Dynamically Planning Path for UAV-Aided Communication Rescue with Nested Monte Carlo Search

The UAV-aided wireless communication system has been playing a more and more crucial role in disaster rescue scenarios. The system needs to optimize, allocate, and dynamically adjust the computational resources to fit the evolving demands of the rescue site and provides computation task offloading services to ground users. The UAV with mobile wireless station acts as a mobile computing center to achieve these. Therefore, planning the flying path of the UAV becomes a challenge due to the dynamically changed demands of the rescue site. To this end, we apply Nested Monte Carlo Search (NMCS) to enhance the efficiency and response speed for the wireless communication rescue system by dynamically planning optimal path for the UAV, given that NMCS can be well adapted to dynamic scenarios. This approach aims to optimize the average throughput of the system with constraints like fairness and energy consumption. Experimental results show that NMCS can significantly improve the average throughput of the system and the rewards compared to the state-of-the-art algorithm, namely Monte Carlo Tree Search (MCTS).