Intelligent Wireless Sensor Networks for Early Fire Warning System

Abstract

Forest and wide area fires are one of the biggest obstacles to the protection of green areas, which are the life guarantee for future generations. These fires, which constitute a major threat to the ecosystem, can lead to human and animal deaths. In this work, an early warning system was proposed to detect forest fires with wireless sensor networks. The wireless sensor network consists of spatially distributed sensors to analyze the situations of the environment for various scenarios. Near real-time data from various sensors are transmitted to the server within mesh network structure via redundant paths using low cost and low power Zigbee. The communication in the system is continued thanks to this mesh design without any loss of data when there is a connection break between the sensor nodes and the beacons. Multi layer feed forward neural networks are utilized for decision making stage by adopting to the conditions of the environment rhythms. By designing a graphical user interface that works on a local machine, the approximate location of the modules, status color codes and status visualization with Google Maps on web server and notification of the relevant authorities are provided.