DEPLOYMENT OF IOT-ENABLED SENSING DEVICES FOR REAL TIME FOREST FIRE PREDICTION SYSTEM USING BOUNDARY CONSTRAINED CNN MODEL

Abstract

Forests are vital to human existence as they maintain ecosystem balance, ensuring sufficient rainfall and other essential resources. Wildfires pose a significant threat to forested areas, capable of rapidly destroying vast numbers of trees and animals within hours. Technologies such as deep learning, the Internet of Things (IoT), and intelligent sensors offer promising solutions for creating a smart wildfire prediction system to help preserve the environment. This work proposes a method for predicting forest fires using boundary-constrained convolutional neural networks (CNNs). The system employs a sensor device integrated with the framework to detect forest fires based on learned reasoning. A 1020-megapixel digital camera has been adapted for monitoring purposes. The sensing device is equipped with two robust sensors: one for smoke detection and another for temperature and humidity observation. Utilizing these sensors, the NodeMCU processor reports data on fumes, humidity, and moisture. The IoT enables wireless alerting by transmitting data. The system collects and stores forest data sent by the sensor unit to a remote cloud computing environment. Acting as a continuous bridge between the sensor assembly and the server, the NodeMCU processor is equipped with WiFi for internet data retrieval. The proposed method effectively recognizes fire indicators and alerts the appropriate parties to take necessary measures to extinguish a forest fire.