Modern infrastructure and a myriad of services rely on positioning and timing information provided by Global Navigation Satellite Systems (GNSS) and in particular the Global Positioning System (GPS). However, given their low received signal power levels, GNSS signals are vulnerable to Radio Frequency Interference (RFI), either from non-intentional or intentional (jamming), sources. Hence, GNSS itself has become a critical infrastructure which must be protected. Since RFI source is unknown a priori, passive localization systems consisting of spatially distributed Sensor Nodes (SNs) are needed to geo-locate the RFI. These systems typically use source Angle of Arrival (AOA), Time Difference of Arrival (TDOA) or a combination of AOA/TDOA measurements which are non-linear in nature, to estimate the RFI position. Also, dynamics associated with the RFI source(s) further complicates the geo-localization process. This paper explores and reports on the use of various Kalman Filters in combining AOA and TDOA measurements for efficient geolocalization and tracking of dynamic and stationary RFI sources based on real measurements from one such geo-localization system. We report on and contrast the geo-localization accuracies and computational complexities of the Extended, Unscented and Single Propagation Unscented Kalman Filters along with the traditional snap-shot approach.