A Spread Spectrum Watermarking Algorithm Based on Local Instruction Statistic

This paper proposes a spread spectrum watermarking algorithm based on local instruction statistic. We select a group of vector instructions in a certain way, the function where a vector instruction occurs most frequently is the vector function, and the watermark will be embedded and detected within the vector functions. Thus each vector function can be independent, one vector function will not affect another, if the code modification made by the attackers is not in the local area, i.e. vector function, it will not affect the detection of watermark. Theoretical analysis and experiment results show that the algorithm has a low influence on the program and can resist a series of attacks.