ABSTRACT

A novel data-hiding methodology, denoted as digital invisible ink (DII), is proposed to implement secure steganography systems. Like the real-world invisible ink, secret messages will be correctly revealed only after the marked works undergo certain prenegotiated manipulations, such as lossy compression and processing. Different from conventional data-hiding schemes where content processing or compression operations are undesirable, distortions caused by prenegotiated manipulations in DII-based schemes are indispensable steps for revealing genuine secrets. The proposed scheme is carried out based on two important data-hiding schemes: spread-spectrum watermarking and frequency- domain quantization watermarking. In some application scenarios, the DII-based steganography system can provide plausible deniability and enhance the secrecy by taking cover with other messages. We show that DII-based schemes are indeed superior to existing plausibly deniable steganography approaches in many aspects. Moreover, potential security holes caused by deniable steganography systems are discussed.


[attachment=10230]
INTRODUCTION
HIDING secrets and only revealing them to certain users may be instincts of humankind. Looking back on history, events involving the art of secret hiding can be found in many aspects of human civilization. Before the digital era, messages
are often hidden with steganographic skills. Steganographic techniques in the real world fall into two branches, linguistic steganography and technical steganography, as introduced in . Linguistic steganography consists of two classes of methods: delivering secret messages via an open code where prior agreements about the true meaning of seemingly harmless phrases, gestures or expressions must be negotiated in advance, as well as semagrams that secrets are expressed in the form of visible but minute graphic details in a drawing or script.
A. Invisible Ink in the Real World
As for technical steganography, writing with invisible ink is the most renowned skill. Certain liquids like lemon juice or milkhave proved popular and effective since ancient times. In general, invisible ink is a substance used in steganographic schemes so that secret messages can be invisibly written on papers. The ink is invisible during writing or soon thereafter. Later on, the hidden message may be developed (made visible) by different methods according to the type of adopted invisible ink. For example, messages written with diluted acid liquids can be developed by heating the paper. Development methods for other types of invisible inks include applying chemical liquids or vapors upon the paper, viewing the paper under ultraviolet light, and so on. Fig. 1 shows a conventional espionage scenario in which invisible ink is used. When a sender wants to deliver some intelligence to a certain receiver over a supervised channel, he can write secret messages on the paper using some acid liquid. Note that, usually, the paper also carries some cover messages written with normal ink because sending a blank sheet of paper might arouse suspicion. The supervisor cannot find any anomaly in the paper under common viewing conditions. When the intended receiver gets the paper, some prenegotiated manipulations, e.g., the heating
operation in Fig. 1, should be performed first so that the secret messages can be revealed. An introduction to invisible ink used by secret operation agents during World-War II can be found in.
Some characteristics of invisible-ink based steganography in the real world are summarized as follows:
1) Prenegotiated manipulations are indispensable steps for the correct extraction of genuine secrets. When certain types of invisible inks are used, corresponding development methods must be performed on the received paper to reveal the hidden message.
2) The received paper may be seriously deformed due to prenegotiated manipulations, e.g., the paper may become lumpy after heating. Since extraction of the genuine message is the real goal, visual quality of the paper after manipulation is often out of concern.
3) Cover messages play an important role as camouflage during delivery of genuine messages. However, spaces where genuine secrets can be invisibly written are always available since hand-written or printed documents always
leave blank spaces for the ease of reading.
B. Steganography versus Steganalysis
After entering the digital era, digital media like images or audio clips serve as good cover objects for carrying secret messages. Therefore, digital data-hiding techniques are adopted to implement steganography systems. In a data-hiding system, the sender embeds some messages, also denoted as watermarks for
some other applications, into the cover work for generating a
Fig. 1.1 Real-world espionage scenario using the invisible ink.
perceptually acceptable stegowork. Afterwards, the receiver extracts the message from the received stego work. In-depth discussions about data-hiding systems can be found in . In the literature of digital steganography, fidelity (the visual quality of the stego work) and capacity (the maximally allowable message
length) are most important performance measures. On the contrary, steganalysis is the practice of attacking steganographic schemes by detecting, destroying or extracting the hidden message, as introduced in. Supervisors of communication channels may adopt adequate steganalysis tools to prevent unexpected communications via transmitted content. Generally speaking, deciding whether a cover work carries hidden messages is a difficult task. When the channel supervisor has the right to cease any doubtful communication, an accurate steganography detection module suffices for all his needs. Alternatively, a channel supervisor may simply introduce imperceptible or acceptable distortions to all incoming contents in expectation of hindering the extraction of potentially hidden messages. In this case, the supervisor should devise versatile and effective noises applicable to all delivered contents, while designers of steganography systems should consider robustness against potential distortions, as well as prescribed fidelity and capacity requirements. Besides, the supervisor may try to eavesdrop and interpret the secret messages. Therefore, cipher modules are often applied to the messages delivering by the sender to prevent the unauthorized receivers from reading the secret message. Note that, in the literature, covert communications are often modeled as “the prisoners’ problems” given different assumptions and the channel supervisor is therefore denoted as a warden, as firstly introduced by . According to definitions specified in , a channel supervisor who can do nothing but only spy on the communication channels is named as a “passive” warden. On the other hand, the supervisor who is allowed to slightly modify the data being sent from the sender to the receiver is thereby an “active” warden. In this paper, most of our discussions focus on the passive-warden scenario. But for the completeness of our works, issues about the proposed methodology against the more difficult “active-warden” configuration is discussed in Section III-E.
C. Plausible Deniability
Plausible deniability is originally a term used in politics. It means the creation of loose and informal chains of commands in government, which allow controversial instructions given by high-ranking officials to be denied if these instructions become public. In the field of cryptography, deniable encryption allows an encrypted message to be decrypted to different meaningful plaintexts, depending on the key used. This allows the sender to have plausible deniability if he is compelled to give up his encryption key. But in strictly-defined modern cryptography, it is almost impossible to design a ciphertext that can be decrypted
to several different meaningful plaintexts. In the literature of steganography, plausible deniability means the capability to deliver some genuine message under the cover of other innocuous messages. When the existence of hidden information
is detected and the sender is forced to reveal the secret message, he can simply turn in one innocuous message and claim that no other information is hidden. As an example, the aforementioned real-world steganography system illustrates
such a behavior. Plausible deniability has been proposed to enhance the security of steganography systems and defend current steganalysis, as described in . In this paper, instead of diving into details of various plausibly deniable schemes, some high-level discussions about implementing plausibly deniable steganographic systems based on generic watermarking techniques, and the comparisons with the proposed system are provided in Section IV-A.
D. The “Digital” Invisible Ink
In this paper, steganography systems based on a “digital” version of invisible ink, denoted as digital invisible ink (DII), are proposed. Since we try to implement a digital version of such invisible-ink system based on existing watermarking schemes, corresponding characteristics of invisible ink listed in Section I-A shall be adequately implemented.
1) Only when the stego work undergoes certain prenegotiated manipulations, hidden messages will be correctly extracted. In subsequent discussions, the prenegotiated manipulations are media processing procedures that always cause distortions to the stego work. Note that in our digital bimplementations, the types and degrees of manipulations are carefully controlled and viewed as keys to achieve better security.
2) To extract the genuine secrets, the intended receiver will deliberately and seriously distort the marked work. But for the channel supervisor or non-intended users, the marked work is still perceptually similar to the original cover work.
3) In the case of plausibly deniable steganography, the payloads extracted by the intended receiver will consist of both a cover message and a genuine message. The intended receiver can easily distinguish between the cover message and the genuine message because he can also extract the cover message solely. In some interesting cases, we will show that the cover message can be devised to help interpretation of the genuine messages.
Note that the idea of digital invisible ink data hiding is firstly revealed in and then briefly exploited in by the authors. In this paper, we thoroughly describe the motivation, implementations and experimental results of digital-invisible-ink
data hiding. Two major data-hiding schemes, the spread-spectrum watermarking and the quantization-based watermarking, are adopted to implement the digital invisible ink and respectively discussed in Sections II and III. In addition to implementation details and experimental results, applications and inherent
limitations of each scheme are also discussed. Section IV gives extensive discussions about plausible deniability and illustrates the superiority of the digital-invisible-ink methodology. Section V concludes this paper and states our future work.