Distributed Event Detection in Wireless Sensor Networks Using Machine Learning

Abstract

In Wireless Sensor Networks (WSN), when an usual event is noticed in the networks, an event is detected through the sensor devices placed at distributed locations. This event detection information is passed to the base station and intelligent decision is taken. We proposed an ensemble distributed machine learning approach for detecting events. This approach works in 3 steps: collection of data, defining levels of fires and division of dataset. Regression and SVM are the approaches used in proposed architecture for detection of events and prediction of forest fires. This method uses regression for calculating the detection accuracy and errors and levels of fires are defined by SVM. The predictors considered in the dataset are significant and thus help in better prediction of forest fires