A Data Model for Storing a Large Amount of Sensor Data
Nobuyoshi Yabuki, Yoshitoku Yoshida · 2006
As the importance of structural health monitoring is increasingly recognized and cheaper sensor networks are becoming available, very large amount of sensor data will be stored in many computers in a distributed manner in the near future. For each database storing sensor data of a single structure or a set of structures administered by a designated office, various changes such as changes or addition of sensors, relocation of sensors, modification of structural members, refurbishment, administrators changes, etc., will give an impact on the data and querying methods to the database systems. From a broader point of view, data stored in many local database systems should be somehow integrated and utilized for generating or discovering knowledge based on the long term large amount of sensor data. To control and correspond to various changes of the sensor environment and to utilize the large amount of sensor data effectively, appropriate data models would be necessary. In this research, a new data model was proposed for solving the problems stated above, on the basis of the event-based relational data model. A data model was developed for storing and reusing experimental data along with their related data generated at E-Defense, the world largest 3D shaking table experiment facility in Japan, as a part of the E-Defense Grid (EDgrid) research project. In this paper, the data model is described and the feasibility of applying the developed data model for sensor data of real structures is discussed.