The Research Of Algorithm For Uncertain RFID Data Cleansing

RFID (Radio Frequency Identification) technology are applied to many fields for its characteristics of without contact and low cost. However, due to the factors of different environment where RFID tags in and the physical characteristics of the RF signal,the collected raw data by the readers is uncertain.And the data is divided into the following three categories: false positive, false negative and redundant reading.This uncertain RFID data restricts the further development and application of RFID.So how to manage and use the large amounts of uncertain data generating in the RFID systems is one of the key issues that need to be addressed currently in the usage of RFID technology.Existing cleaning technologies include processing data smoothly using the sliding window, cleansing data using the temporal correlation between the target objects,and removing the redundant RFID data and so on.Based on the existing data cleansing technology,this paper mainly make a further deep study the cleaning of false negative and redundant readers.(1) False negative cleansing in data layer.We analyze the classical adaptive sliding window algorithm SMURF and improve it.The improved algorithm takes the impact of the speed of RFID tags on the change of the window size into account,and the sliding window is further divided to better detect the transition phenomenon of tags,which can better adjust the sliding window.Experimental results show that the improved algorithm can improve the accuracy of data cleaning.(2) Redundant readers elimination in data layer.This paper proposes an algorithm based on communication information among readers for the applications that the readers are deployed intensively.This algorithm requires the reader that can directly communicates with adjacent reader.Experimental results show that the algorithm can reduce the time complexity and improve the detection rate of redundant readers.(3) False negative cleansing in logic layer.According to the given application scenario,we establish a probabilistic path event model.Then we improve the model by adding the time factor,and use the time decay model to dynamically update the probabilistic path event model.In order to ensure the accuracy of cleaning results,we define the maximum probabilistic path event to fill the false negative. Experimental results show that the improved algorithm improves the accuracy of the cleansing results.