GIWRF-SMOTE: Gini impurity-based weighted random forest with SMOTE for effective malware attack and anomaly detection in IoT-Edge

Abstract

ABSTRACT The Internet of Things (IoT) is a smart technology that has switched the conventional way of living into smart living. As their usage becomes unavoidable, malware attacks in IoT networks have also increased. Many investigations and studies have proposed different methods to detect malware attacks, but these measures have some performance degradation in terms of accuracy, error, and lack of comprehensiveness. The cloud-based IoT infrastructure further creates latency and security problems. The machine learning (ML)-based edge computing can overcome these complications by automating the responses and moving the computation nearer to the network edge, where data is created. In this work, the performance of various prominent ML algorithms, such as logistic regression (LR), naive Bayes (NB), support vector machine (SVM), decision tree (DT), random forest (RF), and k-nearest neighbor (KNN), has been compared to predict malware attack accurately in IoT-edge environment. To enhance the prediction accuracy of the ML algorithms, the unbalanced data is converted into balanced data using the synthetic minority oversampling technique (SMOTE) and optimum features are selected using the Gini impurity-based weighted RF feature selection technique (GIWRF). The investigational results show that among six ML algorithms, RF with GIWRF attained the highest accuracy of 99.39%.