Homophonic Cipher

The homophonic cipher is a substitution cipher that uses a correspondence table between the letters / characters of the plain message and one or more letters / numbers / groups of characters.

Consequently, the same letter may have several possible encryption and the same message will possibly have several possible encrypted versions.

The homophonic cipher helps hide the frequency of letters used in a message, thus making frequency analysis more difficult for a cryptanalyst.

To use the homophonic cipher, the user must first define a correspondence table between the plain text characters and one or more symbols for each character.

During the encryption process, randomly choose from these symbols each time a specific character needs to be encoded. This ensures that the same character is not always represented by the same symbol in the ciphertext.

Use the correspondence table to transform each character / group of characters in the plain message.

By taking 100 numbers, the ideal match is approximately equal to the frequency of each letter in the English language (expressed as a percentage rounded to the nearest unit)

Using 50 numbers, roughly divide the quantities by 2.

When well done, the frequency of characters in a homophonically encrypted message is close to that of a perfectly random text, its index of coincidence too.

In order to simplify the work of the receiver of the message, it is common to use groups of numbers/letters of fixed length (2 or 3 digits).

If the correspondence table is accessible, the figure becomes a classic substitution system.

Using a homophonic cipher makes the ciphertext more resistant to frequency analysis methods.

In cryptography, hiding the frequency of letters in text helps protect against decryption attempts that exploit this information.

If the correspondences are proportional to the frequency of characters in the language of the plain message, then it is impossible to use frequency analysis or coincidence index techniques.

From a portion of known or assumed text, it is possible to guess a few correspondences of some characters but it is often insufficient to decipher the whole message.