11.07.2014 Views

Cryptanalysis of RSA Factorization - Library(ISI Kolkata) - Indian ...

Cryptanalysis of RSA Factorization - Library(ISI Kolkata) - Indian ...

Cryptanalysis of RSA Factorization - Library(ISI Kolkata) - Indian ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Chapter 2: Mathematical Preliminaries 26<br />

In short, one can write the CF representation as [q 1 ,q 2 ,q 3 ,...,q m ]. For example,<br />

take the rational number 34<br />

34<br />

. One can write this as = [0,2,1,10,3], that is,<br />

99 99<br />

0+ 1 99<br />

34<br />

= 0+ 1<br />

2+ 31<br />

34<br />

= 0+ 1<br />

2+ 1 34<br />

31<br />

= 0+<br />

1<br />

2+ 1<br />

1+ 3<br />

31<br />

= 0+<br />

1<br />

2+ 1<br />

1+ 31<br />

1<br />

3<br />

= 0+<br />

1<br />

2+<br />

1<br />

1+ 1<br />

10+ 1 3<br />

.<br />

Now consider a subsequence [0,2,1] <strong>of</strong> [0,2,1,10,3]. Note that 0 + 1 =<br />

2+ 1 1<br />

1<br />

= 33<br />

34<br />

, which is very close to . This indicates that a subsequence <strong>of</strong> a CF<br />

3 99 99<br />

representation may produce a good approximation to the rational number. Any<br />

initial subsequence <strong>of</strong> [q 1 ,q 2 ,q 3 ,...,q m ], i.e, [q 1 ,q 2 ,q 3 ,...,q r ], where 1 ≤ r ≤ m<br />

is called a convergent <strong>of</strong> the original sequence [q 1 ,q 2 ,q 3 ,...,q m ]. For example,<br />

[0,2,1] is a convergent <strong>of</strong> [0,2,1,10,3], which implies that 1 = 33 is a convergent<br />

3 99<br />

. Also note that if the subsequence has a 1 at the end then it may also be<br />

<strong>of</strong> 34<br />

99<br />

representedbyaddingthat1tothepreviousintegerandthusshorteningthelength<br />

<strong>of</strong> the subsequence. For example, both [0,2,1] and [0,3] provide the same rational<br />

number. There are many interesting results about the convergence <strong>of</strong> continued<br />

fractionrepresentations. Forthepurposeathand, weneedthefollowingresult[47].<br />

Theorem 2.10. Suppose gcd(a,b) = gcd(c,d) = 1 and | a b − c d | < 1<br />

2d 2 . Then c d is<br />

represented by one <strong>of</strong> the convergents <strong>of</strong> the continued fraction expansion <strong>of</strong> a b .<br />

It is also important to note that for t bit integers a,b, the CF expression<br />

[q 1 ,q 2 ,q 3 ,...,q m ] <strong>of</strong> a can be computed in O(poly(t)) time and can be stored in<br />

b<br />

O(poly(t)) space. Let us see how this may be utilized to attack <strong>RSA</strong>.<br />

Wiener’s Attack<br />

We have ed ≡ 1 (mod N). So we can write ed = 1 + kφ(N) for some integer k.<br />

So,<br />

e<br />

φ(N) − k d = 1<br />

dφ(N) .

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!