Cryptography Reference
In-Depth Information
13.2 Introduction
Most of the algorithms in this topic hide information in data formats
that are relatively rigid. Image files must describe the color of each
pixel in a well-defined order. Audio files are pretty much required to
describe the sound at each point in time. Hiding the information in
specific places in the noise is a pretty good gamble because the files
aren't going to be rearranged.
Some data is not as rigid. Text documents can have chapters, sec-
tions, paragraphs and even sentences rearranged without changing
the overall meaning. That is to say, the overall meaning of a text doc-
ument isn't changed by many ways of rearranging sentences, para-
graphs, sections or chapters. There's no reason why the elements in
a list can't be scrambled and rescrambled without the reader notic-
ing. Many of the chapters in this topic can be reordered a bit without
hurting the overall flow.
Even files with a strong fixed relationship with time and space can
be rearranged. Image files can be cropped, rotated, or even arranged
in a tile. Songs can be rearranged or edited, even after being fixed in
a file. All of these changes can rearrange files.
When the data can be reordered, an attacker can destroy the hid-
den message without disturbing the cover data. Mikhail Atallah and
Victor Raskin confronted this problemwhen designing amechanism
for hiding information in natural language text. If they hid a bit or
two by changing each sentence, then they could lose the entire mes-
sage if the order of each sentence was rearranged.[ARC
+
01] Many of
the text mechanisms in Chapters 6, 7, and 8 are vulnerable to this
kind of attack. One change at the beginning of the datastream could
confound the decoding of everything after it. Many of the solutions
in Chapter 9 store the bits in random locations dictated by some
pseudo-random stream of values. Destroying the bits early in this
stream can trash the rest of the file.
“There it was, hidden in
alphabetical order”-
Rita Holt
The information stream has two components: the data and the
location where it is stored. Most of the algorithms in this topic con-
centrate on styling and coiffing the data until it assumes the right
disguise. The location where the data goes is rarely more than a tool
for adding some security and the algorithms pay little attention to
the problem.
This chapter focuses on the second component: how to defend
and exploit the location where the information is held. Some of the
algorithms in this chapter insulate data against attackers who might
tweak the file in the hope of dislocating the hidden information un-
derneath. They provide a simple way to establish a canonical order
