Information hiding through noisy channels

Abstract

We consider a scenario where information hiding (IH) is performed through noisy channels. There may arise different situations but one of the most common is the case where the legal IH channel is superior to the attacker IH channel. If a special randomized encoding is used by legal users then it is possible to hide information in the noisy components of the cover message. At the same time, the randomized encoding prevents the secret message to be removed from the stegomessage without any significant distortion of the cover message. If a legal decoder of IH knows the cover message, a randomized encoding procedure does not prevent the error correction of the secret message at the receiving side. The special problem of IH - how to distinguish any binary periodic repetitive sequence from truly random binary noise received on noisy channels - is presented. Application of the randomized encoding technique makes a solution to this problem more difficult and hence complicates a traffic analysis. We consider also how is it possible to "camouflage" IH by natural channel noises independently of the properties of the cover messages probability space, and the application of WMin combination with randomized encoding dedicated to the utilization of noisy channels. If an attacker tries to remove WMb y adding binary noise then an expansion of errors in the cover message results.