Abstract: Real time enterprise is able to provide timely, accurate and detailed information deliberately. The information is about the large volume of data from business systems and Smart Items (radio frequency identification, wireless sensor networks), and most importantly, is about the information behind all the data. There are few powerful methods to retrieve and present the behind information, and complex event processing is effective to do it. The architecture of real time enterprise based on Smart Items is firstly put forth. The clear definition of fundamental concepts of event, event model and hierarchical model are given. All the available event composition methods are almost based on event type, the event instance operators here make it possible to figure out complex event in instance level. The requisite event type operators are also defined. Semantic space is put forth to provide more event context information, and the strategies of event instance using, consuming and aging are given to process all the event instances. Especially, the partition methods of event instances improve the efficiency of event composition significantly. To detect composite event, colored Petri net serves as a good candidate in terms of its properties of concurrency and efficiency, and its mechanism and algorithm are discussed. Application of the mechanism shows that it is effective to provide more accurate and concrete information.

Key words: Complex event processing, Petri net, Radio frequency identification, Real time enterprise, Smart Items