Search

This chapter presents an overview of text-based and binary-based data hiding methods. Text methods, through which secret information is embedded into innocent-looking textual data, are mostly used for steganography. Binary methods are applied to program binary codes: executables and libraries. In binary methods, information is embedded into a binary code so that its functionality is preserved. Data hiding methods for binary codes have been studied intensively to perform watermarking for protecting software from piracy acts. A message can also be embedded into a binary code in a steganographic manner. Another method is also introduced, which is proposed for enhancing the performance of an executable file.

Creation of a stego object by embedding information in a cover object often distorts the cover object. As more information is embedded, more annoying noise is introduced in stego objects. Although reversible embedding methods enable us to restore the original cover object even after embedding, stego objects are not free from distortions. Embedding information does not, however, always result in damaging the contents of the cover object. This chapter introduces data embedding methods that are not based on modification of the contents of cover objects: permutation steganography, metadata steganography, and cover generation methods. This chapter focuses on elaborating the basic principles of these techniques.

Cryptographic filesystems are widely used to protect private files. It is, however, impossible to hide the existence of private information by such filesystems. Steganographic filesystems attempt to address this problem by embedding files imperceptibly into containers. In most steganographic filesystems ever proposed, files are embedded into containers those apparently randomized. Their existence would, however, imply that they include hidden files. This paper presents a new steganographic filesystem manager called Crocus. When a filesystem is to be hidden, it is embedded separately into a set of innocent-looking containers piece by piece. When the filesystem is to be used later, it is reconstructed from the pieces. We can resize or destruct the filesystem if required. Since more than one containers can be used for one filesystem, we can build filesystems those large enough. A prototype system of Crocus has been developed for Linux and a preliminary experiment was performed. The result indicates the effectiveness of the framework of Crocus.