Challenge-response protocol

A challenge-response protocol is a technique for managing source authentication.

As a simple example, consider a user who wants to log in to a computer system. If he sends the actual password, or a hash of it, over the wire, then a wiretapper can get it. Moreover, if an attacker can obtain the list of encrypted passwords from the target system, then he can conduct a dictionary attack and break into any account whose password is in the dictionary.

In a challenge-response protocol, the interaction might go like this instead:

1. the user requests a connection
2. the system generates a random challenge C, using a strong random number generator
3. the system sends C to the user
4. the user encrypts C with a block cipher, using the hash of his password as key
5. the user sends the encrypted challenge E back to the system
6. the system decrypts E, using the stored hash of the user password as key
7. if the result is C, the connection is allowed

There are many variations. The example uses a block cipher, but it can also be done with public key techniques — the user encrypts with his private key in step 4 and the system verifies with the public key in step six. A hash can also be used — steps 4 and 6 are then identical, hashing some combination of data that includes the challenge and the password (or its hash).

The challenge must be large enough to prevent a birthday attack. With a challenge of 2ⁿ bits, an attacker who logs the data from 2^n/2 interactions has a significant chance of seeing a repeated challenge.