Finite-State-Machine Based Distributed Complex Event Processing In Wireless Sensor Network

Wireless sensor networks (WSNs) have provided a new means of perception of thephysical environment, which leads the following problems, how to analyze the datastreams generated by the WSNs and how to identify and deal with the significant eventsfrom the above streams to be the focuses of recent research. Complex Event Processing(CEP) is a technology which can rapidly extract and identify meaningful information fromthe continuous data stream. Most commonly used CEP frameworks are centralized, inwhich the sensor nodes are only used for data collection, simple filtration and transmission,and data is processed on the server later. There are many disadvantages for this kind ofCEP frameworks: node resources are not fully utilized (including storage, computingpower, etc.), huge amount of data need to be transferred in the WSNs, CEP response isslow, and etc.Finite-State-Machine (FSM) based CEP framework for WSN adopts distributedarchitecture, in which some CEP tasks such as event filtering and composing are migratedto the sensor nodes to reduce the amount of data transferred in WSN and accelerate theresponse speed. To support event hierarchy, the network topology is tree-based in thisframework.The CEP engine running on the sensor node implements FSM based complex eventprocessing. This kind of engine provides FSM based model defining interface for users todefine their business logic and corresponding CEP processing processes. When eventoccurs, CEP engine will automatically identify and respond to complex events based onthe current business state and pre-defined CEP FSM logic, thus complete the CEP tasks. Inaddition, the complex event as a result of event composing performed by a CEP engine ina certain level can be the input event for the other CEP engine in a higher level, whichsupports event hierarchy. Finally, the function and performance test on a prototype showsthat the FSM-based sensor network CEP framework has achieved the intended functionand predetermined performance.