Research On The Multi-Queries Approach Over Uncertain RFID Event Streams

In recent years, RFID (radio frequency identification) technology are widelyused in security, logistics, transportation and many other fields, their attention is alsomore and more high. As for the RFID technology research with this trend is ceaselessand thorough, especially the RFID event stream query problem has become a researchhotspot at present. the original data that reading from RFID readers formats to thebasic flow after cleaning, and according to certain logical semantic the basic eventson the flow form a composite event to support more advanced applications, RFIDevent stream query is a query to the system of registration after continuous query wayto extract these complex events, in order to improve the query efficiency, we mustcarry on the optimization of query sharing. In addition the incomplete and inaccuratein the original data and the uncertainty of event semantics and the occurrence timecause the RFID system uncertainty, which due to the event semantic structurecomplexity, the composite event query for the event semantic uncertainty processingbecomes a difficult problem.Aiming at the uncertainty RFID complex event processing system query sharingproblems (including the composite event uncertain problems) are studied in thispaper, with particular attention to a fixed work flow properties of RFID applicationenvironment query sharing problem. For the cost of ordinary looking for the samequery and combining the same sub query In general RFID system query sharingprocess, this paper presents a fixed work flow under the uncertain composite eventquery and probability computing method: For the characteristics of a fixed the workflow of the RFID application environment, we present the form of a work flow eventtree including all possible complex events in the work flow; Through the analysis onthe semantic relationships between the flow nodes, we propose a algorithm to convertthe flow event tree representation for Bayesian network structure, in order to realizethe event complex semantic probability calculation and reasoning, and solve in thequery tree query sharing problem; According to the work flow of change, we present the update method of Bayesian network structure; Finally, we propose the compositeevent query and probability calculation methods based on the Bayesian networkstructure, including the storage structure of the Bayesian network graph, the creationof query index table corresponding to the flow nodes, the methods of query markingthat registered in the system and the marks removal after the end of the query, withthe aid of the Bayesian network structure and node corresponding to the conditionalprobability table to achieve event probability calculation and reasoning methods andcomplete incident query algorithm after event match.Through related experiments can show that, the proposed method cansignificantly improve the RFID composite event query efficiency, at the same timethe query can obtain higher recall and precision.