Metamorphic Malware Detection Using Function Call Graph Analysis

Abstract

Previous work has shown that well-designed metamorphicmalware can evade many commonly-used malware detection techniques, including signature scanning. In this paper, we consider a previously developed score which is based on function call graph analysis. We test this score on challenging classes of metamorphic malware and we show that the resulting detection rates yield an improvement over other comparable techniques. These results indicate that the function call graph score is among the stronger malware scores developed to date.