Deep Reinforcement Learning-Based Task Offloading in Multi-access Edge Computing for Marine IoT

The recent surge in maritime activities has led to a significant demand for marine Internet of Things (M-IoT) devices. These devices are responsible for meeting strict requirements in resource-limited and complex maritime network environments. Advanced communication and computing solutions are imperative for addressing these challenges. Leveraging 6G-based multi-access edge computing (MEC) offers the potential to effectively process large-scale marine data, thereby meeting diverse needs across various marine application scenarios. This paper introduces a latency-sensitive and energy-efficient task offloading scheme (DLES) anchored on deep reinforcement learning (DRL). This scheme was designed to optimize task offloading in a maritime MEC setting, aiming to reduce delays and energy consumption. Simulation results validated the superior performance of the proposed scheme in terms of reduced latency and enhanced energy efficiency.