A Machine Learning Based Method for Improvement of DDoS Attack Defense in IoT Networks

Abstract

The “Internet of Things” (IoT) has been a factor of major impact in facilitating connectivity among a wide range of devices and machines. It offers enhanced user experiences, better management, control, and analysis while facilitating informed decision-making. However, the low processing and storage capacity makes them vulnerable to a variety of network threats. “The Distributed Denial-of-Service (DDoS)” has been perceived as a significant issue in the realm of IoT services, causing disruptions and compromising data flow within and outside of IoT networks. Although there has been several research in this field, accuracy, and efficiency of classification emerge as definitive factors to design a sustainable defense mechanism. Most of the studies are concerned only with attacks targeted to only a specific type of network protocol. The previous research work has been carried out with limited available datasets. The current paper aims to achieve sustainable accuracy in application scenarios using hybrid network traffic which has been missed in all of the earlier research works. In addition to the above, this research uses a custom data set and also facilitates faster detection and blocking without incurring external networking overheads at the edge points of the network.