Fair Communication Coverage and Energy Efficiency Optimization of Multiple UAVs Based on Deep Reinforcement Learning

Unmanned aerial vehicles (UAVs) function as airborne base stations, improving communication networks in challenging scenarios like remote area emergency responses. Drone swarms facilitate data transmission by creating downlinks to ground users within a defined range. However, due to the limited energy and communication range of UAVs, especially in large areas, continuous coverage is challenging, and longduration flight is not feasible. This makes it difficult to control UAV swarms for extended periods to achieve communication coverage while minimizing energy consumption. To tackle this, we introduce MADDPG, a deep reinforcement learning (DRL) algorithm designed to manage UAV swarms by enhancing energy efficiency, coverage, and connectivity. The algorithm maximizes energy efficiency and improves the ability to understand the environment and make decisions using neural networks. Simulation results demonstrate that MADDPG significantly surpasses greedy algorithms and random methods in terms of coverage, equity, and energy utilization.