PhD Defense on “Optimized Strategies for Adaptive Mobility Control of Unmanned Aerial Vehicles in 5G Networks”

Abstract:

The use of Unmanned Aerial Vehicles (UAVs), com- monly referred to as drones, has grown rapidly recently. Due to their flexibility, mobility and adaptable altitude, they are, for ex- ample, used in various applications of the future generation of 5G networks and beyond. On the one hand, UAVs can be used as Aerial Base Stations (ABSs) to assist the traditional cellular infras- tructure and/or extend its coverage by providing on-demand wire- less communication to ground users. On the other hand, UAVs can also operate as a new type of aerial users that use the existing cel- lular network to allow a wide range of beyond visual-line-of-sight applications, such as surveillance, real-time video streaming and package delivery. However, cellular UAV-driven applications in- cur a non-negligible number of technical challenges which have to be addressed to effectively leverage their potential in the next generation of wireless networks. In this thesis, different solutions are proposed in order to face these challenges and to ensure the proper integration of drones in cellular networks. First, we pro- pose a new framework for the 3D positioning of a swarm of UAVs, acting as ABSs to extend the coverage of ground base stations (GBS) and provide wireless communication links to ground users. We consider a multi-hop backhaul scheme that allows ABSs to remain connected to the core network through the GBSs directly, or other deployed ABSs. The proposed Adaptive Social Spider Optimization (SSO)-based framework uses the SSO metaheuris- tic to compute the optimal positions of ABSs in order to maxi- mize their coverage. Next, the potential of using machine learn- ing techniques to predict users’ mobility and enhance the perfor- mance of the proposed positioning solution is studied. As such, the Transformer model and the LSTM (Long-Short Term Memory- based) encoder-decoder model are adopted to accurately estimate the spatio-temporal distribution of mobile users. The mobility prediction results are exploited in conjunction with the place- ment framework to enhance its coverage and adjust the ABSs po- sitions accordingly. Finally, CaPRM (Connectivity-aware Proba- bilistic RoadMap) is proposed for the trajectory optimization of a cellular-connected UAV in a delivery mission within an urban en- vironment with obstacles. On its trajectory, the UAV must main- tain a good connectivity with the cellular network and minimize the total interference signals that the GBSs receive from the UAV. The performance of the proposed solutions are evaluated and val- idated using rigorous experimental studies by providing in depth simulations that were conducted considering various scenarios.