Net-flow Fingerprint Model Based on Optimization Theory
Cheng Lei,Hongqi Zhang,Yi Liu,Xuehui Du
DOI: https://doi.org/10.1007/s13369-016-2073-y
2016-01-01
Abstract:Net-flow fingerprint technique provides better efficiency in areas such as stepping stones detection and anonymous network correlation. Carrier capacity, robustness and invisibility are important indicators of net-flow fingerprint systems. In order to compare the efficiency of different net-flow fingerprint systems and help designing a more efficient system by finding capacity bounds and judging robustness and invisibility level, a net-flow fingerprint model based on optimization theory is proposed. Firstly, the proposed model covers all possible attack effects on net-flow fingerprint by reducing net-flow transformation problems to mergence, substitution and insertion, which improves the applicability of the model. Secondly, the proposed model establishes unified analysis criteria for robustness and invisibility, which, combined with different attack intensities, helps to divide robustness and invisibility into three levels, and then effectively and accurately measures them by goodness-of-fit test. What is more, the proposed model converts robustness, invisibility and net-flow transformation problems into different constraints. The maximum capacity under different conditions can be figured out by layered superposing corresponding constraints. Experimental results have confirmed the correctness, feasibility and expandability of the proposed model.