Considering the classic scenario of Eve trying to listen in to Alice and Bob's conversation, if Eve can intercept messages between Alice and Bob but with no access to their computers would Timing attacks be possible? Am I right in saying it isnt possible because say Alice sends a message to Bob how will Eve listen in to how long Bob's computer will take to decrypt the message?
Asked
Active
Viewed 414 times
1 Answers
0
Eve could not just listen to the communication. If you're talking about RSA I assume that Bob's public key is not secret and that even Eve might know it. Then Eve could try to act as Alice by sending messages encrypted with Bob's public key to Bob. If Bob always answers those messages Eve could use the delay until Eve receives the message to perform a timing attack.
Even if Eve can only listen to the communication and Bob sends a reply to Alice every time he receives a message Eve could measure the time between Alice's message and Bob's reply.
Only if Bob does not reply to Alice's messages a timing attack would not be possible. If Bob must reply to Alice's messages Bob must reply only after a fixed amount of time. For example, if it takes at most 100ms to decrypt a message and Bob decrypts some message in 78ms, Bob has to wait 22ms until he is allowed to send a response.
DanielE
- 701
- 4
- 10
-
"If Bob must reply to Alice's messages Bob must reply only after a fixed amount of time." do you mean that's how Bob and Alice can stop timing attacks? Or is that a rule in the communication because if it is then surely Eve will have no way of knowing the real decryption time.. Also, what if Bob doesn't read the email straight after it is sent.. surely that adds another time? Or am I taking it too literally? – obsessiveCookie Mar 18 '14 at 14:14
-
If Bob is a human who decrypts and reads emails after some time they have arrived there is no way for an attacker to perform a timing attack. A timing attack can only be performed if Bob's mail system receives the mail, the email is automatically decrypted and an automatic response is sent to Alice. This is possible because the time for each step is somehow fixed and only the time to decrypt the message is variable. If Bob starts the decryption manually how should an attacker know the time between receiving and decrypting the email? This knowledge is required for an timing attack. – DanielE Mar 26 '14 at 08:26
-
Thank you for your reply @DanielE. It is very clear to me now :) – obsessiveCookie Mar 26 '14 at 13:08