DKISB: Dynamic Key Instruction Sequence Birthmark for Software Plagiarism Detection

With the burst of open source software, software plagiarism has been a serious threat to the healthy development of software industry. Software birthmark reflecting intrinsic properties of software, is an effective way for the detection of software theft. However, most of the existing software birthmarks face a series of challenges: (1) the absence of source code, (2) diversity of operating systems and programming languages, (3) various automated code obfuscation techniques. In this paper, a dynamic key instruction sequence based software birthmark (DKISB) is proposed. By introducing dynamic data flow analysis into birthmark generation, we are able to produce a high quality birthmark that is closely correlated to program semantics, making it resilient to various kinds of semantic-preserving code obfuscation techniques. Based on the Pin instrumentation framework, a DKISB based software plagiarism detection system is implemented, which generates birthmarks for both the plaintiff and defendant program, and then make the plagiarism decision according to the similarity of their birthmarks. The experimental results show that DKISB is effective to either weak obfuscation techniques like compiler optimization or strong obfuscation techniques provided by tools such as Sand Mark.