Spectrogram-based Frequency Hopping Signal Detection in a Complex Electromagnetic Environment
Zhe Deng, Jing Lei · 2022 7th International Conference on Signal and Image Processing (ICSIP) · 2022
Due to the non-stationary property of frequency hopping (FH) signals, this paper presents a spectrogram-based detection algorithm in a complex electromagnetic environment. The detection algorithm aims to correctly distinguish whether each pixel in the spectrogram is noise or signal, which is similar to semantic segmentation tasks. Hence, we use DeepLabv3 + net to segment the spectrogram of FH signals and compare this method with constant false alarm rate (CFAR) detection. Simulation results demonstrate the comprehensive performance of the DeepLabv3+ net in spectrogram segmentation, which is suitable for practical implementation of FH signal detection.