Baojun Hu

Topic

Wireless Channel Characterization and Robust Localization for Autonomous Aerial and Ground Vehicles

Department of Electrical and Computer Engineering

Date & location

• Friday, August 14, 2026

• 10:30 A.M.

• Virtual Defence

Reviewers

Supervisory Committee

• Dr. Xiaodai Dong, Department of Electrical & Computer Engineering, UVic (Supervisor)

• Dr. Hong-Chuan Yang, Department of Electrical & Computer Engineering, UVic (Member)

• Dr. Yang Shi, Department of Mechanical Engineering, UVic (Outside Member)

External Examiner

• Dr. Tho Le-Ngoc, Department of Electrical & Computer Engineering, McGill University 

Chair of Oral Examination

• Dr. Dennis Hore, Department of Chemistry, UVic

Abstract

The proliferation of autonomous vehicles, both aerial and ground-based, has created an urgent need for robust wireless connectivity that can withstand the challenges of real-world operating environments. This dissertation addresses a fundamental physical phenomenon that threatens both communication and localization in autonomous systems: the blockage of radio waves by conductive or dielectric objects. In aerial platforms, the carbon-fiber airframe and rapidly spinning propellers of giant multi rotor unmanned aerial vehicles (GMR-UAVs) create a time-varying self-blockage effect that severely degrades air-to-ground (A2G) and air-to-air (A2A) communication links. In ground environments, walls, furniture, machinery, and other obstacles obstruct the direct propagation paths between mobile robots and fixed anchors, causing non-line of-sight (NLOS) ranging biases that corrupt positioning accuracy, while magnetic field distortions render conventional electronic compasses unreliable indoors.

This dissertation tackles the blockage challenge through two complementary research thrusts. The first thrust which is urban air mobility (UAM) airframe blockage channel characterization, presents the first comprehensive measurement campaigns on a full-scale, carbon-fiber hexacopter. Outdoor experiments quantify the effects of propeller rotation speed and antenna position on received power, root-mean-square (RMS) delay spread, and frame outage probability, revealing an optimal antenna mounting distance. Anechoic chamber measurements produce a high-resolution spatial map of airframe attenuation, which is subsequently integrated into ray-tracing simulations of New York City urban environments. These simulations demonstrate that the airframe dramatically reduces line-of-sight (LOS) probability, elevates the path loss exponent, and increases shadow fading variability, with the strongest effects in high-rise urban canyons and for A2A links.

The second thrust that is infrastructure-based indoor localization for ground vehicles, develops three complementary solutions that are inherently robust to NLOS conditions. The multi-antenna reverse constellation consistency (MACC) algorithm equips a large robot with three ultra-wideband (UWB) antennas and exploits geometric consistency between the virtually reconstructed and the known physical anchor con stellation to simultaneously identify NLOS-contaminated measurements and estimate absolute azimuth, without requiring training data or additional sensors. The Terahertz (THz) extremely rough diffuse surface (TERDS) concept replaces expensive active THz anchors with completely passive patches of rough material, exploiting natural diffuse scattering to achieve millimeter-level ranging accuracy with only a single active source. A learning-based framework introduces a quantitative soft/hard NLOS distinction and demonstrates that three-class classification improves LOS detection precision by over 21 percentage points compared to conventional binary methods.

Collectively, these contributions provide measurement-based channel models, practical design guidelines, and effective algorithmic solutions that advance the state of knowledge toward reliable, blockage-aware wireless connectivity for autonomous aerial and ground vehicles.