Xenia: Secure and interoperable smart home system with user pattern recognition

Smart home markets are rapidly growing due to the convenience and comfort they bring to their users. However, smart home appliances are still largely heterogeneous, typically requiring their own applications to control them. This usually means that smart home control remains largely user-driven, making the coordination of heterogeneous devices challenging. Moreover, recent cyber-attacks have demonstrated the vulnerability of many of these devices. Therefore, this paper presents Xenia, a secure smart home platform for management and control of smart home devices. In Xenia, APIs are defined for controlling devices through a centralized server. In addition, the devices can be controlled through text or voice chatting in natural English language. An intelligent agent is developed to give this feature human-like traits. To ensure coordination, the server also implements a graph-based machine learning algorithm to automatically detect user patterns around the house, in order to improve coordination of heterogeneous devices. Finally, a comprehensive security architecture is implemented that provides authentication, access control, encryption, intrusion detection, and firewall services to all parts of the system. Experiments demonstrate the veracity of our system operation.