4
$\begingroup$

The answer in this question defined how to calculate the probability of rotational cryptanalysis on modulo multiplication $\odot$. This paper defined an algebraic equation of how to calculate the rotational probability of modular addition $\boxplus$ which is

$$ \textbf{P}(\overrightarrow{X\boxplus Y} = \overrightarrow{X}\boxplus\overrightarrow{Y}) = \frac{1}{4}(1+2^{-r} + 2^{r-n}+ 2^{-n}) $$ where $r$ is rotational value and smaller than modular value $n$

and experimental results of 8-bit ($n$) addition shows similar values of

$$ \begin{array}{c|c} r & f \\ \hline 0 & 65536 \\ 1 & 24768 \\ 2 & 20800 \\ 3 & 19008 \\ 4 & 18496 \\ 5 & 19008 \\ 6 & 20800 \\ 7 & 24768 \\ \end{array} $$

where f is the frequency of $ \overrightarrow{X\boxplus Y} = \overrightarrow{X}\boxplus\overrightarrow{Y} $ for each $r$ . the probability of 1 bit rotation is $\frac{24768}{65536}$ is around $.378$ (nearly same as in the table in ). the other thing is you find reflection of frequency values , 1 bit rotation is equal to 7 bit rotation (2 and 6 so on). The following figure shows a flipped bell distribution of rotational modulo addition.

enter image description here

The next step

I conducted the same experiment (8 bit size) but with the following equation (double rotation).

$$ \overrightarrow{X} \odot \overrightarrow{Y} = \overrightarrow{\overrightarrow{(X\odot Y)}} $$ the following table shows the frequency:

$$ \begin{array}{c|c} r & f \\ \hline 0 & 65536 \\ 1 & 4730 \\ 2 & 1082 \\ 3 & 904 \\ 4 & 1664 \\ 5 & 904 \\ 6 & 1082 \\ 7 & 4730 \\ \end{array} $$

enter image description here from the table and figure, i see there is a reflection of frequency values (suggesting a similar equation of rotational cryptanalysis probability is applied as following:

$$ \textbf{P}(\overrightarrow{X} \odot \overrightarrow{Y} = \overrightarrow{\overrightarrow{(X\odot Y)}}) = \frac{1}{t}(1+2^{-e_0} + 2^{e_0-e_1}+ 2^{-e_1}) $$

Question

if my guess is right , how to calculate the $t , e_0 , e_1$ values?

NOTE Iam happy to share the source code in the question.

$\endgroup$
2
  • 1
    $\begingroup$ Ask a clear question up-front. What does "guess" refer to? How in the world does your "guess" change the fact of how the t,e0,e1 values will be calculated? Which experiment are you referring to? $\endgroup$
    – Patriot
    Oct 8, 2019 at 10:20
  • $\begingroup$ if i understand your comment right , I mapped the equation used in addition expression to multiplication. my guess was t, eo, e1 are only function of rotation constant (r) and size (n) , however , after ploting the tables of rotational addition and multiplication , I found the two plots are different , the addition plot is smooth fliped bell distribution , the multiplication is not (there is a bumb when r =4 f=1644) , I guess the proposed equation is not correct $\endgroup$
    – hardyrama
    Dec 17, 2019 at 9:05

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.