| Type | Diploma thesis | | Code | DIPL-2005-12 | | Title | Compressing Trajectory Data of Moving Objects | | Author | Mihalis Potamias | | Year | 2005 | | Keywords | Data Stream, Spatiotemporal, Moving Objects, Compression, Trajectory, Amnesic, Sampling, Sketches | | Abstract | Over the recent years, the spatial database community has focused its research interests on handling moving objects. The scope of this thesis was to study, develop and experimentally evaluate compression techniques for trajectories of moving objects. The main objectives of compression are on one hand the reduction of data volume and on the other hand the fast computation of approximate answers to queries.
Trajectory data falls naturally under the data stream model. Data streams refer to transient rather than persistent relational information. The model sets several specifications, which the compression techniques must meet. The specifications involve storage, process time and response time complexity, as well as quality guarantees for the approximation.
This thesis addressed three challenging topics. Firstly, we developed sampling techniques based on spatiotemporal heuristics that maintain the most characteristic elements of trajectory information. Secondly, an amnesic tree structure was implemented emphasizing on recent data rather than older data as time progresses. Finally, we applied synopses in order to produce fast approximate answers for aggregate queries over positions of moving objects. This was accomplished by combining the amnesic tree with spatial indices and sketches.
In order to assess the previously described techniques we conducted extensive experiments using synthetic trajectory data produced on the road network of Athens, which yielded very promising results. In addition, the expected performance in terms of computation resources and response accuracy was confirmed. The overall conclusion of this thesis was that compression of moving objects’ trajectories is highly advantageous, effectively achieving trade-off between the allocated resources and the desired accuracy. | | Category | Data Streams | | File | View |
Return to Publications Page
|