Abstract: To improve the accuracy of data cleaning under the circumstances of non-uniform RFID(Radio Frequency Identification) data stream,on the basis of the classical algorithm SMURF(statistical SMoothing for unreliable RFID data),an algorithm was presented based on tag velocity and sliding sub-window to clean RFID data.The method consi-ders the influence of tag velocity on the sliding window adjustment,adjusts the speed dynamically according to the label confidence δ,at the same time further divides the sliding window,and takes statistical sampling for tag data in sub window.The results of the statistical sampling are dealing with the data of entire sliding window together to speed up the detection of tag transition,so that the movement of tag can be determined more accurately.Experimental results show that the algorithm reduces the average error rate and frequency of the phenomenon of positive reading,thereby increasing the accuracy of the data.

Key words: RFID,Data cleaning,Tag velocity,Sliding sub-window,SMURF