Covert Communication Model Based On Network Streaming Media And Research On Its Performance Optimization

As an important branch of information security, steganography is used to achievecovert communication. In contrast to static cover data, streaming media is structured,time-varying, and in real-time built. These make streaming media a good choice forinteractive covert communications. While steganography based on static cover data isseeking the way to collaborate with carriers, the covert communication based onstreaming media have to solve the problem of not only how to embed, but also how tomake a balance among capacity, imperceptibility and reliability.The first problem is capacity. It is possible to synchronically employ severalsteganography algorithms on one piece of structured streaming media in order toincrease capacity. However, an approach to determine whether these algorithms wouldclash each other. After previous static model study of steganography on the cover data,according to the analysis on structures of streaming media, this dissertation definedcover data dimensions and steganographic feature space. Relationship of algorithms inthis space was described as “parallel”, and the capacities of algorithms, vectors,and sequences in this space was determined. It was proved that parallel algorithms cancombine their capacities.The second problem is imperceptibility. After the three levels of the streamingmedia cover data–the communication, the packet, and the feature--there are also threelevels of imperceptibility correspondingly. Existing works on the relationship betweenimperceptibilityand capacity focusedone single hiding algorithm on static cover data. Inthis dissertation, universal properties ofthe imperceptibilitywere concluded, whichoffered some limitations of reasonable quantization of imperceptibilityshould have tocapacities. Then, a model of imperceptibilityquantization based on the results ofsteganalysis was proposed, which uniformed three levels of imperceptibility. At last, byadjusting the secret message allocation on hiding vector, imperceptibilitywas optimized.The third problem is reliability, which was broken by packet lost. After theaforementioned imperceptibility optimization, the codewords set of the covertcommunication channel is randomized. In order to keep this complex channel messagereliable, transport unit positioning scheme was proposed in this dissertation, and an additional retransmission was also presented to recover lost data with vector asthesmallest unit. In order to enhance the overall channel efficiency, thedissertationanalyzed the impact to reliability of parameters.Last but not least, a covert communication prototype system based on streamingmedia steganography was implemented with theoretical results in this dissertation.Actual testswere made to determine the system’s performance, which verified theaforementioned theory.