Tuesday, April 24, 2012

New TRNG

http://arxiv.org/abs/1204.2516

Physical Unclonable Functions (PUFs) are widely used to generate random Numbers. In this paper we propose a new architecture in which an Arbiter Based PUF has been employed as a nonlinear function in Nonlinear Feedback Shift Register (NFSR) to generate true random numbers. The rate of producing the output bit streams is 10 million bits per second. The proposed RNG is able to pass all NIST tests and the entropy of the output stream is 7.999837 bits per byte. The proposed circuit has very low resource usage of 193 Slices that makes it suitable for lightweight applications.
Right now I believe one of the cheaper ways to get TRNGs (True Random Number Generators) is through Michaelson Morley interferometers (an optical TRNG that is essentially equivalent to that experiment with adjustments for high speed generation of numbers.) for >Mbps generation. (And they have a hefty pricetag of $200-$4000.) If this advance really implements a TRNG as claimed it should be significant, especially if it lowers the price.

More on PUF:
http://arxiv.org/abs/1204.0987

The characteristic novelty of what is generally meant by a "physical unclonable function" (PUF) is precisely defined, in order to supply a firm basis for security evaluations and the proposal of new security mechanisms. A PUF is defined as a hardware device which implements a physical function with an output value that changes with its argument. A PUF can be clonable, but a secure PUF must be unclonable. This proposed meaning of a PUF is cleanly delineated from the closely related concepts of "conventional unclonable function", "physically obfuscated key", "random-number generator", "controlled PUF" and "strong PUF". The structure of a systematic security evaluation of a PUF enabled by the proposed formal definition is outlined. Practically all current and novel physical (but not conventional) unclonable physical functions are PUFs by our definition. Thereby the proposed definition captures the existing intuition about what is a PUF and remains flexible enough to encompass further research. In a second part we quantitatively characterize two classes of PUF security mechanisms, the standard one, based on a minimum secret read-out time, and a novel one, based on challenge-dependent erasure of stored information. The new mechanism is shown to allow in principle the construction of a "quantum-PUF", that is absolutely secure while not requiring the storage of an exponentially large secret. The construction of a PUF that is mathematically and physically unclonable in principle does not contradict the laws of physics.