A potential weakness in RFID-based Internet-of-things systems

In recent years, a large body of research has been devoted to the security and privacy of RFID that is expected to become a critical component of IoT (Internet of Things). Most of these studies have been conducted under the assumption that an RFID system consists of the following elements: RFID tags, a reader and a back-end server. However, in IoT scenario it is supposed that a high density of RFID readers will be deployed and networked to the system over the Internet. Hence, a multi-reader RFID environment circumstance, where readers may be mobile handsets like mobile phones, should be involved in the security analysis of RFID based IoT systems. In this paper, we point out that RFID authentication protocols in the IoT need new security mechanisms that consider untrustworthy RFID entities, compromised readers or insecure communication channel between the readers and the back-end servers. Thus, traditional RFID security schemes designed for closed-loop systems cannot fulfill security and privacy demands, if they are directly adapted to the IoT environment. To emphasize this discrimination, we demonstrate that a secure protocol in a closed-loop RFID system may jeopardize the security of the system in this new RFID concept. Furthermore, we address this fault by investigating the security of a recent IoT RFID authentication protocol, named as AKE-MRFID. We exploit security flaws that have gone unnoticed in the design and present three attacks: de-synchronization, replay and reader impersonation attacks. To defend against the aforementioned attacks, we amend the protocol with a stateful variant so that it holds the claimed security properties. (C) 2014 Elsevier B.V. All rights reserved.