As part of the ongoing prototyping and firmware development, the encoding process is continuously enhanced to achieve optimal security and ease of use.
The following example demonstrates how the cipher is used for encoding the seed. The real procedure is not currently disclosed, but the source code will be made available as an open-source solution in the future.
Seeder uses the Vernam Cipher in the following way:
- Starting with our secret text:
"This is a perfect cipher based on an unknown attempt to uncover the secrets of encryption. It’s called Vernam Cipher and is unbreakable, making it perfect for our example. Enjoy decoding."
- Random numbers are generated (e.g., 5, 8, 6, 1, 1, 3).
- The text is repeated multiple times until it exceeds 300 characters, selecting words based on the random numbers:
- 5 = based
- 8 = on
- 6 = secrets
- 1 = is
- 1 = unbreakable
- 3 = perfect
- Every second word has its letters reversed, so the cipher phrase looks like this:
"basedno sretecunbreakableperfect..."
- Remove capital letters, punctuation, and numbers, leaving only selected letters based on randomly chosen positions, resulting in a distorted version of the original text.
- Convert the distorted text into ASCII values, producing a series of numbers like:
122 97 108 111 101 110 97 101 115 111...
- Seed words (e.g., "pillow", "river", "meeting", "apple") are padded with random uppercase letters to a total of 8 characters and combined into one string. The final encrypted string might look like:
"RUFpillowDRriverGBPmeetingDappleDDX..."
- Shift letters based on ASCII values (e.g., "R" shifted by 122, "U" by 97, and so on).
The final encrypted seed looks like this:
X586113V_dSUAVsbKcXEAORwoXR`ENAmSSEELp...
By reversing the process using the same cipher key, the original seed words can be recovered.
Note: Uppercase letters are used as separators or ballast to ensure that the text length remains constant, making it impossible for an attacker to guess word positions or lengths. This is essential, as the seed is generated from 2048 known words, each between 3 to 8 characters long, making decryption unpredictable.
