Comprehensive 5G intrusion detection dataset for network security research

This thesis focuses on the development of a scalable and portable testbed for simulating the 5G core network and creating a high-quality dataset to train Machine Learning (ML) based Intrusion Detection System (IDS) to secure 5G networks from evolved attacks. The testbed allows for the simulation of multiple legitimate user activities and cyberattacks targeting the core network. This environment provides an invaluable resource for generating real-time system logs and metrics, which are later used to create a dataset specifically tailored to Fifth Generation (5G) core networks.

The primary goal of the thesis is to generate a rich dataset that consists of both malicious and binning traffic patterns in a 5G core network. This dataset includes aggregated log counters and key metrics collected from the core network nodes during scenarios of both normal and malicious traffic. A key contribution of this research lies in the innovative approach to dataset creation, which eliminates the dependency on network traces. This approach ensures a lightweight and efficient framework for dataset generation, enabling the creation of IDS with significantly reduced resource consumption. Furthermore, the comprehensiveness of the dataset was evaluated using three different ML models. Hence the dataset will be a valuable resource for researchers and relevant professionals to design and evaluate machine learning models for intrusion detection.