Abstract:
Stream ciphers are known to have less complex hardware design and are generally faster than
the block ciphers. Their employment is necessitated in systems where buffering is limited and
bits must be processed immediately upon arrival or when data comes in unknown lengths like in
wireless communication.
Till the advent of algebraic attacks on stream ciphers, a decade and a half ago, linear feedback shift
register (LFSR) based stream ciphers were quite common due to their efficient implementation.
However, later years saw a host of new stream ciphers based on nonlinear feedback shift registers
(NLFSR), which were highly resistant against traditional algebraic cryptanalysis. At present the
success of algebraic attacks against stream ciphers is only limited to lightweight stream ciphers
or reduced variants of popular stream ciphers.
Side Channel Analysis (SCA) being an implementation attack, requires access to the hardware
implementation of the target stream cipher to capture side channel leakages associated with internal
processing. However to find exact secret key or internal state, a handsome quantity of leakage
information is needed. In contrast to algebraic attacks, research work on SCA against stream
ciphers has been less frequent but more successful.
Combining two cryptanalysis techniques can pay dividends not only in the form of achieving
better attack complexities and facilitating exploitation of new vulnerabilities, but also lowering
the effect of individual weaknesses of attack techniques. Algebraic attacks and SCA can both
be strong candidates for such combination. Rather they are already being applied against block
ciphers in unison for quite some time now. However no published work as of this writing can be
found on combining algebraic cryptanalysis and SCA against stream ciphers.
In this work we propose to combine algebraic attacks with SCA, on stream ciphers, in a manner
that reduces overall attack complexity/ difficulty as compared to isolated application of the two
constituent attacks. This combination attack, termed as Algebraic Side Channel Attack (ASCA),
utilize whatever side channel leakage is available from the target stream cipher’s implementation
in limited time exposure and converts it into algebraic equations. These SCA equations are added
into the system of multivariate polynomial equations obtained from algebraic attack technique.
Union of these equation sets, can then be resolved through any mathematical method/ tool, to find
the unknown variables, i.e., either secret key or internal state of the stream cipher.
Stream ciphers of today, which mostly rely on nonlinearity by design to make them extremely
immune to algebraic cryptanalysis, can also fall prey to ASCA. To demonstrate this, we successfully
attack Crypto-1, Bivium-B, Trivium and Grain stream ciphers using our proposed technique
in 0.158, 11.531, 21.54 and 28.25 seconds respectively.