Dual Prediction and Probabilistic Scheduling for efficient event detection in sensor networks

Energy efficient data collection protocols are required in order to better manage the limited energy, memory and processing capabilities of sensor networks. In applications where data are collected in real time, efficient management of sensor radio assumes critical significance because communication is energy intensive. Moreover, specialised sensors exist which consume even more energy than radio transceivers. This paper proposes a Dual Prediction and Probabilistic Scheduling (DPPS) algorithm to be used for event detection, with improvement of energy efficiency as a central purpose. Whereas previous studies focused solely on recording events within the sensor network environment, DPPS monitors event data while also considering the accuracy of the data collection process. This is done by using the same model for forecasting both at the sensors, where event data are transmitted, and also at the sink which receives and stores such data. By combining time series prediction with probabilistic scheduling, energy is conserved because forecasts within the application´s accuracy constraints reduce the need for unnecessary sensing. Simulation results on real datasets indicate that DPPS improves energy efficiency by reducing sensor usage by up to 40% in comparison to another fundamental stochastic algorithm. DPPS also provides stronger quality guarantees for application users by producing fewer missed events during the monitoring process.