Detecting Embedded Vulnerabilities in Software (DEVIS)

Employing software assurance techniques to detect vulnerabilities in binary executables running on embedded systems presents a variety of unique challenges. First, embedded systems are subject to customized attacks for which signature and heuristic based approaches developed for personal computing and enterprise architectures are not applicable. Second, while vulnerabilities may not be application specific, their actual manifestation on a given hardware architecture is dependent on that architecture and instruction set. To address these challenges RAM Laboratories is developing a Detecting Embedded Vulnerabilities in Software (DEVIS) framework. DEVIS detects, ranks, and locates security flaws in embedded cyber physical infrastructure, such as machinery and control systems. DEVIS is supported by multiple architecture specific toolsets, which optimally combine the use of symbolic execution, fuzzing, and dynamic binary instrumentation techniques to efficiently use available processing and memory resources. Key to this approach is the ability to alternate between these techniques to rapidly search for software states to detect and rank high priority vulnerabilities. DEVIS dynamic vulnerability analysis can be performed both on board instrumented hardware and through the use of an emulated instrumented architectures and instruction sets in virtual settings. Phase II targets ARM and non-ARM architectures, such as the Power Series, Quark, and others.