The National Science Foundation (NSF) funded two teams — one from the State University of New York (SUNY) at Binghamton
and one from the University of California, Riverside — to determine whether it is feasible to build a “Practical Hardware-Assisted Always-On Malware Detection” malware chip. The $275,000 grant from the NSF extends for three years, according to Bleeping Computer.
A Malware Chip Off the Old Block
The study is largely based on a 2014 paper from Columbia University researchers titled “Unsupervised Anomaly-Based Malware Detection Using Hardware Features.” The Columbia team employed “unsupervised machine learning” to create profiles based on data from embedded performance counters.
Researchers then used the profiles to detect “deviations in program behavior that occur as a result of malware exploitation.” The hardware of the central processing unit (CPU) provided the processing power for all the performance detectors. Intel and Clarkson University have also investigated this approach in the past.
The NSF project is specifically based on the two papers presented by the SUNY Binghamton researchers titled “Hardware-Based Malware Detection Using Low-level Architectural Features” and “Ensemble Learning for Low-level Hardware-Supported Malware Detection.”
This isn’t the team’s first foray into the world of security. In fact, two of the researchers involved in the project were part of the crew that discovered the Intel Haswell CPU address space layout randomization (ASLR) bypass technique. They seem to know their stuff.
An Extra Layer of Defense
This new project aims to “modify a CPU chip to include the extra logic to detect anomalies in running processes,” according to Bleeping Computer. The CPU hardware will not deal directly with any discovered anomaly, however. Instead, it notifies local security software, which ultimately determines how to deal with the problem.
“The hardware detector is fast, but is less flexible and comprehensive,” Dmitry Ponomarev, professor of computer science at SUNY Binghamton, explained through the university. “The hardware detector’s role is to find suspicious behavior and better direct the efforts of the software.”
This approach alone cannot catch all security threats. But it can add a layer of defense protection to the CPU on which it is being run.