Fighting Identity Fraud with the Addition of Biometric Techniques

This paper reviews current levels of identify fraud and then addresses a key component in the efforts to combat this fraud, that is authentication. Two key problems exist with authentication. First, authentication systems treat identity as a localized entity when it is actually a distributed concept. Stolen identifiers are not detected because repudiating data for the real person is not locally available. Second, authentication systems use identifying information that is prone to theft. Password, token cards and PIN numbers can be stolen whereas retinal and handwriting patterns cannot readily be “lifted.” We therefore propose that the solution to identify theft involves combining multiple identification methods, some with biometric measures. We describe our research in progress using keystrokes dynamics to identify the typist, perhaps of a personal identification number. Previous work uses time latencies. We are proposing incorporating both time and pressure measures to give us more discriminative and resilient identification.