Learning-Aided UAV 3D Placement and Power Allocation for Sum-Capacity Enhancement Under Varying Altitudes

Unmanned air vehicle (UAV) as an aerial base station (ABS) has attracted the attention of cellular service providers to enable emergency communications. However, the unplanned multiple ABS deployment poses severe interference challenges that degrade the user's performance. To maximize the system sum capacity, we propose the use of K-means and Q-learning assisted 3D ABS Placement and Power allocation algorithm (KQPP). Specifically, we combine the benefits of K-means and Q-learning algorithms to achieve this goal. As a result, we successfully improve the sum capacity by satisfying all the users' minimum data rate requirements. The proposed approach achieves 6bps/Hz and 16bps/Hz higher sum-capacity gain compared to equal power allocation and particle swarm optimization (PSO)-based power allocation schemes, respectively.

Automated summary

This paper proposes a 3D ABS placement and power allocation algorithm assisted by K-means and Q-learning, aiming to improve the system's total capacity. The proposed approach achieves significant performance gains compared to other schemes.