Highly Secure Silicon Physical Unclonable Function

Physical Sciences : Electrical

Available for non-exclusive licensing

Inventor

  • Michael Orshansky, Ph.D. , Electrical and Computer Engineering

Background/unmet need

In many electronic systems, the physical uniqueness of microelectronic hardware components can be exploited to increase the level of security in a device. Physical unclonable functions (PUFs) have been used for over a decade to exploit this uniqueness for that purpose. Broadly speaking, PUFs are circuits that use manufacturing variability to generate a device-specific output which can be seen as the fingerprint of a device.

The two most widely used applications of PUFs are for cryptographic key generations which are produced directly from the unique signature of the hardware. Another application is for authentication: the uniqueness of the components allows generation of challenge-response pairs that serve as unique identifiers of the device.

During the last 10 years, PUFs have become a mainstay in hardware security research and development. However, existing PUFs have some important weaknesses. For example, even the most widely adopted silicon PUFs have shown a vulnerability to attacks that use machine-learning methods to construct a software model of the PUF. This model can then be used to overcome the security guarantees of PUFs.

By exploiting the physics of field-effect transistors (FETs) at the nanometer scale in a fundamentally new way, researchers from the Department of Electrical Engineering (ECE) at UT-Austin have taken a novel approach towards significantly improving PUF resilience against machine-learning attacks.

Invention Description

The resulting invention is a highly unpredictable, strongly non-linear silicon PUF that shows excellent security properties which are superior to those of currently available PUF technologies in terms of secrecy and the number of unique secrets that it can generate.

Benefits/Advantages

  • Greater resistance to modeling attacks
  • Produces a large number of challenge-response pairs
  • Low Power
  • Does not compromise the stability in the output response to environmental variations

Features

  • Based on the essential nonlinearity of terminal current-voltage behavior of FETs at the nanometer scale
  • Does not rely on digital techniques for introducing the nonlinearity
  • Has excellent uniqueness properties measured by inter-class and intra-class Hamming distance

Market potential/applications

Semiconductors and other upstream circuitry suppliers (e.g., for physical systems that require authentication, key generation, software protection, ID management, etc.)

Development Stage

Lab/bench prototype

IP Status