Deep Learning Based Modulation Classification for Underwater Optical Wireless Communication System

Abstract

The Internet of Underwater Things (IoUT) is an emerging technology that facilitates communication and data sharing among underwater equipment. The constrained data transfer capacity and frequent transmission failures in underwater wireless communication channels provide significant issues in the context of IoUT. Modulation categorization is crucial for optimizing spectrum allocation, guaranteeing dependable and adaptable communication, mitigating interference, assuring network security, and enabling various applications in underwater optical wireless communication (UOWC). It has a key role in enhancing the efficiency and user-friendliness of UOWC systems. Deep learning (DL), is an effective classification method that has achieved significant success in various fields of application. Nevertheless, its application in underwater optical wireless systems has not been thoroughly investigated. This work focuses on the utilization of DL in UOWC systems, specifically for the purpose of modulation categorization. A Convolutional Neural Network (CNN) is employed to do the classification task. We transform the unprocessed modulated signals into constellation images with a grid-like structure and then input them into a CNN for training the network. The simulation results demonstrate that the suggested modulation classification strategy based on CNN, provides a comparable level of accuracy in classification without requiring manual selection of features.