Extending the Lifetime of Sensor Networks Using Prediction and Scheduling

Power consumption in wireless sensor networks (WSNs) is a very important issue. Using measured sensor network data, this paper shows that it is possible to conserve a significant amount of energy through the proper use of data prediction and node scheduling without a significant loss in accuracy. Results show that it is possible to increase lifetime by up to 2600% at the cost of increasing average error by 0.5degC for temperature or 1.5% for humidity measurements. The four main design issues tackled are clustering, prediction, scheduling, and spike errors.