Name : Johann Barbier

Institution : Délégation Générale pour l’Armement (DGA)

                        Associated researcher at ESIEA (Laval F-53 000)


Johann Barbier received his PhD in Computer Science from Ecole Polytechnique, Palaiseau, France, in 2007. He received his MS degree in algorithmic in 2003 from Ecole Polytechnique. He was graduated from the French Military Academy of Saint-Cyr, Coëtquidan, France, in 2002 and received a MS degree in Computer Science Engineering.

Johann Barbier joined the DGA in 2003 as an officer and is currently a cryptologist engineer at the cryptology department of the Centre d’ELectronique de l’ARmement (CELAR), Bruz, France. He has been lecturer at the French Military Academy of Saint-Cyr, Coëtquidan, since 2002 and at French Army Signals Academy, Rennes, France, from 2003 to 2008.  He has been an associated researcher at the ESIEA with the Operational Cryptology and Virology Laboratory since 2008.

Johann Barbier has two main research interests. The first one is electronic warfare and more particularly binary stream analysis. He designed and improved algorithms to blindly recover the parameters of error correcting codes from noisy intercepted bit streams. The second one is steganography. He proposed formal models to practically evaluate the security of steganography schemes.


Publications :

J. Barbier and E. Mayer. Non Malleable Scheme Resisting Adaptive Adversaries In Proc. of Digital Watermarking, 7th International Workshop, IWDW 2008, Best Paper Awarded.

J. Barbier and S. Alt. Practical Insecurity for Effective Steganalysis In Proc. of Information Hiding, 10th International Workshop, IH 2008.




Title of Project : Modern Steganography


Steganography is very ancient techniques to hide secret messages inside innocuous covers.  Modern steganography provides a new security goal for privacy protection of digital data: The transmission security (TRANSEC). More than confidentiality, it guaranties the invisibility and so the anonymity of digital communications.  To achieve this, the plaintext is first compressed, then encrypted with a cryptographic algorithm and finally inserted in the cover medium. To embed the ciphertext, positions within the cover medium are pseudo-randomly chosen according to a secret key. The ciphertext is finally written at these positions. As the process is entirely symmetric, the receiver finds the positions according to the secret key and reads the ciphertext at these positions.

From the adversary’s point of view, we have to determine if a given medium contains hidden information or not. The main detection techniques are based on statistical measures performed on the analyzed media. These measures are random variables which do not follow the same density probability law if the medium is a cover or a stego one. The main drawback of such techniques is that steganographic detectors are intrinsically not able to distinguish between stego media which contain only one ciphertext and stego media which contain more than one. To make up for that issue in a forensics context, some techniques estimate the length of the ciphertext. Unfortunately, these techniques are very few and are specific to particular steganographic schemes.