Abstract:
The notion of Deniable Encryption is a cryptographic primitive, which
enables legitimate users to face coercion by dynamic adversaries without
revealing true secret internals of the cryptosystem. Deniable Encryption
provides a way to generate fake internals that correctly explain the cipher text.
When considering existing deniable schemes, two major variations can be
found; schemes based on the concept of Deniable crypto-systems introduced by
R. Canetti et al. and plausible deniable schemes. The schemes based on plausible
deniability are not always depending on cryptographic systems, but rather use
different approaches such as steganography or hardware level hidden volumes.
With the objective of cryptanalysis, this research has been focused on deniable
crypto-systems.
The existing deniable encryption schemes proposed provide different levels
and types of deniability, which makes it difficult to find a common model for the
cryptanalysis. Therefore, This research has narrowed down the cryptanalysis to
full-sender-deniable encryption, which is the strongest notion in sender
deniability.
In order to evaluate the real world implementation of full-sender-deniable
encryption, this research has implemented a crypto-system using sparse-set. This
research has also introduced a new type of sparse-set generation, which provides
better performance compared to the two sparse-set generation methods proposed
by Canetti et al.
Based on the common model of full-sender-deniable encryption, our
cryptanalysis has been focused on three main areas; deniability limitation
already given by Canetti et al., statistical cryptanalysis and cryptanalysis based
on faking algorithm. Since the encryption function of full-sender-deniable
encryption is a public parameter, the adversary can coerce the sender to generate
randomness by further faking and have additional data to detect the original
faking. This is a new scenario that has been considered in this research, where it
can be applicable in situation like rubber hose cryptanalysis.
