Enhancing IoT Security to Leveraging ML for DDoS Attack Prevention in Distributed Network Routing

Abstract

DDoS attacks have become much more frequent especially with the situating of IoT technologies. These attacks are based on compromising weaknesses within the IoT network and focuses on communication links used as paths to flood systems with large volumes of traffic and thereby cause system breakdowns. DDoS attacks lean on centralized control mechanisms and less available bandwidth of IoT infrastructures and the clandestine mobility of the nodes. Traditional measures of security like password and user accounts protection are typically insufficient to counteract such threats. This paper focuses on the system architectures of DDoS attack detection solutions in the IoT networks to evaluate their ability to counter the attack. More specifically, it focuses on the function of machine learning systems, which are designed to analyze previous attacks, in order to stop new attacks. The comparisons with three Machine Learning algorithms—Support Vector Machines, Random Forest, Decision Trees are done with relation to their capability to classify invasion attempts within distributed IoT network routing system. It also assesses the methods for optimising these models such as determination of hyperplane, the clustering of data point as well as the tree structure. Common performance parameters such as confusion matrix, F1 measure, and AUC-ROC are used to ensure the efficiency, effectiveness and to handle the cases with intricate unbalanced datasets to enhance the IoT network Security.