Enhanced Cyberattack Detection in Wireless Sensor Networks Using Gradient Boosting Decision Tree

Wireless Sensor Networks (WSNs) are essential to the Internet of Things despite their power constraints. As sensor deployment has increased, so have security risks, requiring effective attack detection and prevention solutions. Machine Learning (ML) might build strong systems that detect cyber threats in real time. Data preparation, feature extraction, and model training are used to improve WSN security in this research. Preprocessing included data purification, conversion, standardisation, and splitting. Principal Component Analysis (PCA) reduced dimensionality, preserved variance, and improved computing speed. Model training uses Gradient Boosting Decision Trees (GBDT), notable for its classification accuracy in complex datasets. Averaging 93.11 % accuracy, the GBDT model outperformed the BiLSTM and DCNN models. This high accuracy shows GBDT's capacity to identify and mitigate WSN-based IoT security threats. The findings show that machine learning, particularly GBDT, can efficiently identify WSN intrusions to improve IoT network security and resilience. This study establishes complete cybersecurity safeguards for the IoT ecosystem, enabling more secure and robust applications.