A Conditional Tabular Generative Adversarial Network (CTGAN)-based approach to safeguarding artificially created smart IoT settings

Abstract

In the modern world, cyber security is essential since social media and mobile phones are used by all people and might lead to security issues. Machine learning algorithms, and more specifically deep learning algorithms, have seen widespread application in recent years in a variety of sectors, including cyber security. Initially, the stages at which adversarial attack tactics occur, as well as the goal and capabilities of the attacker, are used to characterize them. Next, we classify the ways that adversarial attack and defense techniques are used in the field of cyber security. Finally, we highlight certain aspects of recent research and analyze how recent advancements in other adversarial learning domains may impact future research directions in cyber security. This method depends heavily on using a Conditional Tabular Generative Adversarial Network (CTGAN), a powerful tool that learns from real data patterns to generate realistic synthetic network traffic. Implementing this generated information in a software-defined networking (SDN)-based simulated network environment is a crucial step in our strategy to evaluate and improve the traffic patterns. The resulting dataset's accuracy, using a decision tree, was 0.97, and its less complicated structure was obtained with training and test periods of 0.07 and 0.005 seconds, respectively. The findings demonstrate that synthetic data are appropriate for Internet of Things (IoT) contexts and smart city applications since they are simpler and correctly represent real data.