Adaptive partial-matching steganography for voice over IP using triple M sequences

Although steganographic transparency and steganographic bandwidth are believed to be two conflicting objectives in the design of steganographic systems, it is possible and necessary to strike an optimal balance between them. This paper presents an adaptive partial-matching steganography for voice over IP (VoIP). We introduce the notion of partial similarity value (PSV) to evaluate the partial matching between covers and secret messages. By properly setting a low threshold of PSV and a high threshold of PSV, we can adaptively balance steganographic transparency and bandwidth. Moreover, we employ triple m sequences to eliminate the correlation among secret messages, guide the adaptive embedding process, and encrypt synchronization signaling patterns. In addition, we introduce an improved strategy that takes into account the similarity between not only covers and encrypted messages but also covers and original messages. We evaluate the proposed approach and its improved strategy with ITU-T G.729a as the codec of the cover speech in StegVoIP that is a prototypical covert communication system based on VoIP and compare them with some existing approaches. The experimental results demonstrate that the proposed approaches can provide a better balance between steganographic transparency and bandwidth. Furthermore, the results of delay tests show that they adequately meet the real-time requirement of VoIP.