Reducing vehicle tracking data volume through a network-based approach

The extended availability of GPS makes it possible to keep track of vehicles'movement for various applications in transportation. An approach is proposed to reduce tracking data volume through representing these data with feature points. Feature points can represent tracking data with much less storage space and without a loss of major characteristics than sample points or raw tracking data. A combination of three criteria, i. e. network, velocity, and direction, is used to select feature points from sample points. Network criterion, a novel concept proposed by this paper, is defined as a change of the locational relationships between vehicles and the underlying road network. Experimental and analytical results demonstrate that the proposed approach outperforms existing ones.