Machine Learning

Machine Learning


Advanced Learning Algorithms to Improve Security of Embedded Cyber-Physical Systems

Modern, networked, cyber-physical systems pose a special challenge for Cyber Vulnerability Assessment.  Communication protocols may be inherently vulnerable, may be implemented in an unsafe manner, or may be vulnerable to timing attacks, in which carefully-timed inputs can be used to disrupt system operations.

Developed under AFRL's NOVA program, Adventium Labs’ VOLTA software automates the process of exploring the behavior of implemented systems, including timed behavior.  This automation may save days or weeks of effort by specialized analysts. Because it learns implemented behavior, VOLTA can be applied to legacy systems, even where documentation is not available, or as a design aid to identify vulnerabilities in systems currently being designed or modified.

When completed, VOLTA will join Adventium's other behavior analysis tools on the CAMET® Library, similar to SLICED.