DefinePK

Abstract

Wireless Sensor Networks (WSNs) are increasingly vulnerable to variety of security threats, making the detection of network intrusions critical. This study presents a novel approach for Intrusion Detection (ID) in WSNs by combining the advanced data pre-processing and feature selection techniques. Quantile Recursive-Random Outlier Imputing (QR-ROI) algorithm is used for effective data pre-processing, addressing missing or anomalous values in wsnds dataset, which is collected from the Kaggle repository. To further optimize the model's accuracy and efficiency, feature selection is conducted using the Wrapper Boruta Algorithm, enhanced with Light Gradient Boosting Machine (GBM), known for its speed and performance. The integration of these techniques not only improves the quality of dataset but also enhances the overall predictive accuracy to 94.12  of Intrusion Detection System (IDS). Experimental results demonstrate the efficacy of this approach in distinguishing between normal and malicious activities within WSN environments.