Cryptography Reference
In-Depth Information
property of the Identify algorithm to run over any subset, we obtain a fully
collusion resistant code.
We also note that the above definition leaves open the possibility for a
secret scheme where the Forging algorithm has no access to the whole-code
C generated by the CodeGen algorithm. While keeping the code secret will
prove to be advantageous for the purpose of identifying a traitor as the traitor
coalition has less information in constructing the pirate codeword, there are
many cases where in an actual deployment of fingerprinting codes one would
prefer an open fingperinting code, i.e. having the code publicly available (or
even fixed - uniquely determined by n). A variant of the above definition where
the Forging algorithm is not only given the traitor codewords C
T
= {c
j
| j ∈
T} but also the code C as input gives rise to open fingerprinting codes. Taking
this a bit further, a one may additionally provide the key tk to the attacker
as well; this would be termed a public fingerprinting code.
1.3 Applications to Digital Content Distribution
Fingerprinting codes play an important role in the area of encryption mecha-
nisms for digital content distribution. Encryption mechanisms can be designed
to take advantage of a fingerprinting code by having a key-space for encryption
that is marked following a fingerprinting code. In such case, a user codeword
in the code describes the particular sequence of keys that are assigned to the
user. The encryption of the content is then designed in such a way so that
the recovery of the content requires a valid key sequence. Assuming it is pos-
sible to figure out what keys are stored in a pirate decoder this would provide
a pirate codeword at the code level and the identification of a traitor user
would be achieved by calling the identification algorithm of the underlying
fingerprinting code.
The integration of a fingerprinting code with the encryption mechanism
requires three independent and orthogonal tasks: (i) Designing the content en-
cryption mechanism so that the key-space is distributed among the receivers
according to a fingerprinting code. (ii) Detecting the keys used in the pirate
decoder. (iii) Applying the identification algorithm of the underlying finger-
printing code.
Still, this is not the only way we may apply fingerprinting codes in our set-
ting. To see another possible scenario consider an adversarial scenario where
the pirate is entirely hiding the keys it possesses and rebroadcasts the clear
text content after decrypting it. This would entirely numb any attempt to
catch a traitor on the basis of a decryption key pattern. A different approach
to address this issue that also utilizes fingerprinting codes would apply wa-
termarking to the content itself. Naturally, to make the detection of a traitor
possible, the watermarking should be robust, i.e. it should be hard to remove
or modify the embedded marks of the content without substantial decrease in
