Skip to Main Content
This paper examines the ability of particle filters to provide accurate Doppler-based frequency of arrival (FOA) geolocation of radio frequency (RF) emitters. Most existing non-differential Doppler geolocation techniques simplify their geolocation solution by assuming that the emitter's carrier frequency is unknown but stable over the course of the triangulation. This assumption is often violated by today's commercial devices whose applications allow for significant carrier frequency drift, with the result of erroneous FOA solutions. The proposed approach uses particles to discretely represent a state's hypothesized emitter location and conditionally updates the particle's associated frequency drift based on that location and the observations. The performance of this approach is examined for the case of a relatively slow-moving unmanned aerial vehicle (UAV). The results show it is significantly more accurate and robust than Newton's iterative gradient descent techniques, and closely approaches the FOA Cramer-Rao lower bound (CRLB) for location estimation.