CybersecurityMaking scalable on-chip security pervasive

Published 8 April 2019

For the past decade, cybersecurity threats have moved from high in the software stack to progressively lower levels of the computational hierarchy, working their way towards the underlying hardware. The rise of the Internet of Things (IoT) has driven the creation of a rapidly growing number of accessible devices and a multitude of complex chip designs needed to enable them. A new DARP program focuses on addressing the economic and technical challenges associated with incorporating scalable defense mechanisms into chip designs.

For the past decade, cybersecurity threats have moved from high in the software stack to progressively lower levels of the computational hierarchy, working their way towards the underlying hardware. The rise of the Internet of Things (IoT) has driven the creation of a rapidly growing number of accessible devices and a multitude of complex chip designs needed to enable them. With this rapid growth comes increased opportunity for economic and nation-state adversaries alike to shift their attention to chips that enable complex capabilities across commercial and defense applications. The consequences of a hardware cyberattack are significant as a compromise could potentially impact not millions, but billions of devices.

DARPA notes that despite growing recognition of the issue, there are no common tools, methods, or solutions for chip-level security currently in wide use. This is largely driven by the economic hurdles and technical trade-offs often associated with secure chip design. Incorporating security into chips is a manual, expensive, and cumbersome task that requires significant time and a level of expertise that is not readily available in most chip and system companies. The inclusion of security also often requires certain trade-offs with the typical design objectives, such as size, performance, and power dissipation. Further, modern chip design methods are unforgiving – once a chip is designed, adding security after the fact or making changes to address newly discovered threats is nearly impossible.

“Today, it can take six to nine months to design a modern chip, and twice as long if you want to make that same design secure,” said Serge Leef, a program manager in DARPA’s Microsystems Technology Office (MTO). “While large merchant semiconductor companies are investing in in-house personnel to manually incorporate security into their high-volume silicon, mid-size chip companies, system houses, and start-ups with small design teams who create lower volume chips lack the resources and economic drivers to support the necessary investment in scalable security mechanisms, leaving a majority of today’s chips largely unprotected.”