Performance Analysis of Underwater Acoustic Automatic Modulation Classification System with Machine Learning

Underwater acoustic (UWA) channel is a triple-selective fading channel in time, space and frequency as a result of its lower sound speed and its complicated boundaries in the ocean. It is very difficult to realize a successful classification of the communication signal modulation within a single feature in such harsh channel. The feature extraction is much more sensitive to the environmental parameters and signals. In this paper, an automatic modulation classification (AMC) system based on machine learning is established in order to improve the correction rate of AMC for the several common modulation types. In the case of extracting the feature, the machine learning algorithms, including Support Vector Machine (SVM), k-Nearest Neighbor (KNN), Classification and Regression Tree (CART) are adopted. The performance comparison of these three algorithms is implemented through the simulation data analysis, which would denote how the UWA multipath channel, the signal noise ratio (SNR) affects the performance of the proposed system.