A Novel Approach for Reducing RFID Uncertainty Using Variational Bayesian Inference

Abstract

recently, the applications based on Radio frequency identification (RFID) technologies are increasing. RFID technology helps applications to grow because of the low price of RFID components. The huge amount data collected by RFID may have redundancy and lost data because of quality of data is very low. So, cleansing RFID data is necessary to reduce the number of redundant and lost data. In this paper, we propose a mean-field variational Bayesian inference based on a 3-state RFID model that finds the nearest optimal density distribution to RFID posterior distribution. Our model takes advantage of redundancy and prior knowledge of tags location to improve data quality and take into account the environment constraints of target application to raise the accuracy of cleaned data. Our proposed technique efficiently manages and reduce uncertainty of RFID data in large scale traceability networks. Our implementation validates the proposed approach, and the experimental results show its effectiveness.