On K-gram Based Dynamic Software Birthmarking

As been known to all, it is an undoubted fact that respecting intellectual property rights is the key to develop knowledge economy and enhance the ability of independent innovation. Because the software has the characteristics of high added value, easy to be copied and low-cost, software pirates are increasingly rampant, which leads to the development of software becoming much difficult.The technology of software protection, has outstanding characteristics such as multi-angle research and variety of experimental methods, but main existing methods only ensure software's security during its distributing and running, which has the goal of satisfying the distributing requirement. Therefore, there is no obvious breakthrough against piracy by now. A published survey from the BSA and the IDC in 2007 shows that 35% software installed in personal computers was illegal all over the world in 2006. The direct economic losses were almost 39 billion dollars, which added by 5 billion dollars contrasted with 2005. In China, the pirating rate was 86%, and the direct economy losses were approximate to 3.9 billion dollars. The problem of software pirate needs to be seriously considered since it had blocked the development of software.As a new branch of software protections, software birthmark can prove code theft by identifying intrinsic properties of a program. It has received highly attention from both software manufacturer and academic circles, because this method could prove the copyright by itself and do not interrupt the program's running. In recent years, some algorithms of software birthmark have been proposed, but there are still many questions that should be studied further.This paper presents a novel dynamic software birthmarking scheme based on the k-gram text analysis method. Firstly, we slice the program's excution history with the given input, and get the instruction sequence that denotes the observable behavior of a program. Then a union of k-gram instruction-sequence sets can be gained by k-gram method. Finally, the hash values of all k-gram sub-instruction-sequences denoted as a birthmark are used to identify a program uniquely. In order to evaluate the robustness of this scheme, we design an evaluating framework that contains some experiments of creditability and resilience, which can be good for measuring the performance of software birthmark. The results show that the dynamic k-gram birthmark not only keeps the high creditability such as static k-gram, but also enhances obviously the performance of resilience. There is a good balance between creditability and resilience. This scheme is of benefit to the research of software birthmark based instruction level, and it also can be provided with perfortable researchful value.