Introduction
Every human being who uses the computer also uses a keyboard. The keyboard is either placed separately in front of the monitor, attached inside the laptop or even in the smart phones. Some people write slow, others fast. The typing rhythm might change over time, depending on the mood and time of the day. Biometric keystroke recognition is the technology of recognizing people from the way they are typing. By using different data analysis techniques, it might be that every human being has a unique way of typing.
Keystroke Analysis – State of the Art
A particular way or manner of typing
These words are the definition of keystroke. Every person has his or her own way of typing on keyboards. Several human factors, such as aging, injuries, operations on the hand etc. may change a person’s keystroke style into a slight different typing, either permanent or temporary. Elders usually have slower keystroke the older they get which makes the timings of the keys longer.
Keystroke recognition, however, is obviously a more semi-obtrusive biometric than fingerprint. It gives the possibility to identify human-beings in-front of a computer without any “real” direct explicit interaction with the computer. For example, while a person is typing something on the computer, the computer will extract features and analyze the keystrokes where the user don’t need to think of the authentication.
In case of weak quality features, it would be more sufficient to have a fingerprint as a first time strong authenticator, because that keystroke recognition is still under research to be a strong and robust biometric. However, until now the keystroke recognition can be used as an additional method for increasing security by (un)obtrusive and periodic re-verification of a person identity. To have an idea of what an strong biometric is can be seen in Table \ref{table:strong_biometric}. The EER rates are still quite different than the WS-based ones.
[TABLE strong:biometrics ]
Researching at different methodologies to analyzing the features of keystroke is increasing and become popular area of research in keystroke biometrics. Feature extraction from typings is acrucial for the efficient keystroke recognition. Throughout history, many different features were used such as latency, duration, pressure, etc.
There are two types of keystroke dynamics. The first one is static keystroke dynamics in which the keystrokes are analyzed only at specific times e.g. during login. The second one is continuous keystroke dynamics in which the typing characteristics are analysed during a complete session.Static approaches provide more robust user verification than simple passwords. However staticmethods do not provide continuous security, specifically they cannot detect substitution of the user after the initial verification. Continuous verification monitors the user’s typing behaviorthroughout the session. Therefore it can be used to detect uncharacteristic typing rhythm caused by say drowsiness \cite{16}.A lot of reports can be found on keystroke dynamics dealing with a static authentication. Lesscan be found on Keystroke dynamics based on continuous authentication. In this chapter we are going to talk about the literature concerning those two parts.
More information will be available in week 6…