Securing Smart Homes using a Behavior Analysis based Authentication Approach

Abstract

This paper presents TRICA, a security framework for smart homes. When using controlling apps (e.g., smartphone app), TRICA makes sure that only legitimate users are allowed to control their Internet of Things (IoT) devices. Leveraging User Behavior Analytics (UBA) and Anomaly Detection (AD) techniques, TRI CA collects and processes the historical cyber and physical activities of the user in addition to the historical states of the smart home system to build a One Class Support Vector Machines (OCSVM) model. This model is then used as a baseline from which anomalous commands (i.e., outliers) should be detected and rejected, while normal commands (i.e., targets) should be considered as legitimate and allowed to be executed. Experiments conducted on adapted real-world data properly show the feasibility of such user behavior-based authentication approach. TRICA exhibits low false accept and false reject rates ensuring both security and user convenience, respectively.