Spatial-Temporal Outlier Sensing over Trajectory Data Streams
The increasing capability to track moving vehicles in city roads enable people to probe the dynamics of a city. In this paper, we address the problem of detecting outliers and anomalies sources with trajectory data. Unlike existing anomaly detection methods, both spatial and temporal information are considered to find the potential outliers. We identify anomalies according to not only the individual traffic regular patterns, but also the consistent traffic behaviors with adjacent road segments on road network. To analyze the major sources of anomalies, we then describe the anomalies diffusion process on the basis of information diffusion model. Furthermore, in order to make detection results not limited to only large scale events, the granularity of our detected traffic anomaly is on the level of road segments instead of spatial regions. Finally, experiments on a very large volume of real taxi trajectories in an urban road network show that the proposed approaches outperform the recent state-of-the-art algorithms.
Keywords: Spatial-Temporal Outlier, Data Streams
Download Full-Text
ABOUT THE AUTHORS
Xian Wu
Chinese Academy of Sciences
Chao Huang
Chinese Academy of Sciences
Lin Chen
Chinese Academy of Sciences
Xian Wu
Chinese Academy of Sciences
Chao Huang
Chinese Academy of Sciences
Lin Chen
Chinese Academy of Sciences