A Cloud Database Service Approach to the Management of Sensor Data Zhenguo Cui 1 , Meilan Jiang 1 , Karpjoo Jeong 2,4 , Bomchul Kim 3 1 Department of Advanced Technology Fusion, Konkuk University, Korea 2 Department of Internet and Multimedia Engineering & UBITA Center for Eco-Informatics Konkuk University, Korea 3 Department of Environmental Science, Kongwon National University, Korea 4 Qualcomm Institute, UC San Diego, USA {jinkook 1 , meela 1 , jeongk 2 }@konkuk.ac.kr bkim 3 @kangwon.ac.kr Abstract—The management of sensor data is challenging for most scientists or engineers. A cloud database service is a novel effective approach to such data management. In this paper, we presented a SaaS service that is based on a variant of the O&M model and implemented on Google App Engine. This system was applied for the management of sensor data from the water quality monitoring of the Soyang Lake. Index Terms—Monitoring, Scientific Data Management, Data Model, Cloud Computing, Google App Engine. I. MOTIVATION Monitoring is crucial for many scientific and engineering applications (hereafter, just scientific applications)[1]. There have been lots of R&D efforts for monitoring technologies and systems. Due to recent advances in information technology including sensors and wireless communication, sensor-based real time monitoring is widely used for various applications, these days[2]. However, the management of monitoring data is still challenging for most scientists[3]. The management of monitoring data in scientific domains raises two challenging issues to domain scientists: • Develop their own data management systems for their applications • Maintain those systems on their own. Since scientific applications are diverse in data management requirements, it is almost impossible to use an off-the-shelf data management system without a significant amount of customization or extension. Even if there are suitable data management systems available, the effective administration and maintenance of such systems are still really challenging for most scientists. Therefore, most scientists use simple spreadsheet programs such as MS Excel that support ‘free-style’ data management. However, those spreadsheet programs do not support data modeling explicitly and usually cause ad-hoc data management. As a result, many scientific communities are now facing the challenge of managing, analyzing and sharing a large number of spreadsheet files whose data structures are not well-defined or well-standardized. Recent technological developments in cloud computing and data model standards for observation and measurement provide us with opportunities for addressing the above two challenging issues. They include commercial cloud database services such as Google App Engine (GAE) [4] or Amazon EC2 [5] and well-defined data models such as Sensor Web Enablement (SWE) [6]. Cloud database services allow us to implement data management as an online service like web email. Standards like SWE provide data models, protocols and interfaces that can be used for a variety of scientific applications [7]. In this project, we developed the management of sensor- based real time monitoring data as a cloud database service which are based on a subset of the SWE data model standards (more specifically, similar to O&M in SWE). This system was intended to allow scientists to manage and share their sensor data in an online service like web email which is available anytime, but does not require any system administration and maintenance. II. SCIENTIFIC MONITORING AND DATA STANDARDS These days, many scientists monitor ecosystems, environments, experiments, and products by a variety of sensors. Analyzing such data is now a crucial task in many advanced research projects. A sensor usually generates a measured value of certain property of its associated target at a specific point in time; that is called monitoring or observation. At this point, two types of data must be collected and stored: • Measured data from sensors. Data from the actual sensor measurement of a property is usually a single value or a sequence of values. The structure and format of such data from various sensors are generally pre-defined and usually similar for various applications • Metadata about the observation and measurement. In addition to measured data, the analysis of the measured data requires information about the observation that is called metadata. Such metadata include information about the target object, the property, the procedure, and the sensor.