I wrote my own cipher to encrypt messages. I would like to test a sample ciphered message to see how strong it is. Are there any tools for such task either in Windows or Linux ?
|
To calculate the time to brute force, you actually just have to look at the (effective) key space. If there are $k$ different possible keys, and all of them encrypt differently¹, it takes in average about $k/2$ tries to brute-force your cipher. How long this takes then depends on the speed of your cipher (or an optimized version thereof), and the hardware resources of an attacker. ¹This seems to be trivially the case, but it isn't: While DES uses formally a 64-bit key, 8 of these bits are not used, thus the effective key size is 56 bits, which means that in average $2^{56}/2 = 2^{55}$ tries are enough for brute-force. Of course, with home-brewn ciphers usually the best attack is not a brute-force attack, but some clever cryptanalysis, which then only needs way less time. These ciphers are then called "broken". |
|||
|
|
|
There's no way to verify the strength of a cipher in any automated way, especially given only the ciphertext. Attacking the construction of a cipher is an inherantly intellectual process, which does not fit well to the qualities of computer analysis. For example, I could write a cipher such that $f(m) = rand(0,n)$ where $n$ is the largest integer representable in $sizeof(m)*8$ bits, and any automated analysis would say "this is an awesome cipher", but a human would immediately realise that it's totally useless. If you're writing your own ciphers for real usage, I have only one piece of advice: don't. Creating your own ciphers is fun, and it can give you a learning experience if you're actually doing it based on constructions you understand, but the resulting cipher should never ever go anywhere near a production system. Experts put their ciphers through years (decades, even) of peer review before they're adopted by anyone. If your goal is to learn, do so by breaking ciphers. |
|||
|
|