Subject Area

Electrical, Electronics Engineering

Abstract

A collection of unmanned aerial systems (UAS) can be networked as a cooperative wireless sensor array to geolocate an unknown-location RF emitter using time-based measurements. In operation, however, environmental multipath and hardware errors in sensor positioning and timing can degrade emitter localization accuracy and limit the practicality of single-snapshot solutions. This dissertation evaluates time-of-arrival and time-difference-of-arrival (TOA/TDOA) geolocation for cooperative UAS arrays under realistic error sources and develops geometry-control strategies that actively reduce localization uncertainty through iterative UAS repositioning.

This work studies the Location on a Conic Axis (LOCA) method for emitter localization. Using Monte Carlo simulations with hardware error models and environment-induced excess delay, including multipath statistics consistent with ITU-R P.1411, results show that LOCA remains close to the Cramér–Rao lower bound and benefits consistently from increased sensor redundancy. Across the evaluated mission volumes and error budgets, the most significant improvement occurs when increasing from five sensors, the minimum required for 3D localization, to six sensors, with diminishing returns observed beyond approximately eight sensors.

A primary contribution of this dissertation is the use of a subset-based LOCA solution cloud to provide dispersion-based convergence metrics without requiring knowledge of the true emitter location. These uncertainty metrics are then used to guide reinforcement-learning-based geometry controllers that dynamically reposition the UAS array between localization snapshots to reduce uncertainty and drive the solution cloud toward a compact and consistent region. In comparative policy experiments, active geometry adaptation reduces localization uncertainty within a small number of repositioning steps and achieves substantial reductions in localization error relative to static sensing.

Finally, the learned geometry-control behavior is validated in an EXata digital-twin environment that includes Rician fading and adverse weather. The EXata results demonstrate stable convergence and faster refinement than the baseline simulation, supporting the feasibility of adaptive geometry control for cooperative UAS geolocation under more realistic propagation conditions. Collectively, these results show that for small cooperative UAS arrays operating in dense multipath environments, adaptive geometry becomes a primary mechanism for efficient convergence on an unknown emitter location.

Degree Date

Spring 2026

Document Type

Dissertation

Degree Name

Ph.D.

Department

Electrical and Computer Engineering

Advisor

Mitchell A. Thornton

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Download

Share

COinS