Task-Oriented Transmission and Scheduling for UAV-Based Real-Time Target Tracking in SAGSIN

The Space-Air-Ground-Sea Integrated Network (SAGSIN) offers broad communication coverage, enabling operations in remote regions. However, conventional remote UAV communication solutions require satellite relays, resulting in significant latency that hinders real-time response and decision accuracy for time-sensitive tasks like UAV target tracking and attacking. To address this challenge, we establish a direct communication loop between observation UAVs, the satellite control center, and Reconnaissance-Strike UAVs, reducing reliance on satellite-ground relays. This closed loop ensures continuous control and feedback, making the system highly task-oriented by enabling dynamic adjustments to meet real-time mission demands. Equipped with onboard processing and decision-making capabilities, Reconnaissance-Strike UAVs respond more rapidly and autonomously, enabling quicker, decentralized responses. To ensure data timeliness, we introduce the Age of Incorrect Information (AoII) as a metric to quantify transmission delays, optimizing Observation UAVs' transmission strategies through a Deep Q-Network (DQN). Additionally, Proximal Policy Optimization (PPO) with a task-oriented reward function enhances UAV scheduling. Simulation results demonstrate these strategies significantly improve UAV performance in target tracking and attack, offering a robust solution for complex missions.