19-06-2012, 01:48 PM
Password Authentication with Behavioral Metrics
Password Authentication.pptx (Size: 627.21 KB / Downloads: 35)
PAAKL is a method for password authentication.
PAAKL performs authentication by verifying that password using behavioral metrics extracted from
the typing style of the
rightful owner of a
computer account.
OUR METHOD
This method consists of two phases
Training Phase and
Authentication Phase.
An AKL is constructed from the user’s inputs.
Each input consists of the characters that comprise the user’s password and of the typing style of the particular user.
During the Training Phase, the user is a legitimate user .
During the Authentication Phase, both legitimate and non-legitimate users attempt to gain access to a computer account
TRAINING PHASE
During the Training Phase a user supplies many inputs, each input being the typing of the user’s password. The system records the typed characters and the delays between consecutive keystrokes and forms K-nodes that have the structure
Authentication phase
During the authentication process the user that attempts to access the account enters her password.
The entered password is used to form a K-line Lin similar to the way that it is described earlier. The K-nodes of K-line Lin are matched against the K-nodes of the existing AKL
conclusion
The main feature of our method is that in addition to performing standard password character matching, PAAKL captures the typing style of a user and utilizes this style in the authentication process.
Our findings are that the method allows legitimate users to access their account for the vast majority of cases -- 92.5% while it denies access to illegitimate users in all cases . This is encouraging, especially for the latter case