State of the art security practices use algorithms that are public and 'store' secret bits (key) securely somewhere in the system. Memories, both volatile and non-volative, have been shown to leak information to even a moderately sophisticated attacker. Physical Unclonable Functions or PUFs 'generate' (instead of storing) die-specific secret at every evaluation and promises to raise the bar against many attacks.
PUFs have been an active field of research in the last few years. Researchers have proposed their use in a number of secure IC applications including identification/authentication and encryption key generation. At the core of a silicon PUF implementation is a circuit that generates a die-specific unique randomized set of bits, which should reliably generate the same set of bits across environmental variations and aging. To date, a number of PUF implementations have been proposed, most notably the arbiter, ring oscillator (RO), and SRAM. We have implemented a new PUF based on sense amplifiers (SA) as well a number of conventional designs for comparison on a 65nm bulk CMOS testchip.
In this talk I will briefly talk about the state of the art security practices and how PUFs raise the bar of security in various applications. I will then compare the proposed PUF designs in literature and compare them with our SA based PUF design.
Mudit Bhargava is a PhD student in the Dept of ECE of Carnegie Mellon University. He is advised by Prof. Ken Mai and is currently working in the field of hardware security. After completing his B.Tech. from Dept of ECE, IIT Kharagpur, he worked for 4 years in the Embedded SRAM Design Group at ST Microelectronics, India.
Back to the seminar page