An Ensemble FIR Filtering Framework for Inter-Vehicular Range Estimation Using GNSS Signals

Accurate inter-vehicle range estimation is a critical requirement in Cooperative Positioning (CP) methods. CP methods are preferred over other positioning techniques in location-aware intelligent transportation applications due to increased availability and accuracy of positioning information. In this paper, we propose a new statistical Finite Impulse Response (FIR) filtering framework for estimating the accuracy of inter-vehicle ranges using Global Navigation Satellite System (GNSS) observables. FIR filters have been proven to be better state estimators than traditional Kalman filters in terms of robustness and sensitivity to unknown noise characteristics. However, they are affected by the choice of selecting an optimal value for window size which is usually unknown and involves an intricate procedure to find it. Here, we propose rather a simpler approach to deal with this problem where an ensemble of a priori particles, generated from a Gaussian density function, is used to deal with the window size problem. Only one particle which is closest to current measurement is retained and used in the update step of the filter. The proposed filtering architecture is tested and its enhanced performance is validated using actual field trials data sets.