# Historical algorithm which is frequency analysis resilient

Long ago, I read an article on Wikipedia which described a cipher algorithm, used by the KGB (if I remember well), with the property that all ciphered letters had the same (or nearly the same) probability to appear, without adding symbols to the alphabet of the ciphered text (i.e. unlike how homophonic encryption does). The algorithm was computable by hand.

But I cannot find this article again. Does anyone know the article I mentioned?

• Does "without adding symbols to the ciphered text" mean that the Plaintext message space has the same cardinality of the Ciphertext message space ? Feb 24, 2016 at 16:08
• No, I meant "without adding symbols to the alphabet of the ciphered text". I'll edit my question. Feb 24, 2016 at 16:10
• IMHO without homophones, i.e.the alphabet sizes of plaintext and ciphertext are the same and the length of the texts are identical, it is not feasible to change the probability distribution (in reality), except in cases where some characters happen to be absent in the particular plaintext given. Feb 24, 2016 at 16:11
• The cipher text may be longer than the plaintext, I don't remember. However, your assumption seems false to me : with one-time-pad, all the size are the same and the distribution is uniform. Feb 24, 2016 at 16:16
• You are right but I thought you were considering only "classical" schemes. An ideal OTP would lead to perfect security in the sense of Shannon (though it needs some nice work in practice). Feb 25, 2016 at 10:26

## 1 Answer

You probably mean the VIC cipher.

It can be set up in a way such that the initial substitution that also achieves fractionation through the use of a so-called straddling checkerboard is followed by a transposition or substitution step. These steps make the distribution of symbols in the cipher text more uniform than in the plain text.

An optional subsequent reverse substitution using the same straddling checkerboard leads to a cipher text that contains only symbols of the original alphabet.