Semantic Streams: A Framework for Composable Semantic Interpretation of Sensor Data

We present a framework called Semantic Streams that allows users to pose declarative queries over semantic interpretations of sensor data. For example, instead of querying raw magnetometer data, the user queries whether vehicles are cars or trucks; the system decides which sensor data and which operations to use to infer the type of vehicle. The user can also place constraints on values such as the the amount of energy consumed or the confidence with which the vehicles are classified. We demonstrate how this system can be used on a network of video, magnetometer, and infrared break beam sensors deployed in a parking garage with three simultaneous and independent users.