How to detect cloned tags in a reliable way from incomplete RFID traces

Cloning of RFID tags may lead to considerable financial losses and worse reputation in many commercial applications, while being attractive for adversaries. One way to address tag cloning is to use the visibility that RFID traces provide to detect cloned tags as soon as they enter the system. However, RFID traces always represent historic events without giving certainty where the traced objects currently really are. Furthermore, imperfect read rates can lead to missing reads. As a result, the visibility is not always perfect, which makes detection of cloned tags harder and less reliable. This paper presents a series of probabilistic techniques to enable reliable detection of cloned tags in cases where the visibility is incomplete. Our hypothesis is that the events generated by cloned tags cause rare or abnormal events that can be detected when the process that generates the legitimate events is understood. The presented techniques are studied in a comprehensive simulation study of a real-world pharmaceutical supply chain. Our findings suggest that reliable detection of cloned tags is possible if missing reads are addressed and the supply chain is precisely modeled.