If a protocol is implemented using a poor password, then the password can be guessed and verified from the messages in the protocol run. This is termed as a guessing attack. Published design and analysis efforts always lacked a general definition for guessing attacks. Further, they never considered possible type-flaws in the protocol runs or using messages from other protocols. In this paper, we provide a simple and general definition for guessing attacks. We explain how we implemented our definition in a tool based on constraint solving. Finally, we demonstrate some new guessing attacks that use type-flaws and multiple protocols which we found using our tool.
展开▼