Contextual, Behavioral, and Biometric Signatures for Continuous Authentication

Abstract

Continuous authentication in the Mobile Internet of Things should be based as broadly as possible, since a wide range of factors continuously reveal unexpected correlations. Such factors may include captured events (e.g., password, fingerprint, application start and end, network connect, and disconnect), continuous time series (e.g., gesture, typing rate, accelerometer, GPS, ambient sound, light levels, and time-of-day), and derived behavioral features (e.g., user sociability, browser and application menus, application choice). All these factors have been shown to correlate with the actual user identity, often in surprising combinations. More and more sensors are being deployed in autonomous devices, smart environments and vehicles, enabling even further behavioral and contextual data to be analyzed. The pegs of this continuous authentication “big tent” are moving out further than ever before, bringing it closer to practical uses in our everyday lives.