An Interference Signal Recognition Method Based on Feature Extraction and YOLOv8 in Low INR

To enhance the recognition capability of wireless communication interference signals, this paper proposes an interference signal recognition method based on feature extraction and YOLOv8. First, mathematical models of 11 common types of interference signals are simulated. Then, a feature extraction method based on projection transformation is introduced to construct 2D image datasets of interference signals, generated by bispectrum transform (BT), Hilbert-Huang transform (HHT), and short-time Fourier transform (STFT) respectively. Finally, the YOLOv8 classification network is used for interference signal recognition. Experimental results show that the average recognition accuracy of STFT features is consistently higher than that of other features. Moreover, the proposed method based on STFT features and YOLOv8 attains a recognition rate of 100% when the interference-to-noise ratio (INR) exceeds 4 dB, outperforming other principal methods.