Key points are not available for this paper at this time.
This paper explores the unmanned aerial vehicle (UAV)-assisted maritime Internet of Things (MIoT) system, where the UAV departs from an initial position, flies to an optimal hovering position to transfer energy and collect data from marine IoT-targets (MITs), and then transmits the data packets to the marine base station (MBS). In this system, MITs face challenges such as unstable network connections and limited power. To address these challenges, MIoT focuses on low-power technologies, wireless power transfer (WPT), and energy harvesting (EH). Furthermore, to improve spectral efficiency in maritime environments, the non-orthogonal multiple access (NOMA) technology is employed for data transmission between the MITs and the UAV. We formulate an optimization problem designed to minimize energy consumption while adhering to age of information (AoI) constraints, where AoI measures the freshness of information. The energy consumption is influenced by the number of EH time slots and the number of data collection time slots, necessitating a joint optimization approach. We transform the non-convex problem into a Markov decision process (MDP) and introduce a twin delayed deep deterministic policy gradient-based behavior cloning (TD3-BC) algorithm, designing the corresponding state space, the action space, and the reward function. The simulation results demonstrate that the proposed TD3-BC algorithm outperforms other benchmark algorithms. In addition, it is found that as the UAV velocity increases, the energy consumption initially decreases and then rises again. Moreover, energy consumption increases as the AoI threshold decreases or the number of MITs increases.
Hu et al. (Mon,) studied this question.