Domain Wall Memory: The Next Big Thing in Hardware Security?

Jul 14, 2015  |  Computing, Devices & Systems

University of South Florida researchers recently set out to find a way to give consumers more bandwidth. What they stumbled upon, however, may very well become a valuable hardware network security tool.

An article in IEEE Journal on Emerging and Selected Topics in Circuits and Systems details how the team originally investigated new ways to design cache using domain wall memory (DWM), which is ideal for the application due to its low standby power, fast access time and good endurance. The researchers tested a physics-based model of DWM to determine how it behaves under temperature, radiation and velocity. That’s when they inadvertently discovered that DWM’s characteristics make it a potential asset for hardware security purposes.

Memory consists of densely populated wires which read and write data back and forth. The faster data moves along these nanowires, the faster a processor will work. But similar to fingerprints, no two nanowires in memory are exactly the same due to their random bumps and ridges, and the effect of varied heat levels in the environment. The team found these variations affect DWM’s data access speeds so much that no two wires deliver information at the same time, making them very hard to replicate. Someone would have a lot of difficulty duplicating a chip if every single wire is different and delivers memory at different speeds.

The way DWM is designed makes it extremely hard to copy. Hardware security could greatly benefit from this.

“Our original research sought to design new cache using DWM,” said Anirudh Srikant Iyengar, lead researcher of the group. “But once we determined how difficult hacking a system like this would be, we changed directions and started looking at hardware security. The way DWM is designed makes it extremely hard to copy. Hardware security could greatly benefit from this.”

The researchers propose physically unclonable functions (PUFs) – which are commonly used to encrypt confidential information – that use the characteristics of DWM to generate unique cryptographic keys. This new concept leverages the randomness of DWM’s wire velocities to create a key that is virtually unable to be replicated. Previous keys were able to be decoded through reverse engineering. This addition to PUFs will make it very difficult for adversaries to crack the code.

Researchers are hopeful that it could start being implemented in the next five years.

This concept is still in the development phase, but the researchers are hopeful that it could start being implemented in the next five years. Unique keys like these are commonly used in IP security and counter-plagiarism chips in memory processors. Adding DW dynamics could make hardware security a lot harder to hack, not to mention save companies money and credibility in the long run.

You can also find more articles about “Domain Wall Memory” in IEEE Xplore.

Read the first page of this article for free on IEEE Xplore.

View the full-text article on IEEE Xplore. Read the first page for free. Full article available with purchase or subscription.