In the mutual authentication in the Needham/Schroeder protocol upon which Kerberos is based, the authenticator contained only an encrypted timestamp. The protocol is that Alice sends Bob the authenticator, and then Bob must decrypt the authenticator, add one to the value inside, re-encrypt it, and send it back to Alice. Why was it necessary for Bob to increment the value before re-encrypting it and sending it to Alice? Why isn’t it necessary in Kerberos V5, in the AP_REP message? In Kerberos V4, it is the checksum field (which isn’t really a checksum—see §11.10 Encryption for Integrity Only) that is extracted and incremented. Would it have been just as secure in V4 for Bob to send back the contents of the checksum field encrypted and not incremented?
