Security and Privacy for Data Mining of RFID-Enabled Product Supply Chains

The e-Pedigree used for verifying the authenticity of the products in RFID-enabled product supply chains plays a very important role in product anti-counterfeiting and risk management, but it is also vulnerable to malicious attacks and privacy leakage. While the radio frequency identification (RFID) technology bears merits such as automatic wireless identification without direct eye-sight contact, its security has been one of the main concerns in recent researches such as tag data tampering and cloning. Moreover, privacy leakage of the partners along the supply chains may lead to complete compromise of the whole system, and in consequence all authenticated products may be replaced by the faked ones! Quite different from other conventional databases, datasets in supply chain scenarios are temporally correlated, and every party of the system can only be semi-trusted. In this paper, a system that incorporates merits of both the secure multi-party computing and differential privacy is proposed to address the security and privacy issues, focusing on the vulnerability analysis of the data mining with distributed EPCIS datasets of e-pedigree having temporal relations from multiple range and aggregate queries in typical supply chain scenarios and the related algorithms. Theoretical analysis shows that our proposed system meets perfectly our preset design goals, while some of the other problems leave for future research.