Research And Application On The Integration Of Complex Event Processing And RFID Middleware

RFID middleware is becoming an essential part in companies' RFID application.However,when it comes to real-time processing and deep analyzing of business relationship behind the large-scale data generated by RFID applications,traditional solutions is losing their magic power.The complex event processing(CEP)can describe relationship among complex events based on the universal pattern of language,and the performance of detecting complex events are enhanced greatly based on the memory based calculation methods.Companies will benefit a lot from the integration between RFID middleware in data preprocessing and CEP in real time detecting of business operation under different patterns.In this paper,we focus on integration between RFID middleware and the CEP technology.An integration schema is designed and established based on the self-developed reconfigurable RFID middleware and related open source utils with higher performance,expandability and usability,which is demonstrated by experiments.The main contributions are listed as below:(1)We propose a extendible communication model based on TCP persistent connection,implemented by Netty,a asynchronous network framework,including heartbeat mechanism,connect authentication mechanism,reconnection mechanism and cache retransmission mechanism,which guarantee the efficiency and stable communication between RFID middleware ALE layer and complex events processing system.(2)We introduce a dynamic rule-based scheduling algorithm in which the rule is described by the Espe's EPL language,and we successfully designed the rule-based scheduling engine with usage of Zookeeper based on the scheduling algorithm and implemented the function of loading and unloading rules dynamically during system runnig.(3)Based on the Storm transactional topology,the integration of RFID middleware and complex evens processing system is implemented in distributed environment.The system includes 4 modules: input adaptation layer,stream grouping strategy layer,complex events processing layer and output adaptation layer.In this paper,each module is designed and implemented separately.(4)We implemente the custom stream grouping approach against the RFID events and rule-based scheduling information,at the same time,extend the fault-tolerance mechanism of Storm,which enables automatic recovery of the previous status information after recovery.In the last part of this paper,a test against the integration schema was conducted,i.e.stability test of communication model,functional test of dynamic loading/unloading rules and performance test of complex events processing.In addition,a case study,the application of the integration plan in a 3rd party logistics inventory program included is proposed.