This thesis presents a new method that is able to efficiently select subsets of sub-trajectories from a set of network-constrained vehicle trajectories. The method can be used to improve the quality of histogram-based travel-time estimates for paths in the underlying road network by identifying the trajectories that are most relevant for a given road network path and time.
The thesis shows this with a series of detailed qualitative studies based on real-life trajectory data sets capturing several years of vehicular travel in Northern Denmark. We observe that the quality of travel-time estimates depends heavily on the type of road and whether the road is located in a rural or an urban area.
After demonstrating the quality improvements provided by our method over previous approaches, we design a network-constrained trajectory index to efficiently apply our proposed collection method to compute travel-time histograms. The proposed in-memory index utilizes methods from string processing as a spatial index to identify trajectories in the indexed data that traverse a given path.
All interested parties are welcome. After the defense the department will be hosting a small reception in cluster 3.