
When traditional semiconductor technologies are unable to meet specific nonfunctional requirements of emerging products—including for factors such as energy consumption, peak power dissipation, and chip cost and/or size—new technologies must emerge to fill the gap.
In a recent IEEE Security & Privacy column, “Hardware Security in the Era of Emerging Device and System Technologies,” Associate Editor-in-Chief Nele Mentens advocates for increased research into the security of electronic systems based on new, more-efficient hardware technologies.
A Roadmap of Technology Opportunities
The International Roadmap for Devices and Systems (IRDS) uses a 15-year horizon to provide predictions and guidance on the expected opportunities and limitations of traditional and emerging technologies.
Among the various International Focus Teams (IFTs) that the IRDS Executive Summary describes are the following:
- The “More Moore” IFT: elaborates on challenges and solutions for continuing Moore’s law and focuses on the down-scaling of semiconductor features.
- The “More than Moore” and “Beyond CMOS” IFTs: tackle emerging technologies that follow alternative pathways rather than focusing on Moore.
In the latter case, examples of alternative technologies include
- novel memory technologies, such as magnetic memories and resistive memories; and
- novel logic devices, such as those based on carbon nanotubes or spin waves
Commercial products are already using some of these emerging technologies, yet research on the hardware security properties of these products … is still in its infancy.
Security: The Limits of Existing Knowledge
In her column, Mentens asks a key question:
- Can existing knowledge on traditional technologies be used to reason about emerging technologies?
She then posits that answering this question requires revisiting hardware security’s fundamental building blocks: hardware roots of trust (RoTs).
Synopsys—a global leader in hardware IP cores for security—cites several key RoTs, including the following three:
- Hardware cryptographic accelerators
- True random number generators (TRNGs)
- Secure storage
To this list, Mentens adds
- physically unclonable functions (PUFs) as an alternative for secure key storage.
The thing these components share, says Mentens, is that their desired hardware security properties “are highly dependent on their physical behavior, which is determined by the underlying technology.”
Generating New RoTs: Two Promising Technologies
Mentens highlights two emerging technologies as suitable for generating hardware RoTs for commercial products: resistive memories and flexible electronics.
Resistive Memories
A type of emerging memory device, resistive memories are nonvolatile; they both consume less power and work faster than traditional Flash memories.
Mentens also cites three additional qualities of resistive memories:
- They are more expensive to fabricate than Flash memories.
- They have great potential to improve the computing system, particularly in high-performance neuromorphic computing.
- Because they differ completely from the semiconductor-based mechanisms that traditional volatile or nonvolatile memories rely on, security properties for these traditional memories do not necessarily hold for resistive memories.
For its commercially available resistive memory, Crossbar mentions various secure storage properties, including
- protection against read-out,
- unclonability, and
- the ability to build PUFs
Although researchers also offer novel ways to design PUFs based on resistive memories, Mentens recommends a thorough analysis of security properties based on open source models of resistive memories before they become widespread.
Flexible Electronics
Mertens expects flexible electronics to become an increasingly important topic in hardware security research. These thin, ultralight chips are built on mechanically flexible substrates such as plastics, metal foil, flexible glass, and paper. Their advantages include
- Low cost
- Short development cycle
Although not suitable for high-performance systems, Mertens says flexible electronics are perfect for Internet of Things devices, especially in
- Wearables
- Smart packaging
- Logistics
- Product authentication
In terms of built-in RoTs, research suggests that, for TRNGs and PUFs, flexible electronic chips based on inkjet printing could use random ink dispersion as a source of randomness or intrinsic variation.
Digging Deeper
In addition to discussing various other types of RoTs for flexible electronics, Mertens issues a call for researchers in security, hardware design, and emerging technologies to collaborate on work that will enable secure deployment of these technologies at scale for a range of applications.
To read more, check out “Hardware Security in the Era of Emerging Device and System Technologies.”