Research On RFID Data Cleaning Method

RFID (Radio Frequency Identification) technology is now widely used in many fields, such as supply chain, asset tracing, health care, for its small size and no needing to contact. Whereas the unreliable data restricts the RFID technology from being farther developed. The three kinds of unreliable data in RFID are: false negative, false positive and redundant readings.This paper studied the RFID cleaning method and then proposed a cleaning method focus on dealing the three kinds of unreliable data and mobile tags, the work are as follows:This paper proposed a three-layered architecture—KFBC, to solve the three kinds of unreliable reading: the reader layer deals with the false negative and false positive reading from individual readers; the proximity group layer is to solve the redundant RFID readers; and the data table layer is to eliminate the redundant RFID data.In reader layer, this paper proposed a cleaning method based on Kalman Filter, to solve false negative and false positive reading from individual readers, and the transition time respond delay problem of mobile tags: to form the KAL-RFID update equation, the KAL-RFID method use the measuring value and the estimated value, which is generated from the observation and prediction equations; to accurate measurement value, this paper proposed a method by filtering false negative reading below threshold in a read cycle; and to get the estimated value, this paper proposed a method of fitting history reader rate based on the least squares fitting method, and propose a method to select fitting data.In the proximity group, to eliminate redundant reader, this paper proposed a method by composing the adjacent reader into one proximity, and removes the reading from illegal reader according the read rate, and proposed a effect algorithm to solve the frequent switching problem. In the table layer, this paper proposed a method to deal with the issue of data redundancy, by using four-tuple , and gives a method to determine Time_In and Time_Out.In the end, this paper proposed the concepts of Accuracy and Transition Respond Time to value the cleaning results. As experiment shows the reasonable and the efficiency of the KAL-RFID method we proposed.