Drift detection and calibration of sensor networks

Wireless sensor network (WSN) is becoming more and more important with the development of Internet-of-things (IoT), it has been widely used to sense and monitor an area of interest in many scenarios. Since sensors in WSN can suffer from drift or errors with the increase of deployment time, the trustworthiness of sensor data will be affected. In this paper, an algorithm is proposed to detect and calibrate drift of sensors using clustering method and Kalman filter. The proposed method neither requires a densely deployed assumption nor needs prior knowledge of data models but utilizing data correlation among sensors. It also allows multiple sensors to drift at the same time. The results show that proposed algorithm can successfully run to improve the trustworthiness of collected data, and is also robust even in a high-noise scenario.