P9.2 - Fast and Efficient Dual-Forecasting Algorithm for Wireless Sensor Networks
Florian Strakosch, Faouzi Derbel · Proceedings SENSOR 2015 · 2015
Data transmission is by far the most energy consuming operation a wireless sensor node needs to perform.Hence, the most effective method of extending the lifetime of such a usually battery-powered device is to reduce the amount of sending intervals.One way of treating this issue is by using system identification and forecasting.That basically implies to take a reasonable amount of sample data for building a close-to-perfect system model and to perform a one-step prediction out of the given information on sensor node and gateway simultaneously.For doing so measurements of temperature, humidity, force and friction have been taken and analyzed.The novelty of our approach is the simplicity of our algorithm compared to other works, its measureable energy savings in real life environments and its potential for online parameter estimation.