Energy-Efficient UAV Communications with Interference Management: Deep Learning Framework

Abstract

In this paper, an interference-aware energy- efficient scheme for a network of coexisting aerial-terrestrial cellular users is proposed. In particular, each aerial user aims at achieving a trade-off between maximizing energy efficiency and spectral efficiency while minimizing the incurred interference on the terrestrial users along its path. To provide a solution, we first formulate the energy efficiency problem for UAVs as an optimization problem by considering different key performance indicators (KPIs) for the network, coexisting terrestrial users, and UAVs as aerial users. Then, leveraging tools from deep learning, we transform this problem into a deep queue learning problem and present a learning-powered solution that incorporates the KPIs of interest in the design of the reward function to solve energy efficiency maximization for aerial users while minimizing interference to terrestrial users. A broad set of simulations have been conducted in order to investigate how the altitude of UAVs, and the tolerable level of interference, shape the optimal energy-efficient policy in the network. Simulation results show that the proposed scheme achieves better energy and spectral efficiency for UAV and less interference to terrestrial users incurred from aerial users. The obtained results further provide insights on the benefits of leveraging intelligent energy-efficient scheme. For example, a significant increase in the energy efficiency of aerial users with respect to increases in their spectral efficiency, while a considerable decrease in incurred interference to the terrestrial users is achieved in comparison to the non-learning scheme.