Process-level localization of RFID tags using probabilistic models

The reliable tracking of goods in logistic applications is of great importance in order to maintain an efficient operation and supply chain visibility. RFID technology is a potential enabler by offering bulk detection of tagged products even without a direct line of sight connection throughout the supply chain. However, tracking in RFID enabled supply chains is difficult due to the uncertainties introduced by noisy RFID observations. In this paper we present a probabilistic model for an RFID driven business process that can be used to estimate the location of products in the supply chain based on observed RFID read events and prior knowledge of the underlying business process. We justify the chosen model using real-world data and present simulation results of a typical problem in RFID driven supply chains.