SDPA: Sensor Data Processing Architecture for Modeling Semantic Data from Sensor Steams

With the rapid deployment of a number of sensors, it is crucial to efficiently manage their data streams with heterogeneous properties. To achieve various sensor applications such as discovery and mashup, a method of retrieving meaningful information from raw sensor data is required. However, it is hard to analyze and represent the sensor data since sensors generate streaming data of different patterns and continuously transmit the observations to servers in real-time. In this paper, we propose a sensor data processing architecture to retrieve meaningful information from raw sensor data. In particular, we adopt a machine leaning strategy for sensor data analysis. Semantic sensor data are modeled based on ontologies. The processed semantic data construct a semantic knowledge base, which allows a user to make the best use of sensor information. We present an evaluation of our approach by using real-world datasets and experimental results.