A Fault Detection System for Large Scale Sensor Networks Considering Reliability of Sensor Data

In recent years, sensors have become smaller, more affordable, and widely used. Also, Internet spreads all over the globe, and made possible transfer large data instantly. Thereby, large scale weather sensor networks have been constructible. In this kind of sensor network, we need to consider load balancing, network redundancy, and reliability of sensor data. Using P2P technology is one of the solutions for load balancing and network redundancy. Fault detection is a method for calculating reliability of sensor data. We propose a system that can calculate reliability of sensor data. This system can install method of fault detection by plug-in. It can install and update not only existing method but also new method. In other words, this system is a sustainable fault detection system. In these days, pinpoint heavy rain is occurring frequently. In this paper, we validate a fault detection system at a pinpoint heavy rain. We consider a validation, and improve a plug-in that can be useful at a pinpoint heavy rain. We install the improved plug-in to system, and experiment it using real pinpoint heavy rain data. The result shows that our system can detect 100% of data fault at pinpoint heavy rain which occurred in August 29th, 2008.