A Framework for Deploying UAV-Based Mobile Base Stations to Enhance Mobile Network Coverage Using Deep Reinforcement Learning

In wireless communications, small cells are low-power base stations used to increase network coverage and capacity in areas where traditional cell towers are impractical or uneconomical. Unmanned Aerial Vehicles (UAVs) are utilized for the rapid deployment of these cells in remote locations, disaster-stricken areas, or crowded events. In this paper, we demonstrate that Mobile Base Stations (MBS) mounted on UAVs can maximize user coverage in affected areas with a minimum number of devices. Due to the limited access of UAVs to network information, a decentralized control strategy is employed, where each UAV uses only local information for optimal ۳D deployment. The optimization problem is NP-hard due to its non-convexity and the coupling of sub-problems, making it unsolvable by standard solvers. By reducing the constraints and formulating the problem, a decentralized reinforcement learning algorithm is developed to increase the channel capacity of the UAVs. This algorithm facilitates the optimal deployment of UAVs using local information and achieves high MIMO channel capacity. Simulations demonstrate the convergence, effectiveness, and computational efficiency of this algorithm, which has been evaluated through detailed analysis, confirming its ability to optimize network coverage and capacity.

کلیدواژه ها:

UAV ، Small Cells ، Coverage Enhancement ، Reinforcement Learning ، Q-Learning