A team of researchers from the Research Institute of
Computer Science and Random Systems (IRISA) in France made up of Annelie
Heuser, Matthieu Mastio, Duy-Phuc Pham, and Damien Marion have released a new
paper detailing how they were able to accomplish this feat.
Unlike other malware detection systems that rely on
software, IRISA's new system uses an oscilloscope (Picoscope 6407) and an
H-Field probe in addition to a Raspberry Pi 2B to scan devices for specific
electromagnetic (EM) waves.
Using EM waves to detect malware
IRISA's new malware detection system relies on an external
oscilloscope and an H-Field probe to scan devices by looking for specific waves
that indicate the presence of malware on an affected device. However, the
research team was also able to “obtain precise knowledge about malware type and
identity” from these scans.
At the same time, since no additional software needs to be
installed, many obfuscation techniques used by hackers and cybercriminals are
completely ineffective and can even be detected and analyzed by the system.
While the Raspberry Pi used in this project was trained
using both safe and malicious data sets, the researchers also used Convolution
Neural Networks (CNN) to evaluate the data for threats. In the end, the model
used to train IRISA's malware detection system was up to 99.82 percent
effective during testing.
While their detection system is far from being commercially
deployed, it could potentially make things much more difficult for malware
writers as they will also now need to figure out a way to hide their malicious
code from EM scanners.