Abstract
Complex event processing (CEP) is a promising technology, which can evaluate the business logic from a large number of continuous real-time data streams. There are a few issues that are still remaining, such as linearization of processing realization mode and event pattern description, which would severely affect the detection efficiency of CEP. This paper proposes a CEP system that can efficiently share the associated event pattern and simultaneously process multiple event sources. First, an event presentation model and a complete set of event operators are summarized, on which a multi event pattern specification language with parallel output capability is proposed. Then, an extended colored Petri-net (CEP-PNm) is defined, which can effectively realize the proposed language and share the same nodes between event patterns in a variety of connection types. At last, by comparison with the traditional CEP system, a parallel processing system called PM-CEPs is introduced. The experimental results show that PM-CEPs can improve complex event processing performance horizontally and vertically through event pattern share and parallel processing.