To happen or not to happen: towards an open distributed complex event processing system

In order to facilitate development of applications that are real-world aware through arbitrary sensing devices, we develop a complex event processing system (CEP). This CEP is open to any kind of sensor, distributes event processing to optimize resource utilization, supports complex temporal and spatial events, and provides the possibility to query both the complex events as well as deviations from these events. The latter is especially useful for application domains with regular patterns in which a deviation from such a pattern represents an important event. The system reads queries that describe what is expected as normal behavior. This query is distributed to the data sources, like sensors, which are interesting for the query. In this paper, we define the event model and a finite automaton that models the queries. We discuss possibilities of distributing this finite automaton, and propose two causality operators and a syntax for these operators. We also show preliminary results from our first prototype.