A Complex Event Processing Language And Its Implementation Method

With the popularity of cloud computing and digital information technology in every industry and real-time monitoring system has been widely used, these systems will produce various types of event information constantly. These events usually have low monomer value, but these events will produce a lot of valuable information by the paradigmatic and specific rules analyzing. Since most of the value of real time data will decrease over time, it is required to find the target information in the event stream accurately and efficiently according to the detection rules. The object of the Complex Event Processing(CEP) technology is to detect the event sequence in time and accurately, CEP can meet the requirement of the real-time event detection system.CEP is one of the key technologies to support large data processing. CEP continuously check the event information that meets the requirements from the event stream by event driven thinking, which detection rules are formulated according to the temporal relations of events and paradigmatic relation. CEP also needs to adapt the requirements of low delay, high throughput of large amounts of data processing. At present, many researchers proposed many kinds of CEP languages and CEP engines, event detection capability of these languages have more or less limitations, and most of the CEP engines only supports a single execution environment, cannot use more computing resources in a cluster so that these engines has performance bottleneck.In order to enhance the capability of complex event detection in distributed environment, a new type of complex event processing language CEStream is designed. This language takes semi-structured stream data as data model, and also provides a feature for complex event detecting, which named regular tree pattern matching. This feature support structural connection and regular expression matching. Moreover, by aiming at distributed multi-sources of event stream, CEStream provides another feature that matches each pattern matching result once again by regular expression which described time sequence for each result. This feature supports the demand of multi-sources composite complex event detecting, and has the good ability of event processing. To implement CEStream language, a distributed processing engine system is developed, which based on stream data processing cluster and remote querying agent. This system provides a feature that uses remote querying agent to detect events by regular tree pattern, and provides another feature that uses stream processing cluster to process multi-sources composite complex events. Experiments show that this system can achieve the CEStream language, which effectively limits the amount of communication between the nodes, and makes full use of the computing power of the cluster. The event detection capability and the overall performance of this system meet the goal of complex event processing technology, and this system can meet the application requirements of real-time event detection system.