[Under construction]2014 AFOSR MURIDevelopment of Universal Security Theory for Evaluation and Design of Nanoscale DevicesPO: Dr. Tristan Nguyen, PI: Dr. Mark Tehranipoor, University of ConnecticutWebsite: TBD
Context: Over the past several decades, the development of nano-scale devices has been driven by scalability and performance with little attention to security. As a result, the semiconductor community is largely unprepared to identify and address the security challenges and vulnerabilities of existing and next-generation nano-scale devices. Given the pervasiveness, even universality, of electronics in products, systems, and infrastructure affecting national security, commerce, energy, and transportation, such an oversight has wide-ranging implications. By elevating security as a fundamental design parameter at nano-scale, we have the opportunity to reverse this trend and provide a more secure foundation for future systems.
Objective: The objective of this proposal is to develop a universal security theory for the evaluation and design of nano-scale devices. Our universal security framework intends to (a) explore the unique features of nano-scale devices and how they may contribute to security, (b) establish abstract models that encapsulate the physical characteristics and functionality of nanoscale devices, (c) develop concrete heuristics that merge the abstract models of nano-scale devices, (d) theoretically evaluate primitives driven by nano-scale devices in terms of universal properties/metrics that we develop, and (e) exploit what is learned to (re)design nano-scale devices with security primitives/attacks in mind and vice versa. Such an improvement in the state-of-the-art would have a significant impact on DOD core missions including (1) substantially enhancing the security of the electronic systems in military and space applications, and (2) greatly improving the security of DOD’s electronic component supply chain.
Approach: A key element to our approach is a powerful, universally applicable concept called a Black Box Statistical Device Model (BBSDM). A BBSDM expresses a device’s input and output behavior based on a set of hidden parameters (caused by manufacturing variability). A BBSDM can be used to simulate security primitives (e.g., PUFs, TRNGs, COAs, etc.) that are composed of multiple devices together with a logical interface. In our framework we use this “simulated view” to evaluate universal security properties like observability, controllability, tamper-evidence, and unclonability in terms of associated metrics. We shall use our framework as a guide for developing nano-scale devices with security primitives/attacks in mind and vice versa. Specifically, we will explore the unique features of nano-scale devices and demonstrate how they can improve security, create new capabilities, and accelerate performance of secure implementations. As one concrete realization, we will focus on new implementations of primitives that integrate phase change memory (PCM) cells with a logical CMOS interface in 3D. This represents a fresh direction for PCM technology that goes beyond its current purpose as non-volatile resistive memory.
Team The research team consists of recognized leaders in the domains of security and nanoscale devices and offers a strong record of scholarship and innovation. UCONN’s CHASE Center houses $3M in some of the most advanced equipment available for nano-device analysis, characterization, and testing. The combined infrastructure and personnel resources of the team constitute an exceptional, if not unique, capability in the field of security for nano-scale devices.