Securing IoT Networks Against Cyber Threats Using an Improved Whale Optimization Algorithm

Abstract

The advancement of Internet of Things (IoT) technology has enhanced efficiency and innovation, but it also presents new security challenges, particularly against sophisticated botnet cyberattacks. These attacks are too difficult for conventional security measures to handle, which emphasizes the need for better security solutions. This study presents a machine learning (ML) framework for securing IoT networks, using a modified whale optimization algorithm (WOA) for feature selection with an adaptive crossover function. The framework, tested on the N-BaIoT dataset, shows notable improvements in performance metrics across various classifiers for both binary and multiclass classification tasks. Particularly, classifiers such as random forest (RF), light gradient boosting machine (LGBM), adaptive boosting (AdaBoost), and histogram-based gradient boosting (HGB) achieved 100% accuracy, precision, and f1-score in binary classification. For multiclass classification tasks, the extreme gradient boosting (XGBoost), gradient boosting trees (GBTs), and artificial neural network (ANN) also demonstrated enhanced performance, with the XGBoost achieving 99.41% accuracy, precision, and f1-score.