Low duty-cycling with spline-based curve fitting of sensor data in wireless sensor networks

Traditionally, sensors in wireless sensor networks are designed to collect data from the area of interest and forward it to the base-station. In periodic sensing, the prior knowledge about the data collected by sensor helps in making the sensor more sophisticated. The proposed scheme verified through the simulations helps to conserve the highly constrained resources of the network through pro-active decision making by the sensor-node. A spline curve fitting model built using historical data of the sensor is installed on the sensor and the user-node. The model helps to predict the current observed value knowing the past readings of the sensor. If relative-error between the calculated and the observed value by sensor is less than certain threshold, the sensor could schedule itself to stay idle instead of being in transmission mode. The same model installed on the user-node could be used to obtain the approximated observed value. This paper contributes by exploring an untapped area for node scheduling in wireless sensor networks. The proposed scheme uses a decentralized scheduling algorithm which is generic and easy to implement.