Stretching the limits of steganography

Abstract

We present a number of insights into information hiding. It was widely believed that public key steganography was impossible; we show how to do it. We then look at a number of possible approaches to the theoretical security of hidden communications. This turns out to hinge on the inefficiency of practical compression algorithms, and one of the most important parameters is whether the opponent is active or passive (i.e., whether the censor can add noise, or will merely allow or disallow a whole messages). However, there are covertexts whose compression characteristics are such that even an active opponent cannot always eliminate hidden channels completely.