Multi-UAV Trajectory Control for AoI and Energy Efficiency

The patent implements dynamic trajectory optimization for UAVs using deep Q-network reinforcement learning. The system continuously adjusts UAV flight paths in real-time based on changing network conditions, user distribution, and energy states. This dynamic adaptation allows UAVs to maximize coverage area while minimizing energy consumption by avoiding redundant movements and optimizing patrol routes according to actual network demands rather than following fixed predetermined paths.

The system dynamically changes multiple parameters including UAV trajectory coordinates, flight speed, altitude, and communication power transmission levels. By adjusting these parameters in real-time based on network conditions and energy constraints, the system optimizes the balance between coverage area and energy consumption. The deep Q-network learns optimal parameter combinations to maximize network utility while staying within battery capacity limits.

The patent introduces a ground base station as an intermediary that assists the UAVs in autonomous navigation. The ground station provides reference information, trajectory suggestions, and coordination to multiple UAVs, reducing the computational burden on individual UAVs. This distributed architecture where the ground station handles complex coordination tasks while UAVs execute localized navigation simplifies the overall system complexity while maintaining autonomous operation capabilities and extensive coverage.

The patent divides the large-scale coverage area into multiple trajectory points or waypoints that UAVs visit sequentially. This segmentation transforms the complex problem of continuous area coverage into a series of simpler navigation tasks between discrete points. The deep Q-network learns optimal sequences of visiting these segmented points, making the autonomous control problem more manageable while still achieving comprehensive area coverage.