A novel steganography approach for voice over IP

In homeland defense and security, secure data transfer is still critical challenging due to the open nature of Internet. One of the solutions which came to the rescue is the VoIP (Voice over IP) steganography. VoIP is unquestionably the most popular real-time service in IP networks today. To date, existing VoIP steganography research commonly focus on information hiding in the LSB bits of network audio streams. However, this approach may raise serious security threat, where the hidden information may be easily removed, detected and attacked. Towards this issue, we propose AVIS, a novel Adaptive VoIP steganography approach to hide information within network audio streams. AVIS consists of three parts, named VAMI, VADDI and VODO. VAMI works by dynamically selecting multiple bits based on the VoIP vector value, VADDI dynamically changes embedding intervals to avoid detection and attacking, and VODO try to change the neighbor bits to offset the sound distortion. Also, we evaluate the effectiveness of this approach with G.711 as the codec of the cover speech in Linphone , a famous open-source VoIP software. The experimental results demonstrate that our approach provides better performance than the traditional one.