Fortifying IoT Infrastructure Using Machine Learning for DDoS Attack within Distributed Computing-based Routing in Networks

• Published in: Qubahan Academic Journal Vol. 4; no. 2; pp. 569 - 581
• Main Authors: Abdulkadhum Abbas, Sharaf Aldeen, Abdu Ibrahim, Abdullahi
• Format: Journal Article
• Language: English
• Published: Qubahan 30.06.2024
• Subjects: Cyber Attacks; Cybersecurity; IDs
• DOI: 10.48161/qaj.v4n2a581
• ISSN: 2709-8206

The DDoS, Also known as the Denial-of- Service cyber-attack, is now widely used, especially after such technologies as the IoT (Internet of Things) became mainstream and data traffic prefers link routes. Their effectiveness is limited by the attacks being controlled by the center, the limited data transmission capacity and also the viruses' ability to operate under-the-roof while using the mobile nodes which helps them move covertly. On the other hand, if we consider the conventional security approaches, these security devices mostly use traditional security protocols such as password encryption and user authentication respectively. The aim of this paper is to analyze the architecture of attack detection that are deployed in the IoT network and correspondingly demonstrate whether their function is to track and follow or even attack subjects. Moreover, the paper demonstrates the proper job of the detectors in keeping the networks to be safe. The algorithms that are tended to do the machine learning which is about past occurrences of these attacks and then soon to come up with new solutions which somehow or very likely could control or minimize attacks that might be prejudicial are the typical way that attacks are prevented. This research aims to compare the key machine learning approaches, Namely Support Vector Machines (SVM), Random Forest (RF) and Decision Trees (DT), in their ability to classify Intrusion Detection Systems (IDS) via routing networks over distributed computing systems. In addition, Algorithms perform quality control to determine the optimal hyperplane for the given data, Find neighboring data points and preserve the structure of the tree. We evaluate these algorithms using metrics such as the confusion matrix, F1 score, and AUC-ROC to determine their performance in managing imbalanced datasets and generating meaningful insights. Our results indicate that Random Forest outperforms the other models, achieving an accuracy of 99.2%, a false positive rate of 0.8%, and an AUC-ROC of 0.997.