Fault Detection in Forest Fire Monitoring Using Negative Selection Approach

Forests are the lifelines of any economy. Forest fires are uncontrolled and unpredictable once occur, identified after their wider spread in the forest. Forest fire detection using Wireless Sensor Network (WSN) is a significant application to monitor it. However, in some situations, unreliable/incorrect data may be sent by sensor nodes due to some fault, resulting in significant problems. Such faulty sensor nodes are unknown to the network and that causes major problems. Hence, the Detection of faulty nodes, if any, is essential and a prerequisite. To fulfill this necessity, faulty node, and forest fire detection is carried out in this research. A Negative Selection Approach (NSA) is introduced in the first phase for faulty node detection and a threshold-based approach is introduced for forest fire detection in the second phase of the proposed work. The experimental findings manifest that the proposed methodology performs efficiently in different scenarios. An NSA-based fault detection approach provides better accuracy with an average of 98.18%, an average of 3.56% False Alarm Rate (FAR), and an average of 1.81 % False Positive Rate (FPR).