Efficient RFID Data Imputation by Analyzing the Correlations of Monitored Objects
Source: Northeastern University
As a promising technology for tracing the product and human flows, Radio Frequency Identification (RFID) has received much attention within database community. However, the problem of missing readings restricts the application of RFID. Some RFID data cleaning algorithms have therefore been proposed to address this problem. Nevertheless, most of them fill up missing readings simply based on the historical readings of independent monitored objects. While, the correlations (spatio-temporal closeness) among the monitored objects are ignored. The paper observes that the spatio-temporal correlations of monitored objects are very useful for imputing the missing RFID readings. This paper proposes a data imputation model for RFID by efficiently maintaining and analyzing the correlations of the monitored objects.