Urgency-Aware QoE-Driven UAV Trajectory Planning for Multi-Priority Post-Disaster Networks via Deep Reinforcement Learning
In disaster scenarios where terrestrial communication infrastructure is compromised, Unmanned Aerial Vehicles (UAVs) provide a rapid solution for restoring connectivity. However, conventional trajectory planning methods often treat users uniformly, neglecting heterogeneous urgency requirements in emergency environments. This paper proposes an urgencyaware, Quality of Experience (QoE)-driven Deep Reinforcement Learning (DRL) framework for UAV trajectory optimization in multi-priority post-disaster networks. We introduce a configurable priority scaling factor that augments the standardized Mean Opinion Score (MOS) model to explicitly incorporate the influence of urgency on user satisfaction. Additionally, a minimum service-duration mechanism enforces sustained prioritization for critical users. By encoding urgency requirements directly into the satisfaction model, the proposed framework enables the learning-based policy to adaptively and flexibly optimize the trajectory under dynamic disaster conditions. Simulation results demonstrate significant improvements in responsiveness to critical users while maintaining overall network effectiveness.