XGBoost based Fault Detection for Wireless Sensor Networks

Diagnosing faults is a highly complex problem in the realm of Wireless Sensor Networks (WSNs). In Wireless Sensor Networks (WSN), sensor nodes are deployed in a random manner within a challenging environment, resulting in a significantly higher risk of faults occurring in these nodes compared to traditional networks. The network's functionality diminishes when the quantity of defective sensor nodes escalates within the network. The current fault detection method suffers from a high rate of false alarms and low accuracy in detecting faults, resulting in a continuous decline in network performance and quality of service. This research presents a machine learning approach using the XGBoost technique to identify problematic nodes in the sensor circuit of a network. The proposed technique is implemented using Google Colab, and its performance is assessed based on metrics such as false alarm rate, false positive rate, and accuracy. The simulation results clearly demonstrate the exceptional efficacy of the suggested method in identifying various forms of faults.