Performance Enhancement in Underwater Optical Wireless Communication using Deep Reinforcement Learning Approach

Recent years have witnessed increased studies focusing on underwater optical wireless communication, also known as UOWC. This is due to the proliferation of new uses for this technology, including military defense, ecological control, and marine ecological preservation. A developing method for direct transmission over the water surface is UOWC. Nevertheless, because of restricted energy supplies and an architecture that is relatively variable due to water flow motion, it is challenging to offer low-consumption and dependable routing in UOWC because of the high directionality of optical rays and the harsh marine conditions. It is necessary to provide enhanced communication for UOWC. As a result, a deep reinforcement learning (DRL) strategy is suggested to improve underwater communication. To better respond to a changing ecosystem and extend system longevity, the network is initially represented as a deep decentralized network, with residual power and connection reliability considered while designing the routing mechanism. The effectiveness of the proposed system is assessed by comparison with existing systems. Several assessment standards, including packet delivery ratio, error rate, energy consumption, and computation time. The findings from the simulation demonstrate that the proposed system can deliver dependable communication reliably and effectively.