Estimating which Object Type a Sensor Node is Attached to in Ubiquitous Sensor Environment

Abstract

By simply attaching sensor nodes to physical objects with no information about the objects, the method proposed in this article infers the type of the physical indoor objects and the states they are in. Assuming that an object has its own states that have transitions represented by a state transition diagram, we prepare the state transition diagrams for such indoor objects as a door, a drawer, a chair, and a locker. The method determines the presumed state transition diagram from prepared diagrams that matches sensor data collected from people’s daily living for a certain period. A 2 week experiment shows that the method achieves high accuracy of inferring objects to which sensor nodes are attached. The method allows us to introduce ubiquitous sensor environments by simply attaching sensor nodes to physical objects around us.