Network Forensics Analysis of Cyber Attacks Carried Out Over Wireless Networks Using Machine Learning Methods

Abstract

As technology advances, the frequency of attacks targeting technological devices has surged. This rise in cyber threats poses a constant risk to the devices we rely on. Any device connected to a network becomes vulnerable to exploitation by attackers. Given the extensive interconnectedness of devices in network environments, this research endeavors to address this pressing issue. The aim of this study is to analyze and classify network traffic generated during potential cyber attacks using various classification algorithms. By subjecting a simulated environment to different cyber attack scenarios, we extract the distinctive features of network packets generated during these attacks. Subsequently, we employ widely used classification algorithms to train and analyze the obtained data. For the comparison of models, more than 7000 attack data instances were employed. At the conclusion of the comparison, the Gradient Boosting algorithm achieved the highest accuracy value, reaching 91%, whereas the Naive Bayes algorithm obtained the lowest accuracy, reaching 74%.