Modulation Recognition of Underwater Acoustic Communication Signals Based on Frequency-Domain Recurrence Plot and CBAM-ResNet

The modulation recognition of underwater acoustic signals plays a crucial role in various fields such as ocean exploration, military communication, and underwater surveillance. However, challenges such as multipath effects, noise interference, and signal attenuation in underwater channels make modulation recognition particularly difficult. To address these issues, this paper proposes a novel Frequency Domain Recurrence Plot (FDRP) algorithm, which constructs a similarity matrix of frequency domain signals to visualize the spectral characteristics in the form of images. Furthermore, a Convolutional Block Attention Module (CBAM) is incorporated into the traditional ResNet architecture to fully leverage the feature extraction capabilities of convolutional neural networks. Experimental results from lake trials demonstrate that the features constructed based on FDRP exhibit superior recognition performance compared to other methods.