A stegosystem with blind decoder based on a noisy channel

We consider an unconventional form of steganography (SG) in which the attacker can see the cover signal only after that signal has passed through a noisy channel. Some examples of such scenario are presented. But in contrast to our previous investigation of this problem, in cases where only an informed decoder has been studied, we investigate in the current paper a blind decoder. This approach is very important for the cases when the cover message cannot be got by the receiver of the stegosignal in a simple manner. In order to increase the embedding rate we propose the use of an informed encoder based on quantization projective modulation (QPD). The error probability of the extracted bits is investigated both within a theoretical approach as well as through computer simulation. Since an application of the QPD to a stegosystem may result, generally speaking, into a security reduction, we consider several new attacks on such stegosystems and their efficiency is estimated. In order to provide a greater security, we propose the use of the so called spread-time stegosystem where the embedding of secret information into samples of cover message is performed into several samples controlled by a stegokey.