Research On The Data Integration Technology For Massive RFID Data Based On RFID Middleware And The Application In Decision Analysis Of Modern Logistics

Radio frequency identification (RFID) technology has been widely used in industry,medical care and transportation and so on,when the RFID tags and readers are widely used,some urgent problems will surely emerge.For example,different equipments from different enterprises will obviously own the different interfaces.How to unify the interfaces between application systems and equipments,how to handle and effectively utilize the vast amouts of data generated by RFID system and so on.If these problems aren’t resolved properly,the development of RFID technology will be strictly obstructed.RFID middleware is developed for this problem and how to develop an effectively RFID middleware is the main focus for enterprises.Another important issue is how to process the massive path data generated in data mining field of RFID.The most important work in quick access technology of the RFID Database is mining the frequent path to supply the information of object movement patterns for people, like mining the data of frequent paths.This paper carried on deep research in RFID path data, mined the movement path with higher frequency and offered advantageous help for users to understand the information of the object’s moving trend and rule.With the wide application of RFID technology, mining RFID trajectory data becomes very important.Mining RFID trajectory data can show the object potentially moving trend,so far,there have been a lot of existing trajectory clustering algorithms,but most of these algorithms are designed for static data,and similar to RFID technology in the field of track data,these algorithms are impractical.Therefore,this paper carried on deep research in RFID trajectory data.The main contributions and innovations of this dissertation are as follows:1) The paper designs and implements a set of RFID middleware using agents and XML,by solving the problems of packing the different interfaces of bottom equipments and isomerous data,using agent as a surrogate,providing unified programming interface and intergrating isomerous data between equipments using XML.2) In order to prove the feasibility of RFID middleware that we have designed,we apply it for the existing RFID data management system. Whenever the items get into the reader radio frequency range,the reader can read automatically labeled data,and transform the heterogeneous data into XML format.It can be compatible with different RFID reader,which prove the feasibility and validity of RFID middleware that we have designed.3) In order to provide better services for users or decision makers of corporation,we propose the algorithm for mining frequent paths based on MP-tree.The algorithm is implemented on the tree which avoided a large number of candidate itemsets generated and reduced the number of scanning database.On the basis of movement patterns,we put forward the corresponding prediction algorithm.According to the movement patterns,we can predict the next location,next arrived time,and even both of them together.4) We put forward the trajectory clustering algorithm aimed at incremental properties of RFID trajectory data,each new path is simplified into a directed linear segments in order to find clusters of trajectory subparts,set of similar trajectory line segments are stored in partition-clusters which take much smaller space than raw trajectories.The integrity-clustering is performed on the set of partition-clusters rather than on all trajectories over the whole time span. The integrity-clusters are generated efficiently to show clustering results of trajectories.