Vigenere applied to encrypted data

Question

What happens if you encrypt data with a good cypher that produces a cyphertext that looks random (say, AES but avoiding ECB mode), and then apply a Vigenere cypher to the encrypted random-looking data?

1. Encrypt data so it looks random (for example with AES-CBC)
2. Choose a long random binary key for Vigenere (for example: cc3e87ba3f43...)
3. XOR that key repeatedly, as in a Vigenere cypher (although we modify bytes instead of letters)

Questions: how strong is that Vigenere cypher, and how can it be broken without having to bruteforce all the Vigenere keys? If it can't be broken without bruteforcing all the Vigenere keys, does that mean that in this scenario it becomes a safe cypher, and with a key longer than 256 bits we could obtain a cypher that is even stronger than AES?

Answer 1

If you already encrypted with a strong cipher (e.g AES-256) and a strong secret key, the data is secure without Vigenere applied on top. Adding a Vigenere layer on top won't hurt security but what is it's purpose?

Vigenere is usually very easy to break using frequency analasis however if the data input into Vigenere is undistinguisable from random data removing the Vigenere is challenging. In fact if we encrypt with Vigenere truelu random data, the fact the Vigenere key is repeated doesn't help us, we still can't learn the key or peal the Vigenere encryption.

Answer 2

The answer is somewhere between a bad idea and perfect security. If the key is as long as the message or even longer, your Vigenere cypher becomes a One Time Pad. The latter is perfectly secure, but not good for practice.

One the other hand: Normal Vigenere is easy and fast to solve computationally. Therefore you may not gain any security benefits in general.

Note: The combination of cryptographic applications can also lead to a completely insecure scheme. Moreover, the current security of AES-256 is so strong that it does not need to be improved (especially not for private use).