Deep Reinforcement Learning Based Reliable Data Transmission Scheme for Internet of Underwater Things in 5G and Beyond Networks

Abstract

The Internet of Underwater Things (IoUT) is an interconnected communication ecosystem for underwater devices in maritime environments. IoUT devices range from seabed sensors to ships and boats in oceans which help in diverse practical applications of underwater information ranging from marine transportation, navigation, and underwater exploration to disaster prevention with intelligent monitoring. There exist significant obstacles within the present architecture of the IoUT. This article proposes three distinct methodologies aimed at enhancing the robustness and reliability of the communication architecture for 5G and future networks. Firstly, the employment of the Deep Reinforcement Learning approach is advocated for the identification of anomalies and faulty sensors. Secondly, energy consumption is identified as a crucial factor for IoUT devices. To address this concern, the integration of renewable energy sources, namely oxygen rechargeable batteries, with nano-generators is employed. This integration allows for the replenishment of the batteries through the harnessing of kinetic energy from water. Thirdly, the process by which information can be efficiently transmitted from IoUT sensors located on the seabed to the final terrestrial network is delineated, with an emphasis on minimizing the number of hops required. Additionally, a case study is presented, focusing on the mitigation of interference within the IoUT. Lastly, the article delves into the various research challenges and unresolved matters that persist within the realm of IoUT.