Dynamic Frameproof Codes

by Maura Paterson

Abstract:

There are many schemes in the literature for protecting digital data
from piracy by the use of digital fingerprinting, such as frameproof
codes, which prevent traitorous users from colluding to frame an
innocent user, and traitor-tracing schemes, which enable the
identification of users involved in piracy. The concept of traitor
tracing has been applied to a digital broadcast setting in the form of
dynamic traitor-tracing schemes and sequential traitor-tracing
schemes, which could be used to combat piracy of pay-TV broadcasts,
for example. In this thesis we explore the possibility of extending
the properties of frameproof codes to this dynamic model.

We investigate the construction of l-sequential c-frameproof codes,
which prevent framing without requiring information obtained from a
pirate broadcast. We show that they are closely related to the
ordinary frameproof codes, which enables us to construct examples of
these schemes and to establish bounds on the number of users they
support. We then define l-dynamic c-frameproof codes that can prevent
framing more efficiently than the sequential codes through the use of
the pirate broadcast information.  We give constructions for schemes
supporting an optimal number of users in the cases where the number c
of users colluding in piracy satisfies c greater than or equal to 2 or
c=1.

Finally we consider sliding-window l-dynamic frameproof codes that
provide ongoing protection against framing by making use of the pirate
broadcast.  We provide constructions of such schemes and establish
bounds on the number of users they support. In the case of a binary
alphabet we use geometric structures to describe constructions, and
provide new bounds.  We then go on to provide two families of
constructions based on particular parameters, and we show that some of
these constructions are optimal for the given parameters.