A greedy data matching for vehicular localization with temporal-spatial weighting factor

In vehicular localization, there is an increasing interest in the fusion algorithm between the relative local sensing estimate (LSE) of ranging sensors and the absolute remote sensing estimate (RSE) of a GPS receiver. A challenging issues of the fusion algorithm is to find a matched pair corresponding to the same vehicle between two sets of estimates. To aim this, we present a greedy data matching (GDM) that finds an optimal set of pairs, consisting of a sensing estimate and a GPS estimate, based on the closeness of these estimates. To represent the closeness in spatial domain, we employ the Mahalanobis distance metric based on the position and the speed estimates. To cope with the temporal randomness of these estimates, we also present a moving-average weighting factor which can provide a robustness to the random error of the estimates. From the numerical results, we show that the GDM outperforms the existing schemes in terms of the correct matching probability.