Research On Dynamic Software Birthmark Based On Double Granularity

With the wanton spread of pirated software, software copyright protection has become an important direction of software protection technology research. Software birthmark becomes a technology that can prove the software copyright after software watermarking. Unlike the software watermarking, software birthmark technology extracts the only inherent attributes of software to prove the software copyright. It does not need to embed additional information into the software, and does not change the size of the software. The attributes are intrinsic and unique.This thesis summarizes and analyses the existing software birthmark extraction algorithm and similarity calculation method at home and abroad. In view of the existing birthmarks are focused on single granularity attribute or static multiple attribute, a double granularity dynamic software birthmark based on function level and instruction level is proposed.First of all, the thesis proposed a dynamic software birthmark based on the instruction of basic block, which is an improvement dynamic birthmark of the k-gram instruction sequence. The improved birthmark does not depend on the size of the K setting artificially, but on the existing rules of basic block. This kind of block is a unit, and though they contain different number of instructions, they represent the semantics of a program. The instruction level birthmark is fine-grained birthmark.Secondly, the thesis puts forward a dynamic birthmark based on system function call frequency, which is coarse-grained birthmark. The two levels of granularity of birthmark constitute the double granularity dynamic birthmark, and the similarity is the mean of two kind of birthmark similarity.Finally, the proposed double granularity birthmark in this thesis is simulated by the Pin instrumentation tool under VS2010. Due to the dynamic birthmarks and input has the very big relations, in order to ensure the stability of the experimental results, the experiment were conducted in multiple input. The experimental results show that the birthmark has very high reliability and credibility. And comparing with the existing algorithm, it is also improved.