Answers are grounded in this paper and cite their sources.
In disaster-stricken areas, unmanned aerial vehicles (UAVs) can be employed as aerial base stations (ABSs) to facilitate communication control. However, complex and dynamic scenarios pose significant challenges for UAVs in providing emergency communication services. Previous studies have primarily focused on the fixed position of ground users (GUs) for a certain time interval with fixed scenes, where UAV trajectories are with fixed length. Only algorithms designed based on appropriate environmental models can be better applied to real emergency scenarios. Therefore, this paper introduces the Emergency Scenario Force Model (ESFM), a complex and dynamic environmental model, and proposes a framework that enhances the performance of multi-agent proximal policy optimization-based (MAPPO-based) reinforcement learning approaches in real-world scenarios. Furthermore, we explored different reward design methods for performance improvement. Through comparative analysis of various degrees of randomness and reward functions, we present the results of the generalization test for different networks. The results indicate that the proposed method effectively resolves communication issues in more complex and dynamic environments and demonstrates significant robustness.
No takes yet. Share an insight, caveat, or question.
Qiu et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: