Preventing simple replay attacks

Now we look at a simple replay attack. After a protocol session of an initiator $A$ and a responder $B$, an intruder $I$ stores all the messages sent in the session. Then, $I$ tries to re-send the packets to $B$ to impersonate as $A$. If $I$ can trick $B$ to finish its session believing it is talking to $A$, then the protocol is flawed and is discarded. Similarly, $I$ can launch the simple replay attack to $A$ as well. The purpose for this attack is just to check whether nonces are used in a correct way. The intruder does not try to encrypt or decrypt messages or alter the received messages and hence is very efficient.