Physical-Interface-Based IoT Service Identification

Connected devices are exponentially invading our real world environment, making it smart (smart home / building / city / etc) and forming the Internet of Things (IoT). They are paving the way for major technological breakthroughs that will deeply change our way of life. In order to make this announced revolution comes true, some strong issues remain to be solved somehow. A major one relies on our ability to build existing or new services from this ocean of surrounding connected devices for any purpose, meaning whenever, wherever, whatever our technological skills and above all, whatever our unlimited service expectations. In such a context, this paper presents an end-to-end approach enabling the identification and future deployment of personalized IoT services using connected devices in various smart environments. It relies on a key postulate: IoT services are first driven by their physical interfaces interacting with the smart environments. To support this vision, we first defined a rich abstraction framework relying on attributed typed graphs. Then we developed what represents the cornerstone of the presented methodology: an innovative physical interface based algorithm enabling the identification and thus the classification of IoT services. We illustrate how this algorithm is working on some use-cases involving video-type interfaces. Finally, we discuss the future developments and the potential benefits of our approach to fulfill our vision of a world of smart environments where people could use their connected objects for any usage they could envisage and beyond.