Multi-Feature Based Recognition of Bionic Underwater Acoustic Bio-LFM Communication Signals

The underwater acoustic Bio-LFM communication signal is characterized by its robustness and prevalence in bionic camouflage underwater acoustic communication. However, certain signals share similar characteristics with Bio-LFM signals, the use of single-domain features for recognition is prone to interference, which subsequently affects the effectiveness of underwater adaptive communication and reconnaissance. In this study, we propose a multi-feature-based recognition method of Bio-LFM signals under multiple types of interfering signals. First, the time-frequency, auto-correlation, and connected component features are extracted to form a feature set. Then, the support vector machine classifier is used for training and optimization to realize the recognition of Bio-LFM signal. The proposed method demonstrates a robust recognition rate of bionic signals, even in the presence of interference from dolphin whistles, sweep spread carrier signals, and linear frequency modulation signals. The test under sea trial channel proves the effectiveness and generalization ability of the proposed recognition method.