An Efficient Adaptive Sampling Technique based on the Kalman Filter for Sensor Monitoring

Abstract

In sensor network environments, each sensor measures the physical environments according to the sampling period, and transmits a sensor reading to the base station. Thus, the sample period influences against importance resources such as a network bandwidth, and a battery power. In this paper, we propose new adaptive sampling technique that adjusts the sampling period of a sensor with respect to the features of sensor readings. The proposed technique predicts a future readings based on KF (Kalman Fiter). By using the differences of actual readings and estimated reading, we identify the importance of sensor readings, and then, we adjust the sampling period according to the importance. In our experiments, we demonstrate the effectiveness of our technique.