摘要:
This application proposes the use of Cellular Non-Linear Networks (CNNs) as physical unclonable functions (PUFs). We argue that analogue circuits offer higher security than existing digital PUFs and that the CNN paradigm allows us to build large, unclonable, and scalable analogue PUFs, which still show a stable and repeatable input-output behaviour. CNNs are dynamical arrays of locally-interconnected cells, with a cell dynamics that depends upon the interconnection strengths to their neighbours. They can be designed to evolve in time according to partial differential equations. If this equation describes a physical phenomenon, then the CNN can simulate a complex physical system on-chip. This can be exploited to create electrical PUFs with high relevant structural information content. To illustrate our paradigm at work, we design a circuit that directly emulates nonlinear wave propagation phenomena in a random media. It effectively translates the complexity of optical PUFs into electrical circuits. This, leading to better practicality, while maintaining or even improving the security properties of their optical counterparts.
摘要:
A system for security purposes comprising: - an inner structure that is accessible by a plurality of terminals, - wherein the system allows a measurement on the inner structure of the system by using a challenge signal comprising a plurality of input signals applied in parallel to the terminals and by receiving a response signal dependent on a setting of the challenge signal and dependent on the inner structure of the system, wherein - the system comprises at least two features of the group containing: > the ability to process non-binary input signals, > a bandwidth at the terminals and an information content ensuring a incomplete readout of the information content within a predefined access time period, > a spatial and/or logical disorder of the inner structure, > a non-linear relation between the input signals and output signals appearing at the terminals, and that - from the knowledge of a subset of the predefined settings and of associated response signals, the response signal associated with a particular setting outside the subset is not predictable with a probability higher than the probability for guessing the response signal.