Study On Steganography Of VoIP Speech

With the rapid development of network and communication technology, traditional speech communication is being gradually replaced by more cost-efficient and effective net-work telephony (VoIP). How to exploit covert channels in VoIP and eliminate latent security risk within become the hotspot of covert communication technology. And the transmission media, VoIP speech, as a important source of covert communication channel, has lots of covert channels available therein, thus is very important for the research of covert commu-nication technique in VoIP and whole speech communication field. Differing from infor-mation hiding of traditional static covers (e.g. image, audio and video), VoIP system is often under an opening network environment, lots of latent risk exists and the covert com-munication is vulnerable to potential attack. On the other side, speech communication has the feature of low latency, low redundancy and instability, that makes the enforcement of steganography technique harder.Transmitted speech in VoIP channel is compressed and coded from traditional sampled pulse coded modulation speech signal, thus there is little redundancy available, which is conflict with the redundancy requirement of steganography. Otherwise, once the speech is encoded into frames, there will be large discrepancies among frame parameters. If tradition-al least significant bits (LSB) method is adopted, more additional noise will be brought in. However, if we could utilize the specialities of VoIP and combine speech steganography with kinds of stcgo coding techniques to reduce thc modification towards speech cover during the steganography procedure, steganography performance will be largely improved. Moreover, studying on the steganograpy features of all parameter bits which affect the speech frame to employ more effective embedding techniques is significant for speech covers. This paper focuses on LSB replacement based methods in VoIP environment to increase covert com-munication capacity, to reduce distortion introduced by steganography and to enhance the security of steganography.(1) Combining stego coding with embedding method can increase the embedding ef-ficiency of covers and reduce introduced distortion. In hybrid-coding based speech codec, different speech parameter bits have different weights, so there will be heavy distortion-s and low embedding efficiency when embedding with stego coding directly. As a result, CLFW method is proposed to combine the embedding with stego coding effectively so as to enhance the embedding efficiency and reduce introduced distortion. In CLFW method, LSB arc selected according to different impacts on speech quality using PESQ method, then divide them into different parts. CLFW adopts a simple window mechanism for selecting to attain the required bits for stego coding. By doing this, the inputting bits for stego coding are increased and the noise resistance of them are more similar, thus stego coding is more flexible as well as increased embedding efficiency and low distortion are obtained.(2) The steganography of speech frame can be more fine-grained to further improve the steganography performance. State-based "least significant digit" steganography method is the extension of LSB steganography. The notion of "bit" in LSB is extended to "digit" in multi-ary system, the selection of LSB is replaced by least significant state accordingly. This method is more fine-grained compared with LSB method. In the embedding and extracting procedure, different multi-ary system are chosen for different numbers of states selected. The combination of multi-ary arithmetic and state-based strategy increases steganography capacity a lot with the same distortion. If the existing bits for embedding is limited, this method has less effect on speech quality compared with other methods. Otherwise, for the enhancement of embedding precision the choosing of embedding is more flexible, that makes the steganography more secure.(3) Besides steganography using redundancy inside speech frames, the redundancy pro-duced by frame bitrate alteration during the transition in VoIP communication can also be exploited for hiding. This method is better than steganography method based on frame modification theoretically. One strategy is to decrease the frame coding bilrate and replace higher bitrate frame with lower one, so the remained space can be used for steganography. This method is applicable for transmission in encrypted RTP packets. The other strate- gy is to embed secret message in the switching of frame bitrate, the advantage is that the stego frame can be decoded by normal codecs correctly and the disadvantage is that it has a low embedding capacity. Both of them are implemented in a covert communication system called StegVoIP, and the designed target is attained.The above methods are all concentrated on spacial domain towards speech stcganog-raphy. Compared with steganography methods in transform domain, this kind of method is simple and feasible. What’s more, they are universal methods for different kinds of speech codecs. Otherwise, by virtue of the complexity of VoIP system and the double key mechanism inside steganography, the security can also be assured.