Trust-driven approach to enhance early forest fire detection using machine learning

Area of Science:

• Environmental Science
• Computer Science
• Artificial Intelligence

Background:

• Forest fires present significant threats to ecosystems and human communities.
• Early detection is crucial for mitigating adverse environmental and climatic impacts.
• Existing detection systems require enhancement in reliability and speed.

Purpose of the Study:

• To develop a real-time Universal Trust Model (UTM) for early forest fire detection (FFD).
• To improve the reliability and reduce the detection time of forest fire identification systems.
• To integrate intelligent wireless sensor networks (WSN) with machine learning for robust fire detection.

Main Methods:

• Implemented an intelligent WSN with clustered sensor nodes for extensive forest coverage.
• Developed a UTM calculating trust ratings based on communication, energy, and data factors for sensor nodes.
• Utilized a machine learning regression model analyzing temperature, humidity, and CO2 for enhanced detection precision.

Main Results:

• The proposed UTM system demonstrated a high data processing rate.
• Achieved a reduced time delay in fire detection compared to existing systems.
• Experimental validation with 7200 samples confirmed the system's efficacy in early-stage forest fire detection.

Conclusions:

• The UTM system offers a robust and accurate solution for early forest fire detection.
• Combining trust mechanisms with machine learning significantly advances fire detection capabilities.
• The system is a promising solution for prompt forest fire detection and prevention, especially under challenging conditions.