Making the Ciphertext String Itself Truly Random – The Conjecture.

I have made the point recently that the ciphertext string is never
random and indeed the clever cryptanalyst who knows this will try and
determine to what extent that is instantaneously true so that he may
mount a statistical mapping attack using that as the basis (an
advanced Kasiski/Babbage attack).

My ploy of studiously making the ciphertext string truly random or at
least very close to that is the appropriate foil to this last ditch
attack by the cryptanalyst.

I claim that this is an innovative new approach in cryptography.

Bearing in mind that all current number-theoretic cryptography
ridiculously puts the plaintext and the cipher text as neighbours on
the same straight line and the only thing saving the ciphertext from
becoming totally transparent is the disparity (there for the finding
also by the same cryptanalyst) of the relative addresses on the number
line of these two.

No. 1 attack is the inversion attack or ciphertext only attack, let
the next attack be indexed No 2 attack and is based on numerical
linear analysis and linear differential analysis using ordinary
mathematics.

The final attack No 3 seeks to circumvent both of these i.e. No’s 1 ,
2 in one fell swoop by attempting to map the ciphertext directly to
the plaintext it represents on a statistical basis. My answer to that
is to make the ciphertext truly random i.e. having no repeats so that
each element of ciphertext has equal probability of being the right
one and the attack then collapses into total uncertainty.

Having put so much thought into the cipher algorithm fearing
ciphertext only attack as the most likely one and to which the cipher
is most vulnerable, the cryptographer may forget or underestimate or
even be totally unable, to foil this third attack.

I want to address this last attack here and describe how it is
counteracted in my cryptography.

First of all the highly transparent number line situation where the
ciphertext is a sitting duck waiting to be shot at by a cryptanalyst
is foiled by the concept of ‘displacement’ where by the numbers that
represent the ciphertext are scattered wildly throughout three-
dimensional space instead of being addresses in the same street so to
speak.

Attack No 3 is foiled by making the ciphertext scientifically random.
This has to be structured from the very beginning in the cipher design
by a multiplicity of component keys only one of which must be truly
random i.e. the finished ciphertext string.

It is interesting to note that when a number of truly random keysets
i.e. having zero repeats, are factors in an encryption process they
lose their randomness when they operate on each other during the
process i.e. they can surface with such copious repeats that by the
definition they are not now random any more. This securing randomness
ploy needs to be prodigiously crafted into the ciphertext by the
design algorithm.

At the present time I have found by means of casual operational
research that the number of keysets should be three or four at least
and the numeric ciphertext needs to be as large as possible with seven
digits per number being a minimum. Clearly, such a rule-of-thumb
needs to be ratified more scientifically (deserves proper research)
but the interesting thing is that there is a new area for research
here and also, this obviates the need for generating separate random
keys beforehand something that has preoccupied cryptographers for over
a century. The nub of the matter seems to be the encryption algorithm
generates its own randomness internally without needing difficult-to-
find dedicated random keys to be provided from the outside as a
precondition.

There is a price to be paid for large-number ciphertext of course but
so be it if it means secure communications that are independent of
computer power for all time.

Recapping on the displacement cipher being called “Skew Line
Encryptions”

1) This form of attack i.e. ciphertext only attack is prevented by a
mathematical one-way function derived from a change-of origin being
given to the position vector that comprises the ciphertext.

2) This form of attack is foiled by the ciphertext string being made a
discontinuous string of wildly disparate elements.

3) This form of attack is foiled by the ciphertext string being made
truly random right through the creation stage.

This post is made with the best intentions, I am merely reporting back
to you things I have found while working on my own cryptography that I
think might be of interest to others. I am not being patronising or
pontificating to anybody.


- adacrypt
.

Relevant Pages