Study On Covert Communication Based On Streaming Media

With the vast development and quick popularity of streaming media application, information hiding, an innovative and challenging research topic, has attracted increasing attention. Covert communication based on streaming media (CCSM) is a typical subject of this topic. It focuses on how to hide covert information into some ostensible streaming media to achieve the purpose of secure transmission, and provides a novel and effective solution for information security problem, especially secure communication problem.Due to the difference between various branches and the diversity of hiding carriers, there has been no universal information hiding theory so far. On the basis of existing theoretical results and characteristics of streaming media, the basic principles of CCSM are investigated using information theory and optimal theory. An application model for CCSM is presented, and many important conclusions are drawn through discussions of security, steganographic capacity and correlations between cover information and covert information, which provide a theoretical basis for design of CCSM methods and prototype systems.In order to resist statistical attack while maintaining real time performance of streaming media, two covert communication methods based on streaming encryption are proposed for covert transmission of short instant messages and covert files respectively. They are described with the proposed compounded pseudorandom sequence based on Linear Congruential and Mersenne Twister and m sequence as key streams respectively. However, they can also interchange the adopted pseudorandom sequences with good results. Further, according to their application requirements, synchronization mechanisms are designed by incorporating protocol steganography techniques, which can effectively resolve the synchronization problem of keys and covert information.From the perspective of similarity theory, it is firstly postulated that steganographic performance can be significantly enhanced by increasing the similarity between cover information and covert information. The measurable similarity, partial similarity and relevant measure criteria are completely defined. In view of the inter-restricted and interactional relationship between steganographic transparency and hiding capacity, three covert communication schemes based on partial similarity are presented. The first one simply employs one partial similarity threshold to guide the embedding process; the second one introduces triple m sequences and describes an adaptive partial-matching approach using two partial similarity thresholds; the third one is an improved version of the second one, which takes account of the similarity not only between the cover information and the encrypted covert information but also between the cover information and the original covert information. Although the proposed schemes differ in their performances, all of them can adaptively achieve the acceptable balance between perceptual transparency and hiding capacity while maintaining the real time performance of streaming media. We can choose an appropriate scheme in terms of specific characteristics of the cover and desired steganographic performance.Due to trading hiding capacity for perceptual transparency, the existing steganographic encoding algorithms are not totally suited for CCSM. For that reason, a steganographic encoding algorithm based on digital logic transformations is proposed. According to the adopted transformations, it can be divided into three categories:logic operations based algorithm, shift operations based algorithm and hybrid algorithm that combines the operations of the first two. The main idea behind them is to increase the similarity between cover information and embedded information via logic transformations to enhance the steganographic transparency. Essentially, they can reduce the total number of changed bits by decreasing the effective bit-change rate. Therefore, they can enhance the steganographic transparency while maintaining the maximum hiding capacity of streaming media. In addition, in order to dynamically adjust steganographic transparency and hiding capacity, a dynamic matrix encoding algorithm is presented. Compared with the traditional matrix encoding algorithm, it can optionally adjust embedding rate and bit-change rate by dynamically regulating the size of each covert message vector. Also, it inherits the advantage of high embedding efficiency. However, it should be pointed out that the dynamic matrix encoding algorithm is a probabilistic approach, whose error is inversely proportional to hiding capacity. Thus, this algorithm is not suited for applications with very small hiding capacity, but it is very reliable in long-time covert communications.Finally, design principles of CCSM system are investigated. Incorporating the proposed covert communication methods, a prototype VoIP-based covert communication system (named StegVoIP) is designed and implemented. The test results show that, StegVoIP can effectively transmit short instant messages and small covert files while maintaining the normal VoIP conversation.