Neural Network-Based Hardware Architecture multi- sensor for Monitoring Forest fires in Wireless Sensors Network

In this paper, we present and evaluate multi-sensor hardware architecture for monitoring forest fires using wireless sensors network (WSN). In the proposed architecture, each node is equipped with various sensors for temperature and carbon monoxide in order to improve accuracy detection. A method of neural network is applied to the data processing detected by the different sensors. This hardware multi-sensor architecture was assed for monitoring forest fires, especially designed to enable low power and higher precision in WSN. In this paper we show that neural network is a powerful and accurate mechanism which can successfully be applied not only to fire detection but to any event detection sensor network application. Compared to using crisp values, neural network maintains a high accuracy level despite fluctuations in the sensor values. This helps to decrease the number of false detection, while still providing accurate event detection. The proposed hardware architecture is planned to reach high performance running in FPGAs circuit. Synthesis results and relevant performance comparisons with related works are presented.