New Encryption Algorithm

I have coded an unique encryption method in C.
This method make use of cellular automatas on the files, at the bit level.
Giving a key A of size T this method offers :

256^( T / 8 ) possibilities for an enemy to search in.

giving a 1024bit key, this method is practically unbreakable.
It does NOT RELY ON ALGEBRA FORMULA at all !

The key is divided in bytes. Each byte reffers to a different Cellular Automata (if you don't know CA, you should read A New Kind of Science by Wolfram, or just google this)
Then we apply one step of each CA, so the file is encrypted. I have made this method reversible.

This method of encryption is pretty slow, I have done 500k / min on my Athlon XP 2600+

What do you think of that ?
You can email me at nicolas.said@la-machine.net

 

Log in or Sign up to hide all adverts.

How are you certain there are no shortcuts?

 

Log in or Sign up to hide all adverts.

All steps of the transformation process are independant (so we can use the conditional probability formula to find the nb of different crypted sequences possible).
It's also a bitlevel encryption algorithm, I don't know if it have been done before...

There is only two way to find the correct key :
- Guess the right code
- Have the right code

 

Log in or Sign up to hide all adverts.

Lol! Of course, but the amount of guesses or processing power needed may be greately reduced by flaws in the algorithm.
Like, for example: assume you encrypt a file by doing:
y = x * 5
y*4
y/2
y/2

This is stupid of course but you get the point. In this case it is sufficiently to divide the result by 5. There is no need to do the rest.

Can I see your code?

 

Camion, If your program is unique then you should be careful, encryption algorythms are usually the types of code that programmers seek software patents for to stop people having open access to reversing the encryption.

Also seeking such things as software patents will introduce you to certain areas of laws that countries were trying to pass in regards to the use of encryption.

Notibly I'm a bit rust but the laws went like the U.S. was allowed to keylog/spy on a user, where as the U.K. had to get a legal warrant to ask for the encryption keys from a user (and there was some discussion on the software company having a "backdoor key" purely generated for government bodies.)

 

I don't know the details of your algorithm, obviously, but are you positive that it's reversible? The way I'm imagining it works, the output of the encryption may not necessarily be unique to a given input.

 

http://crc.stanford.edu/crc_papers/alyamaniDAC03.pdf

You can use cellular automata for many things, even for creating music. Similar transformations are used in the Turing deformations on those pesky little characters some sites use when you register (bots cannot 'see' those).

I agree with malkiri on the problem of reversibility, tho...

 

Nice work. Can the source be posted?

 

Reversibility is achieved by adding a dependance on the step back in the rule.
The reverse rule is just the inverse of the rule.

Such rule can be seen at page 437 in "A New Kind of Science"

 

I'll see if I can find that, but not all rules are invertible, that's the whole point, or not without guaruantees of freedom of 'collissions' etc. (two different reversed rules, identical outcome).

Think of the simple modula rule, and similar 'trapdoor' routines.

In order to have it accepted as a valid encryption routine you'd have to cough up the math proving this at one point or another.

But again, interesting.

Please Register or Log in to view the hidden image!

 

Collision detection takes time, so... Well, still working on it.

Please Register or Log in to view the hidden image!