MSTM: A novel map matching approach for low-sampling-rate trajectories

Map matching is an important technique that matches user trajectories to the real road networks on a digital map. It is crucial and has been more and more widely used in various fields, such as route plan, traffic forecast, location-based services and so on. However, most existing algorithms are less effective when applied to low-sampling-rate trajectories. In this paper, we propose MSTM, a novel approach that can effectively match the low-sampling-rate trajectory to road networks. In MSTM, we first partition the trajectories into trajectory segments according to the stay points. Then we construct a map-searching tree by conditional extend and prune operations, which contains all the candidate paths. Finally, by considering the spatial and temporal information of trajectories, we evaluate each branch path in the map-searching tree and choose the one with the highest score as the result. Extensive experiments on real-world datasets show that, compared with two classic approaches, MSTM outperforms ST-Matching and IVMM in terms of matching accuracy as well as efficiency.