Public keys and their protocols

Question

I'm having difficulties understanding exactly what this protocol means:

$S \to D : \{N_S , S\}K_D$

$D \to S : \{N_S , N_D \}K_S$

$S \to D : \{N_D \}K_D$

"where $S$ represents the supervisor’s console and $D$ represents the door controller and the other symbols have their usual security protocol meanings."

I understand $N$ to be a nonce, a randomly generated number/hash that will only be used once.

I'm assuming $K$ is the public key?

Could someone explain this protocol for me, I’ve looked everywhere and cannot find anything like this.

Answer 1

It seems like an authentication protocol. In the first message, $S$ identifies herself. In the 2nd message $D$ proves knowledge of her private key (otherwise she could not return $N_s$. In the third message, $S$ proves knowledge of her private key.

You asked in a comment about replay attacks. There are two possibilities, the attacker is impersonating $S$ or $D$.

For $S$, the attacker might replay a captured 1st message. $D$ would see it, decrypt, return $N_S$ with a randomly generated $N_D$. The attacker could not possibly know what $N_D$ is and could not send the last message.

For impersonating $D$, $S$ would send the first message, but the attacker couldn't decrypt it and therefore could not send the 2nd message.

So, it seems the protocol is secure against replay attacks.

Answer 2

Under the assumption that $K_X$ represents $X$'s public key, this is the three-message version of the Needham-Schroeder public-key protocol. Many mutual authentication protocols derive from this design.

Wikipedia has the seven-message version of the protocol. The essential difference between that version and the three-message version is that the seven-message version explicitly includes the certificate exchanges for the public keys. The three-message version simply assumes that the correct public keys are known.

The protocol was proven secure using a formalism called BAN logic, which is hardly used nowadays. Almost 20 years after the protocol was proposed it was shown that it is vulnerable to a man-in-the-middle attack if the adversary has his own public/private key pair, see this paper by Gavin Lowe.

Answer 3

This looks like a symmetric-key protocol: $S$ first sends a nonce $N_S$ along with its name encrypted under the key $K_D$. Then $D$ responds with an encryption of $N_S$ along with another nonce $N_D$, this time encrypted under $K_S$. Finally $S$ sends $N_D$ to $D$, this time encrypted under $K_D$.