A Small-Object Detection Based Scheme for Multiplexed Frequency Hopping Recognition in Complex Electromagnetic Interference

A significant challenge in Frequency hopping (FH) communications is accurately identifying the target FH signals within the multiplexed spectrum. This letter proposes a recognition scheme that combines time-frequency analysis with deep learning to overcome the degradation of recognizing multiple FH signals amidst complex electromagnetic interference. The scheme utilizes the synchroextracting transform to obtain the spectrogram of the mixed signal by aggregating the energy, then employs RFRM-CenterNet with enhanced small-object detection capability to extract the features and to precisely identify FH signals. Simulation results demonstrate that the scheme improves the accuracy and robustness of multiplexing FH recognition.