Data Mining Approach to Predict Forest Fire Using Fog Computing

Fires caused in forests are known to be one of the most hazardous environmental issues which cannot be neglected. Fast and quick prediction is the only way by which we can at least face it with readily available fire extinguishing resources. To accomplish this, one of the best ways is to use automatic tools based on locally placed sensors, such as rain, thermal readings, relative dampness and wind. In this work, we inquire a Data Mining (DM) approach to predict the area prone to forest fire. SVM (Support Vector Machine) algorithm is used for live sensor data. SVM algorithm utilizes four sensor inputs related to weather info data (i.e. thermal data, relative dampness, rain and wind velocity) and it is feasible to predict the burned land zone due to small fires, which are periodic or frequent. Awareness about those zones are utilized for developing and deploying fire fighting resource management. The data received from the sensor devices are processed in individual fog nodes and the cumulative data is collected upon and is used for further analysis. The data transfer happens wirelessly via Zig-Bee tool. The results obtained is utilized to predict the areas which are prone to and will be affected by sudden outburst of forest fire.