Stéganographie résistante à la détection du secret par conservation de l’intégrité des médias de couverture

Abstract

Steganography is a branch of cybersecurity that aims to conceal the existence of a communication by embedding a secret message within covert media such as text, images, audio, or video. Traditional steganography approaches rely on modifying the media, which poses an increasing risk of detection when faced with media forensic ana lysis techniques. However, the coverless approach avoids any modification of the media by selecting media whose content is similar to the secret message after applying a hash function. This approach offers a promising alternative but is still limited in terms of ca pacity, robustness, and flexibility (difficult to adapt to different secret message contexts or various media types). This thesis proposes a new paradigm of steganography without modification of the medium based on the dissociation between the content of the me dium and the secret information, using a system of indexing files in directories and an indirection mechanism for storing and sharing files in multi-cloud environments. The model is based on four fundamental properties : the use of any type of files (text, image, video, etc.) as potential medium, an indexing of the media from the digital descriptors of files (modification date, file type, file name, file size, etc.), an indirection between the two communicating parties ensured by the distribution of the files and their sharing through the storage services of public clouds and finally the reconstruction of the message based on the order of the selected files and the intervening clouds. The experimental results show a better resistance to the alteration of the medium and a great flexibility compared to methods based on the coverless. In addition, the analysis of the cost of using cloud storage spaces is null. Future work will involve reducing the size of the file database required to encrypt the secret by proposing a method that both reduces the number of files per directory and increases the insertion capacity. Then, optimize the method by reducing the number of files moved within the multi-cloud storage environment during concealment.