P7: A sensor monitoring and management framework for industrial sensor networks

Many applications based-on sensor networks have recently been founded in industrial monitoring area. We motivate our technique in the context of the problem of sensor fault detection, sensor state recognition and health management. We use time windows to format the recent sensor data and the reduction results can be used to detect the data-centric faults. We divide sensors into groups and define the state transformation space to represent the normal sensor status. And we propose a lifetime prediction method for sensor management.We propose the framework of P7 which evaluates the sensor data from seven perspectives and give a ranking operator to each dimension respectively. Finally we figure out a health degree for the sensor. A system based-on P7 is proposed and we test this application for 14 months. The results of our case study indicate that P7 can detect not less than 90% sensor faults successfully and help the user to maintenance the sensor easily.