Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Cryptanalysis of this New Cryptography

9 views
Skip to first unread message

adacrypt

unread,
Feb 5, 2012, 9:08:50 AM2/5/12
to
I have made my pitch quite confidently for both vector and scalar
cryptography recently and I can demonstrate three ciphers to hand that
use this new cryptography.

I want to demonstrate the cryptography now from a cryptanalysts
perspective and I want to show how strong the crypto is under any
known form of attack.

In particular I want to present statistical figures for the ciphertext
frequency of both crypto types (they are very close to each other)
that readers can use to judge for them selves.

I am going to use the vector cryptography of “Skew Line Encryptions “
only as my example in presenting the relative frequencies of the
plaintext and corresponding ciphertext in a file of 10852 characters.

The file that I tested contained 10852 alphanumeric characters.

The ciphertext frequency resolves to,

7859 non-repeats i.e. 7859 absolute elements
1216 once-repeated i.e. 2432 absolute elements
165 twice-repeated i.e. 495 absolute elements
14 triple repeats i.e.56 absolute elements
2 four-time repeats i.e. 10 absolute elements.

The plaintext resolves to,

‘ ‘…1827 (spacebar),’“’ ….8, ‘’’..…23, ‘(‘ …..13
‘)’…..13, ‘,’....115, ‘-‘.….28, ‘.’…80, ‘/’ …..3
‘0’….4, ‘1’….9, ‘2’….13, ‘3’….6, ‘4’….1, ‘5’….14
‘6’….14, ‘7’…..3, ‘8’…..3, ‘9’…..3, ‘:’….2, ‘;’….2
‘<’…..2, ‘>’…..2, ‘?’….1, ‘A’…20, ‘B’….3, ‘C’….10
‘D’….6, ‘E’….13, ‘F’….5, ‘G’….2, ‘H’….7, ‘I’…..37
‘J’…..1, ‘L’….10, ‘M’…14, ‘N’….4, ‘O’….7, ‘P’….4
‘R’….6, ‘S’….16, ‘T’….27, ‘U’….30, ‘W’…3, ‘X’….5
‘Y’….4, ‘a’…682, ‘b’…122, ‘c’…372, ‘d’…309
‘e’…1084, ‘f’….171, ‘g’….166, ‘h’….361
‘i’…..576, ‘j’…..8, ‘k’….26, ‘l’…..306, ‘m’…206
‘n’….613, ‘o’….607, ‘p’….192, ‘q’…..6, ‘r’…..555
‘s’…..563, ‘t’…...882, ‘u’…..236, ‘v’…..88, ‘w’….98
‘x’…..72, ‘y’…..126 ‘z’……2

Clearly, there is no correlation whatever between the ciphertext and
the plaintext frequencies that might enable statistical mapping on a
probabilistic basis.

This is not a one-off event – this exceptionally high pseudo
randomness in the ciphertext is designed to be sustainable and can
always be counted on in this cryptography – it is a feature of the
cipher design..


Other common attacks.

1) The most direct attack is the ciphertext only attack and this is
annulled by the strength of the encryption algorithm which is
demonstrably unbreakable in all of these ciphers.

2) The next attack is a numerical attack by linear analysis or
differential analysis on the ciphertext. This is foiled by the
ciphertext in this cryptography being a string of ‘jump’
discontinuities. The latter totally rules out that form of attack.

3) The next attack is in using the ciphertext as a key in its own
decryption by attempting a statistical mapping operation of the
ciphertext onto the plaintext in a probalistic attempt. This attack
is foiled by the randomness of the ciphertext string when it is viewed
as a possible key in the hands of an adversary. That attack is
impossible as is shown in the demonstration above by ensuring a high
degree of ‘pseudo’ randomness in the cipher text itself (albeit the
keys that produce the ciphertext are truly random themselves the
ciphertext in general is never random in itself. This cryptography
seeks to achieve that as aa bonus of the cipher design.)

I am claiming that this is a bonus feature of this cryptography in
that the ciphertext has such a high pseudo-randomness about it that
there is no danger of statistical mapping of the ciphertext to a
probabilistic messagetext.

The frequencies demonstrated are sustainable and can be availed of in
all future cases.

PS - this file test was a complete afterthought - it is an honest test
completely unplanned and unbiased.

- adacrypt


adacrypt

unread,
Feb 5, 2012, 2:40:53 PM2/5/12
to
Further Comment.

This experiment was conducted on just one column of coefficients in
the packages of change-of origin vectors in the computer program i.e.
the ‘j’ coefficients column. The test recorded the randomness of that
column on the basis of non-repeats found in that coefficient alone to
be 72% i.e. there were 7859 non-repeats out of a possible 10852
elements thus giving 72 %.

It did not take into account however if the remaining two coefficients
of any perceived repeated (non qualifying) vector were also equal i.e.
the (i) and (k) coefficients of a so-called repeat should also be
equal in order for it to be verified as being a bona fide repeat.
That was not done and statistically it is very unlikely to be the case
that they were in fact that many real repeats on that basis.

It can be expected therefore that if this experiment is repeated and
conducted so that in the case of a perceived repeat that the other two
(not checked) coefficients are also checked (a bit difficult) before
recording it as a repeat, then the pseudo randomness as a percentage
of true randomness would be very near to perfectly true random i.e.
100%. I would expect 98%.

In any case this is a very secure cipher that is being understated by
the results of the experiment described on the ciphertext. A cipher
that produces 100% random ciphertext that was itself randomly
generated originally using two random key sets is unheard of surely.
It can be quite feasibly achieved with a bit of care in testing.

- adacrypt
0 new messages