A Combined Static and Dynamic Software Birthmark Based on Component Dependence Graph

Software birthmarking provides an effective approach to detect software theft by computing the similarity of unique characteristics between the suspected program and the original. In this paper, we present and empirically evaluate a novel birthmarking technique which uniquely identifies a program based on static and dynamic component dependence graphs of it. To argue the advantage of the technique, the credibility and reliability against semantics-preserving transformations are evaluated. Experimental results show that our technique is more stable than the WPP birthmark proposed by Myles and Collberg. Additionally, it complements the previously proposed birthmarking techniques which are only static or dynamic.