Objectives: Design and develop a reference architecture for database-centric analysis of large spatiotemporal data streams comprised of trajectory events.
Background: Efficient content-based analysis of time trails, i.e., streams of (Who, What, When, Where) events, poses new challenges for query optimization. It is the key component to achieve the required scalability into the hundreds of terabytes of events that should be mined and inspected at near real-time requirements.
Description of work: The driving applications are the trajectory streams from TomTom and KNMI. They both call for innovative techniques to handle both bulk loads, enable short circuit responses, and integrate seamlessly with large trajectory repositories. The major task is to tackle the challenge posed by using domain specific optimization techniques with “low-level” dynamic optimization techniques and transparent access to foreign file repositories.
Traditional cost-based query optimization techniques are not applicable in this context as the shape and characteristics of a good query execution plan depend on the actual data distribution. Instead, these characteristics change over time. The query plan is changed accordingly, preferably in an automatic self-organizing way. These optimization techniques form a significant contribution to support the content analysis in WP-1, WP-3. In turn, the instant stream mining techniques as developed in WP-1 and applied in WP-3 will deliver valuable input to novel data-dependent self-tuning optimization techniques in particular of continuous queries over data streams in general.