Network Watermarking Model Based on Optimization Theory

In order to compare the efficiency of different network watermarking systems and help designing a more efficient system by finding capacity bounds and judging robustness and invisibility level, a network watermarking model based on optimization theory is proposed. Firstly, it covers attack effects on network watermarking by inducing network transformation problems. Secondly, the proposed model establishes unified analysis criteria for robustness and invisibility, which helps to divide robustness and invisibility into three levels. Besides, combined with different attack intensities, it can effectively measure robustness and invisibility by goodness-to-fit test. Thirdly, the proposed model converts robustness, invisibility and network transformation problems into different constraints. Therefore, the maximum capacity under different conditions can be figured out by layered superposing corresponding constraints. At last, experimental results verifies the correctness and feasibility of the proposed model.