Trajectory Planning and Task Offloading for Delay and Energy Optimization in UAV Corridors

This paper introduces the concept of UAV (Unmanned Aerial Vehicle) corridors as structured, wirelessoptimized flight paths to reduce collision risks and enhance safety in urban environments. To address challenges such as signal interference and collisions at shared altitudes, we formulate the UAV trajectory planning and task offloading problem as a continuous-time integer nonlinear programming problem, considering communication constraints, time delay, task offloading, and energy consumption. A novel Signal-Interference Proximal Policy Optimization (SI-PPO) algorithm is proposed, incorporating a reward mechanism based on signal interference to improve UAV coordination in dense urban areas. Simulation results demonstrate that SI-PPO significantly outperforms traditional methods such as Deep Q-Learning (DQN), Simulated Annealing (SA), Greedy Algorithm, and Deep Deterministic Policy Gradient (DDPG) in terms of trajectory optimization, delay reduction, and energy efficiency.