CRYPTO '02 Proceedings of the 22nd Annual International Cryptology Conference on Advances in Cryptology
Attacks on Steganographic Systems
IH '99 Proceedings of the Third International Workshop on Information Hiding
A Concrete Security Treatment of Symmetric Encryption
FOCS '97 Proceedings of the 38th Annual Symposium on Foundations of Computer Science
An information-theoretic model for steganography
Information and Computation
Toward a theory of steganography
Toward a theory of steganography
The ultimate steganalysis benchmark?
Proceedings of the 9th workshop on Multimedia & security
On steganographic chosen covertext security
ICALP'05 Proceedings of the 32nd international conference on Automata, Languages and Programming
Upper and lower bounds on black-box steganography
TCC'05 Proceedings of the Second international conference on Theory of Cryptography
Universal JPEG steganalysis in the compressed frequency domain
IWDW'06 Proceedings of the 5th international conference on Digital Watermarking
| Hi-index | 0.00 |
In this paper, we propose to link practical steganalysis and classical security models. In one hand, some steganography schemes are proved to be secure in some security models but in practice, they cannot be used in real-life because of too many constrains. In the other hand, lots of practical steganography algorithms have been broken by effective steganalysis but without any connection with standard models of attackers. So, we introduce two new types of adversaries to simulate real attackers, the IND-SSA adversary and the IND-USA adversary. The IND-SSA adversary emulates specific steganalysis whereas the IND-USA attacker stands for universal steganalysis. We also define the games and the security models associated with and formalize discrimination attacks and discriminant steganalysis. Then, we connect these new models with the hierarchy of classical security models. Using models introduced by C. Cachin, S. Katzenbeisser and F. Petitcolas, and finally by N. Hopper, we show how effective steganalysis gives us a lower bound on the insecurity of the steganalyzed steganography schemes. We also point out that steganography schemes which are not secure in these new models are also not secure in the classical models of security.