Social Behaviometrics for Personalized Devices in the Internet of Things Era

As the integration of smart mobile devices to the Internet of Things (IoT) applications is becoming widespread, mobile device usage, interactions with other devices, and mobility patterns of users carry significant amount of information about the daily routines of the users who are in possession of these devices. This rich set of data, if observed over a time period, can be used to effectively verify a user. In previous works, verification of users on personalized electronic devices via biometric properties, such as fingerprint and iris, has been successfully employed to increase the security of access. However, with the integration of social networks with the IoT infrastructure and their popularity on smart handheld devices, identification based on behavior over social networks is emerging as a novel concept. In this paper, we propose an intelligent add-on for the smart devices to enable continuous verification of users. In the experiments, we use data from built-in sensors and usage statistics of five different social networking applications on mobile devices. The collected feature set is aggregated over time and analyzed using machine learning techniques. We show that when smart devices are equipped with continuous verification intelligence, it is possible to verify users with less than 10% false rejection probabilities, and the users can keep using the devices with no interruption for biometric authentication 90% of the time. In the case of anomalous behavioral patterns, the proposed system can verify genuine users with up to 97% success ratio using an aggregated behavior pattern on five different social network applications.