Distributed Task Scheduling and Edge Collaboration for Network Reliability Improvement in LEO enabled Smart Grids

Low Earth orbit (LEO) satellites has been foreseen as a promising paradigm to enable reliable connectivity in the Internet of Things (IoT) enabled smart grids. Considering the limited coverage of the traditional base station (BS), in this paper, we put the edge computing servers on the LEO satellites to realize ubiquitous service coverage and wireless transmission in the power grid environment. However, the inherent high mobility of LEO satellites poses challenges in maintaining stable communication links and efficiently managing the heterogeneous network resources among a myriad of IoT devices, including those in smart grids. To address the dynamic challenges posed by the high mobility of LEO satellites, we have meticulously evaluated several deep reinforcement learning algorithms. Ultimately, we opted for the Proximal Policy Optimization (PPO) algorithm due to its robustness and adaptability in handling such dynamic environments. Our simulation results substantiate the efficacy of the PPO-based solution, showcasing its ability to not only enhance the overall performance of LEO-enabled IoT networks by significantly reducing user latency and energy consumption but also to substantially improve the efficiency of heterogeneous network management.