Abstract

Real-time online coral sensor and climatic data are vital for long term ecological observatory research. It is difficult to maintain a continuous real-time online data stream using various underwater sensors feeding to a server. To solve this difficulty, we integrated an event detection and decision support system (DSS) to help researchers to make decisions about sensor maintenance and to proactively prevent system malfunctions. This paper describes the components of pre-processing data in-situ, using behavior learning from historical data, creating anomaly detection, and verifying sensor data. This online system is flexible, for both health monitoring and decision-making. When data loss or sensor malfunctions occur, then an automatic email will be sent to researchers, to show both sensor health and anomaly reports. Researchers will then make a decision on who will fix the problems, and plan for maintenance on site. The integrated decision support system and event detection uses various tools and techniques, such as Mathematica for analysis and open source software (PHP, MySQL, and Highcharts) for visualization. This is the first practical system that uses heterogeneous sensors integrated with flexible cloud storage, which shows both alerts regarding system health and anomaly reports through social media, and helps researchers in decision-making, in order to reduce maintenance time and costs, and to enable long term use of sensor data in order to expand understanding on climate change.