single fully connected layer was used to classify the radiation source. Comparison experimenls with ORACLE, CNN-DLRF and IQCNet on the FIT/CorleXlab radio frequency fingerprint recognition datasel show lhat the proposed network improves the recognition accuracy and reduces the computational effort to some extent. © 2024 National University of Defense Technology. All rights reserved.