GROUND USER LOCALIZATION IN UAV-BASED WIRELESS NETWORKS: FROM CHANNEL CHARACTERIZATION TO SYSTEM DESIGN

Unmanned aerial vehicles (UAVs) are widely used in numerous applications thanks to their excellent mobility and fast deployments. One promising application is to let UAVs serve as flying base stations for providing reliable and cost-effective wireless communications. Such UAV-assisted communications are interesting for extending the range of an existing network, serving remote areas, or temporarily replacing cellular infrastructure in disaster scenarios. Unlike traditional terrestrial cellular networks where base stations are fixed to the ground and serve nearby cells, a UAV base station can be moved and deployed on demand. Its service coverage area is entirely dependent on the location of the UAV itself. This characteristic leads to a key challenge for UAV-assisted communications: the system should localize ground users before deploying UAV base stations to ensure optimal capacity and energy efficiency.In this thesis, we focus on an advanced localization technique that has a great potential to help UAV platforms to localize ground targets based on their emitted radio signals. This so-called “virtual antenna array” technique uses bearing measurements to estimate the Direction-of-Arrival (DoA) of the targets. Unlike conventional DoA estimation methods that require large, expensive antenna arrays, the virtual antenna array is implemented with a mobile single-antenna receiver. While the receiver is moving and receiving signals simultaneously, the DoA can be estimated by capturing the phase interferometry of the intercepted signal at several locations along the receiver trajectory.The first part of this thesis is to verify the suitability of channels in UAV-assisted communication for the application of DoA-based localization methods. We investigate channel characteristics of UAV-assisted communications by performing extensive simulations using ray-tracing simulations. The results show that the angular spread of the UAV platform is very limited, and the power-weighted average angle of all multipath is close to the actual azimuthal angle. Both features are beneficial for DoA estimation, which proves that the application of virtual array methods on UAV platforms is very promising.We then focused on improving the accuracy and the robustness of the proposed method. The main difficulties of the virtual array are two-fold: 1) estimate the relative positions of the receiver during its movement to reconstruct the array manifold that is necessary for array signal processing; 2) address the phase distortion caused by the Local Oscillator (LO) frequency offset between the transmitter and the receiver. The first challenge is addressed by implementing inertial navigation with inertial sensors, and the second challenge is addressed by the two frequency offset compensation methods presented in this thesis. We then perform a comprehensive nonlinear observability analysis for the system to provide guidelines for designing feasible and user-friendly receiver trajectories. The proposed system is implemented on a software-defined radio testbed and tested in an indoor environment. Finally, the proposed virtual array system is implemented and evaluated in UAV-based communication networks. We have performed extensive simulations to investigate the DoA estimation accuracy by considering different system configurations. Parameters including the LO quality, the frequency offset compensation method, the SNR, and the UAV tra- jectories are evaluated. Simulation results show that the virtual array can be a promising technology for UAVs to localize ground targets.Doctorat en Sciences de l'ingénieur et technologie