An Efficient and Robust Framework for IoT Security using Machine Learning Techniques

Spotting and approximation of malicious node(s) in sensor based network is an open challenge. The proposed research work presented here primarily focuses on identification and estimation of malicious nodes within IoT networks following a machine learning-based models. The SensorNetGuard dataset was employed for the development and testing of the machine learning models such as Decision Tree (DT), Support Vector Machines (SVM), K-Nearest Neighbor (KNN), and Random Forest (RF) etc. The presented model here has been developed and evaluated using Python libraries like Scikit-learn, Seaborn, Matplotlib, and Pandas. In this work, Random Forest model has been emerged as a most effective model in detecting malicious nodes and shows an accuracy, recall, ROC AUC, precision, and F1-score of 99.99% and Cohen’s Kappa of 0.99. This depicts the capability of machine learning performance toward real-time IoT security. The SensorNetGuard dataset will be publicly available on platforms like IEEE DataPort and Kaggle to enable further research.