Efficient CNN-Based Automatic Modulation Classification in UWA Communication Systems Using Constellation Diagrams and Gabor Filtering

Underwater Acoustic (UWA) communication channels are varying in nature due to various underwater conditions. The efficiency of the UWA communication system can be increased by changing transmission parameters over UWA channels based on the state of the channel. The aim of Automatic Modulation Classification (AMC) is to recognize efficiently the modulation schemes from signals received over UWA channels. In this paper, we propose an efficient method that combines equalization, Gabor filtering of constellation diagrams, thresholding and deep convolutional neural networks (CNNs) for modulation classification over the Signal-to-Noise Ratio (SNR) range of -10 to 30 dB in UWA systems. Simulation results of the proposed method indicated an acceptable performance over a wide SNR range.