Keystroke dynamics on a mobile handset: a feasibility study

With the introduction of third generation phones, a technological transition is occurring in which the devices begin to have similar functionality to that of current personal digital assistants. The ability of these phones to store sensitive information, such as financial records, digital certificates and company records, makes them desirable targets for impostors. Current security for mobile phones is provided by the Personal Identification Number (PIN), which has weaknesses from both technological and end-user perspectives. As such, non-intrusive and stronger subscriber authentication techniques are required. This paper details the feasibility of one such technique, the use of keystroke dynamics. This feasibility study comprises a number of investigations into the ability of neural networks to authenticate users successfully based upon their interactions with a mobile phone keypad. The initial results are promising with individual users’ classification performing as well as 0% false rejection and 1.3% false acceptance.