Our society has become increasingly dependent on electronic information exchange between personal devices and the cloud. Unfortunately, the number of identity and secure information leaks is on the rise. Many of the security breaches are due to insecure access channels to the cloud. The security problem is likely to be exacerbated in the Internet-of-Things (IoT) era where billions of devices in our homes, offices and cars are digitally connected. On the other hand, the Integrated Circuit's global and distributed supply chain has introduced hardware security issues, such as hardware Trojans, counterfeiting, etc. In such context, each IoT device should be equipped with a unique signature that can be used for authentication by the cloud and the data transmission between devices and the cloud should be securely encrypted.
Physical Unclonable Function (PUF) is proposed to serve as the unique device signature for authentication and to generate the cryptographic keys for data encryption. Today's implementations of PUFs with silicon technology are either vulnerable to modeling attacks or side-channel attacks, and they are not robust under environmental variations. In this project, the PIs propose resistive random access memory (RRAM) technology to design the hardware security primitives, including weak PUF, strong PUF and true random number generator, to overcome the limitations of today's silicon design. The PIs exploit RRAM's intrinsic physical randomness as an entropy source. The design targets are small area, low power, high reliability and high tamper resistance. The design and fabrication of RRAM test chips are part of the proposed research.
