2007-03-20

Speaking of steganography...

One of the true odd ducks of the internet, Shannon Larratt, created a good proof of concept for a randomized steganography implemented by using Markov chains. His post on his blog Zetastic.com two years ago showed his method and some sample meta code for the process. Also included is a demo Win32 app that performs the functions. One of my goals of re-learning C++ lately had been to write a simple function or library to preform a Markov
chain encoding for data. The main draw back to using this process is the vast increase of space needed to store the output. Every bit will be represented by a entire word of several bytes. One way to help with that may be to use a large randomized pad of data. Then use that instead of text for the seed.

This is one form for steganography that may be useful as a publicly visible data. Through a judicious use of seed text it may be possible to use a text that is both large, and contains very few "identifier" words that would reveal the original text. Thus, without the original text (seed/key) it may be impossible to generate the Markov chains and build the separate sets needed to decode the data.

blog comments powered by Disqus