Improving the quality of RFID Data by Utilising a Bayesian Network Cleaning Method

Abstract

Radio Frequency Identification (RFID) is a technology used to identify automatically a cluster of objects within a specified parameter. This technology has promised a means to cut cost of time and money in manual labor and to allow greater efficiency in numerous workplaces. However, there are various problems such as missed readings which hinder wide scale adoption of RFID systems. To this end we propose a system that utilises a Bayesian Network applied at a Deferred stage to impute and restore missed readings. Experimental results have shown that the optimal random threshold is 15/% and that the DefBayNet method improves missed data restoration process when compared with the state-of-the-art method.