The title of this paper is somewhat misleading. I expected a survey of numerous security techniques for embedded systems, but in reality the authors only explore physical unclonable functions (PUFs), essentially hardware circuits that produce globally unique values. Expectations aren’t set properly. A better title would have been something like “Digital PUFs for IoT Devices.” Also, the authors keep referencing computer-aided design (CAD) and low-powered or unpowered devices as if the discussion were going to head in a new direction, but these allusions are just red herrings.
As for the core of the paper, the authors believe that hardware-based security beats software-based security, and that hardware security would be a perfect solution for Internet of Things (IoT) devices. Their goal is to initiate a “quest for conceptually new hardware security primitives,” in order to fully implement a more robust security platform. They seed this quest with a discussion of the aforementioned PUFs, starting with analog versions and moving to recent digital implementations. Their key point: let’s design a new generation of digital PUFs that can be used “to create all expected [security] protocols at a fraction of hardware and energy requirements.” IoT devices could be made immune, at least at their core security levels, to all of the software attacks that proliferate today. That’s an idea worth pursuing.
This paper is only seven pages long, and there are some mechanical problems, but it introduces a provocative new concept for improving embedded security that deserves attention. I hope the abstract can overcome the title’s shortcomings and generate widespread interest and discussion.