Font Size: a A A

Research And Implementation Of RFID Data Cleaning Algorithm Based On Middleware

Posted on:2018-01-04Degree:MasterType:Thesis
Country:ChinaCandidate:W W ShenFull Text:PDF
GTID:2348330536479927Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Due to the performance of RFID equipment is vulnerable to the surrounding environment(such as electronic equipment,metal appliances can influence on the radio-frequency signal),RFID data collected is unreliable,and its accuracy is only 60% to 70% according to the relevant research.The unreliability is mainly characterized by the following four categories: false positive,false negative,dirty data and redundant reading.False negative and redundant reading are the main source of RFID data reliability.Large unreliable data is transmitted to hypervisor of the enterprise,which causes serious burden to the system.RFID middleware system as independent levels share the processing of the data,and data cleaning technology is the key technology of middleware for RFID data preprocessing,and it is one of the key problems that RFID technology needs to solve urgently.This thesis focuses on the problems of false negative and redundant reading of the RFID data in the middleware.The existing data cleaning technologies of false negative are the sliding window time smooth algorithm and online pipeline frame cleaning technology;data cleaning techniques of redundant reading mainly include RRE algorithm and LEO algorithm.Based on the above algorithm,this thesis makes a deep research and some improvements.The main work of this thesis includes:(1)For redundant-reader elimination of the intensive deployment of the reader environment,combined with the layered eliminate optimization technology and neighboring coverage density algorithm,this thesis proposed redundant-reader elimination algorithm L-NCD.Simulation results show that more redundant readers are eliminated and the stability of the algorithm is better.(2)For false negative cleaning in the RFID middleware,we analyse the classic adaptive sliding window algorithm SMURF and improve it.The novel proposed data cleaning algorithm bases on dynamic performance of tags,which combines the dynamic variation of tags and the analysis of confidence.Experimental results show that the proposed method has the lowest average error rate compared with other algorithms.(3)A RFID middleware prototype system is designed and implemented using JAVA.The key modules of the system are analyzed in detail.Finally,the L-NCD algorithm and VSMURF algorithm are applied to the system and we present the system experiment results.
Keywords/Search Tags:RFID, middleware, data cleaning, redundant-reader elimination, sliding window
PDF Full Text Request
Related items