Prediction of Compromised IoT Infrastructure Using Machine Learning

Abstract

The rapid growth of the Internet of Things (IoT) has led to big advancements in Fog Computing, Smart Cities, and Industry 4.0. These areas handle complex data and need strong protection against cyberattacks. To make IoT devices use power more efficiently, the IETF created the RPL protocol (Routing Protocol for Low-Power and Lossy Networks). However, due to its sophisticated design it is vulnerable to attacks. One of the most successful attacks against this protocol is flooding attacks, which leads to resource exhaustion in nodes. Consequently, there is a pressing need for new security methods. Nonetheless, there is a scarcity of readily accessible, comprehensive, and organized datasets specifically tailored for IoT, as well as benchmark datasets, to train and assess machine learning models. Therefore, the primary focus of this research is on creating new labeled IoT-specific datasets with the COOJA simulator and processing these packets with machine learning algorithms. Decision Tree, Random Forest, K-Nearest Neighbor and Artificial Neural Network algorithms were compared for identifying the Flooding Attacks. The average accuracy obtained for each of the above algorithm is 89.24%, 91.128%, 89.25% and 89.73% respectively.