Secure adaptive Steganography

Basic data for this project

Description

Adaptive steganography is a hiding technique to embed secret messages into covers by selecting the most secure embedding positions within a heterogeneous cover signal. Current adaptive techniques are designed without considering that differences in the detectability between different parts of a cover in fact leak side information to the detector. This enables a passive adversary to separate parts with presumably high embedding density from parts with presumably low embedding density, or to compare between both parts. This project for the first time combines signal processing and game theory to study distribution functions for embedding changes which anticipate such adversarial strategies. This way, we expect novel theorems on the achievable security of adaptive steganography and new insights for the construction of secure adaptive embedding functions.