UAV 3D Trajectory Design in Space-Air-Sea Integrated Networks for Marine IoT Systems Using Deep Reinforcement Learning

Marine Internet of Things (MIoT) systems have undergone significant development and are becoming increasingly vital in marine environments. Enhancing connectivity is crucial for these systems, and low Earth orbit (LEO) satellites currently play a key role in supporting MIoT systems by providing high reliability and reduced communication delays. However, the curvature of the Earth and the limited capacity and power of MIoT devices can result in weak LEO satellite signals over marine areas. To address this challenge, unmanned aerial vehicles (UAVs) provide an effective solution by bridging the communication gap between LEO satellites and MIoT devices, thereby enhancing connectivity in these regions. However, maintaining reliable communication remains a challenge due to the high mobility of UAVs. To address these issues, this paper introduces space-air-sea integrated networks (SASIN) by optimizing the 3D trajectory of UAVs. In this work, we propose a double deep Q-network (DDQN) approach to enhance the UAVs' trajectory, aiming to maximize communication coverage while guaranteeing satisfactory average data rates for MIoT devices. The simulation results indicate that the proposed method significantly surpasses baseline methods in improving communication coverage while ensuring acceptable average data rates for MIoT devices.