Enhancing the Performance of UAV Communications in Cellular Networks

Abstract

The use of cellular networks as a communication infrastructure for Unmanned Aerial Vehicles (UAVs) has become the current trend. This would mainly enable beyond visual line-of-sight applications and allow UAVs to benefit from the latest evolutions achieved in cellular networks.Despite the advantages that cellular networks can bring to UAVs, several issues still need to be addressed. Indeed, cellular networks are deployed to serve ground user equipment (UEs), whereas UAVs' aerial communications are characterized by different channel conditions. Field evaluations have shown that flying UAVs can experience poor link quality, or even negatively affect ground communication. In addition, UAV applications can be deployed in a challenging environment characterized by different types of QoS (Quality of Services). For instance, UAVs can be deployed to provide network connectivity to ground devices, whereas each one of the latter requires sending two types of traffics with different QoS, at the same time. Furthermore, the consideration of cellular networks for UAVs can bring more opportunities that merit to be explored to enhance the communications, mainly in terms of taking advantage of the presence of several UAVs and Mobile Network Operators (MNOs).

The main objective of the dissertation is to contribute to enhancing the performance of UAV communications in cellular networks. The contributions of this dissertation can be divided into six categories. First, as aerial communication presents different channel conditions, we are interested in modeling UAV communications in cellular networks and deriving expressions that define the performance indicators. All the contributions build from these expressions, and target performing network optimization to enhance UAV communications in cellular networks. Second, we consider a cellular network deployed to serve both UEs and UAVs, and we investigate their co-existence by enhancing their underlying performances. Next, we focus on supporting the co-existence of several QoS types in UAV communications. To this end, we consider the scenario where UAVs are deployed to provide network connectivity to ground devices, where each one of the latter requires two different QoS types. In the fourth and the fifth categories, we focus on new opportunities that cellular networks can bring to UAV communications. In particular, we investigate the possibility of taking advantage of the presence of several UAVs and MNOs in a way to enhance the performance of UAV communications. Finally, we explore the use of machine learning in order to enable fast optimization and enhance the performance of UAV communications.

All the contributions of this dissertation have been validated with a series of performance evaluations.