Perturbation hiding and the batch steganography problem

Abstract

The batch steganography problem is how best to split a steganographic payload between multiple covers. This paper makes some progress towards an information-theoretic analysis of batch steganography by describing a novel mathematical abstraction we call perturbation hiding. As well as providing a new challenge for information hiding research, it brings into focus the information asymmetry in steganalysis of multiple objects: Kerckhoffs' Principle must be interpreted carefully. Our main result is the solution of the perturbation hiding problem for a certain class of distributions, and the implication for batch steganographic embedding. However, numerical computations show that the result does not hold for all distributions, and we provide some additional asymptotic results to help explore the problem more widely. © 2008 Springer Berlin Heidelberg.