In this contributed post the Universidade de Aveiro, Instituto de Telecomunicações, Portugal, explains how nProbe has be successfully used in 5G networks.
Introduction
As networks evolve to meet the demands of modern connectivity, the need for intelligent traffic monitoring and anomaly detection becomes increasingly critical. In the context of 5G networks, where high-speed data transfer and low latency are paramount, Machine Learning (ML)-based solutions provide a robust mechanism for detecting anomalies and ensuring network reliability.
Our project leverages nProbe, a high-performance NetFlow/IPFIX probe, to extract a comprehensive set of 43 traffic flow features from packet capture (pcap) files. Instead of real-world PCAP files, we utilize pre-collected data from the UNSW-NB15 dataset, a widely used dataset for network anomaly detection research, which contains ground truth-labeled network traffic.
Background and Related Work
Our work builds upon previous studies that contribute to the detection of cyberattacks and network anomalies. Notable references include:
-
Creation of the UNSW-NB15 dataset:
-
“UNSW-NB15: A Comprehensive Data Set for Network Intrusion Detection Systems” (Nour Moustafa & Jill Slay, IEEE).
-
Available at: https://ieeexplore.ieee.org/abstract/document/7348942
-
-
Extraction of NetFlow features:
-
“A Study on Network Flow Features for Intrusion Detection” (Mohanad Sarhan, Siamak Layeghy & Marius Portmann, Springer).
-
-
Anomaly Detection:
-
“Cyberattacks Detection in IoT-based Smart City Network Traffic” (Article on Towards AI)
-
Why Use nProbe?
nProbe enables efficient extraction of network traffic metadata, providing a detailed view of communication patterns. Unlike traditional packet inspection tools, it is lightweight, scalable, and supports real-time data export, making it ideal for 5G network environments.
By analyzing these features, our ML models can detect patterns and anomalies that may indicate security threats, network congestion, or performance degradation.
This project is carried out by university students as part of an academic initiative aimed at applying and expanding existing research in the context of Network Data Analytics Functions (NWDAFs). Anomaly detection serves as a chosen proof-of-concept scenario due to its practical applicability and ease of implementation. As students, we face challenges such as limited resources and the need to rigorously replicate prior studies while ensuring our contributions add value to the research community. Despite these challenges, our work highlights the importance of hands-on experimentation and learning in real-world network security applications.
Our project replicates the methods used in the aforementioned studies, applying them within the context of a Network Data Analytics Function (NWDAF) in 5G networks. By combining the capabilities of nProbe with Machine Learning, we unlock powerful insights into network behavior, making 5G networks more secure and resilient.
For more details on the project, visit our documentation: https://rodrigoabreu22.github.io/PI_Microsite/