In order to address security issues effectively at nano-scale and develop new devices that are better suited for security and resiliency against attacks we investigate to identify a set of universal security properties of devices S={S1, S2, …} that correspond to the hardness or infeasibility of attacking physical security primitives: Each security primitive consists of some logical interface that connects to an ensemble of interconnected devices. We define a primitive’s composition heuristic as the way the primitives are designed with respect to the features of its logical interface and composition of devices. The composition heuristics define different implementations of a given security primitive. In our proposed research we will develop a set of possible composition heuristics C={C1, C2, …} and evaluate their resulting security. In order to evaluate the security of a composition of devices we need to know which subset in S of security properties of the devices plays a role in this analysis.

Task B.1. Property Identification and Metrics
The diversity of the security primitives and attacks makes the development of a universal model for evaluating existing and the next generation of nanodevices quite challenging. The main objective of this task is to clearly define what the most relevant properties are for security primitives and attacks. Ultimately, the goal is to identify which security properties (set S) are needed from each nano-device in order to attain a certain primitive. In other words, we will determine which nano-device is best suited as a TRNG or PUF, which one is more resistant to side-channel attacks, which one is more expensive to be cloned, how to design new nano-devices to maximize certain properties to build a stronger security primitive, etc.

Task B.2. Composition Heuristics and their Evaluation
Security cannot be based solely on single devices. If we want to design a secure primitive, we need to build more complex structures that compose several devices and which can only be accessed through a logical interface.
In this task, we will design and analyze composition heuristics that have a balanced set of properties. I.e., given a composition, what is the underlying implicit hardness assumption (also given the aggregated side channel information), what interface components (post-one-way function, pre-one-way function, privacy amplification, error correction and fuzzy extraction, access control, proof of execution, etc.) are needed and what is the resulting area size, and how does noise propagate (i.e., do our composition theorems express “nice” behavior)?
As an example of one promising concrete line of research is to provide novel composition heuristics for achieving error-resiliency and attack robustness in TRNGs.