This paper examines the performance of the Doppler Particle Filter (DPF) algorithm to locate a push-to-talk handset from the its radio frequency transmissions recorded on-board an aircraft. The technical challenge resides in the mathematical solution to solve for the stationary radio location based on the received Doppler shifted frequency measurements from an emitter with a non-stationary or drifting carrier frequency. The proposed DPF solution is motivated by the failure of the standard techniques (e.g. Gauss-Newton iterative descent technique, Extended Kalman Filter) to this application. The DPF is based on a particular implementation of a particle filter that tracks the carrier frequency at the same time as estimating the radio location and whose approach has been described in previous papers. Its location performance is examined on data collected on both continuous and intermittent radio operation. The search area was 10×10 km and for most of the data collected the closest the sensor got to the target was 2-3 km. For these ranges, the results show median miss distances of 55 m for continuous transmission and 131 m for intermittent transmissions over almost 150 different collections. Its performance is also compared and shown to approach the FOA Cramer-Rao lower bound (CRLB).
展开▼
